From lennart at karssen.org  Mon Feb  3 17:01:44 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Mon, 03 Feb 2014 17:01:44 +0100
Subject: [GenABEL-dev] Submitted updated version of MetABEL to CRAN
Message-ID: <52EFBD68.2000809@karssen.org>

Dear list,

Today I fixed a small 'bug' in MetABEL (values show on the vertical axis
of the forest plot [1]). While at it I also converted the documentation
to roxygen format and made some other small improvements to the layout
of the forest plots.

I've just uploaded these changes to CRAN.


Best,

Lennart.
-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140203/f62977ae/attachment.sig>

From yurii.aulchenko at gmail.com  Mon Feb  3 18:46:21 2014
From: yurii.aulchenko at gmail.com (Yury Aulchenko)
Date: Mon, 3 Feb 2014 18:46:21 +0100
Subject: [GenABEL-dev] Submitted updated version of MetABEL to CRAN
In-Reply-To: <52EFBD68.2000809@karssen.org>
References: <52EFBD68.2000809@karssen.org>
Message-ID: <43BF8E02-33B1-4E2E-85EC-7D73A2CC9808@gmail.com>

note that they do not like too frequent updates (or at least that was the case some time before)...

Y

On Feb 3, 2014, at 17:01, L.C. Karssen <lennart at karssen.org> wrote:

> Dear list,
> 
> Today I fixed a small 'bug' in MetABEL (values show on the vertical axis
> of the forest plot [1]). While at it I also converted the documentation
> to roxygen format and made some other small improvements to the layout
> of the forest plots.
> 
> I've just uploaded these changes to CRAN.
> 
> 
> Best,
> 
> Lennart.
> -- 
> *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
> L.C. Karssen
> Utrecht
> The Netherlands
> 
> lennart at karssen.org
> http://blog.karssen.org
> GPG key ID: A88F554A
> -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
> 
> _______________________________________________
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel


From lennart at karssen.org  Mon Feb  3 20:55:19 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Mon, 03 Feb 2014 20:55:19 +0100
Subject: [GenABEL-dev] Submitted updated version of MetABEL to CRAN
In-Reply-To: <43BF8E02-33B1-4E2E-85EC-7D73A2CC9808@gmail.com>
References: <52EFBD68.2000809@karssen.org>
 <43BF8E02-33B1-4E2E-85EC-7D73A2CC9808@gmail.com>
Message-ID: <52EFF427.60704@karssen.org>

Hi Yurii,

Thanks. I'll keep that in mind. I decided to submit anyway, since I
don't expect any other changes any time soon.

Anyway, Kurt mailed me that the package has been accepted.


Best,

Lennart.

On 03-02-14 18:46, Yury Aulchenko wrote:
> note that they do not like too frequent updates (or at least that was the case some time before)...
> 
> Y
> 
> On Feb 3, 2014, at 17:01, L.C. Karssen <lennart at karssen.org> wrote:
> 
>> Dear list,
>>
>> Today I fixed a small 'bug' in MetABEL (values show on the vertical axis
>> of the forest plot [1]). While at it I also converted the documentation
>> to roxygen format and made some other small improvements to the layout
>> of the forest plots.
>>
>> I've just uploaded these changes to CRAN.
>>
>>
>> Best,
>>
>> Lennart.
>> -- 
>> *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
>> L.C. Karssen
>> Utrecht
>> The Netherlands
>>
>> lennart at karssen.org
>> http://blog.karssen.org
>> GPG key ID: A88F554A
>> -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
>>
>> _______________________________________________
>> genabel-devel mailing list
>> genabel-devel at lists.r-forge.r-project.org
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
> 

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140203/e04b14be/attachment.sig>

From yurii.aulchenko at gmail.com  Mon Feb  3 20:56:31 2014
From: yurii.aulchenko at gmail.com (Yurii Aulchenko)
Date: Mon, 3 Feb 2014 20:56:31 +0100
Subject: [GenABEL-dev] Submitted updated version of MetABEL to CRAN
In-Reply-To: <52EFF427.60704@karssen.org>
References: <52EFBD68.2000809@karssen.org>
 <43BF8E02-33B1-4E2E-85EC-7D73A2CC9808@gmail.com>
 <52EFF427.60704@karssen.org>
Message-ID: <CAHX9t6JTENW5gBqJeS=RZEj4Z2O4yFFRWUmxv1abk5D5r=KTCg@mail.gmail.com>

Good news! - thank you!

Yurii

PS we actually have a nasty bug in MetA :(


On Mon, Feb 3, 2014 at 8:55 PM, L.C. Karssen <lennart at karssen.org> wrote:

> Hi Yurii,
>
> Thanks. I'll keep that in mind. I decided to submit anyway, since I
> don't expect any other changes any time soon.
>
> Anyway, Kurt mailed me that the package has been accepted.
>
>
> Best,
>
> Lennart.
>
> On 03-02-14 18:46, Yury Aulchenko wrote:
> > note that they do not like too frequent updates (or at least that was
> the case some time before)...
> >
> > Y
> >
> > On Feb 3, 2014, at 17:01, L.C. Karssen <lennart at karssen.org> wrote:
> >
> >> Dear list,
> >>
> >> Today I fixed a small 'bug' in MetABEL (values show on the vertical axis
> >> of the forest plot [1]). While at it I also converted the documentation
> >> to roxygen format and made some other small improvements to the layout
> >> of the forest plots.
> >>
> >> I've just uploaded these changes to CRAN.
> >>
> >>
> >> Best,
> >>
> >> Lennart.
> >> --
> >> *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
> >> L.C. Karssen
> >> Utrecht
> >> The Netherlands
> >>
> >> lennart at karssen.org
> >> http://blog.karssen.org
> >> GPG key ID: A88F554A
> >> -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
> >>
> >> _______________________________________________
> >> genabel-devel mailing list
> >> genabel-devel at lists.r-forge.r-project.org
> >>
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
> >
>
> --
> *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
> L.C. Karssen
> Utrecht
> The Netherlands
>
> lennart at karssen.org
> http://blog.karssen.org
> GPG key ID: A88F554A
> -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
>
>


-- 
-----------------------------------------------------
Yurii S. Aulchenko

[ LinkedIn <http://nl.linkedin.com/in/yuriiaulchenko> ] [
Twitter<http://twitter.com/YuriiAulchenko>] [
Blog <http://yurii-aulchenko.blogspot.nl/> ]
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140203/827cee34/attachment.html>

From lennart at karssen.org  Mon Feb  3 21:00:11 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Mon, 03 Feb 2014 21:00:11 +0100
Subject: [GenABEL-dev] Submitted updated version of MetABEL to CRAN
In-Reply-To: <CAHX9t6JTENW5gBqJeS=RZEj4Z2O4yFFRWUmxv1abk5D5r=KTCg@mail.gmail.com>
References: <52EFBD68.2000809@karssen.org>	<43BF8E02-33B1-4E2E-85EC-7D73A2CC9808@gmail.com>	<52EFF427.60704@karssen.org>
 <CAHX9t6JTENW5gBqJeS=RZEj4Z2O4yFFRWUmxv1abk5D5r=KTCg@mail.gmail.com>
Message-ID: <52EFF54B.7040205@karssen.org>

Hmm, I didn't see a bug report. Or didn't I look carefully enough?


Lennart.

On 03-02-14 20:56, Yurii Aulchenko wrote:
> Good news! - thank you!
> 
> Yurii
> 
> PS we actually have a nasty bug in MetA :(
> 
> 
> On Mon, Feb 3, 2014 at 8:55 PM, L.C. Karssen <lennart at karssen.org
> <mailto:lennart at karssen.org>> wrote:
> 
>     Hi Yurii,
> 
>     Thanks. I'll keep that in mind. I decided to submit anyway, since I
>     don't expect any other changes any time soon.
> 
>     Anyway, Kurt mailed me that the package has been accepted.
> 
> 
>     Best,
> 
>     Lennart.
> 
>     On 03-02-14 18:46, Yury Aulchenko wrote:
>     > note that they do not like too frequent updates (or at least that
>     was the case some time before)...
>     >
>     > Y
>     >
>     > On Feb 3, 2014, at 17:01, L.C. Karssen <lennart at karssen.org
>     <mailto:lennart at karssen.org>> wrote:
>     >
>     >> Dear list,
>     >>
>     >> Today I fixed a small 'bug' in MetABEL (values show on the
>     vertical axis
>     >> of the forest plot [1]). While at it I also converted the
>     documentation
>     >> to roxygen format and made some other small improvements to the
>     layout
>     >> of the forest plots.
>     >>
>     >> I've just uploaded these changes to CRAN.
>     >>
>     >>
>     >> Best,
>     >>
>     >> Lennart.
>     >> --
>     >> *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
>     >> L.C. Karssen
>     >> Utrecht
>     >> The Netherlands
>     >>
>     >> lennart at karssen.org <mailto:lennart at karssen.org>
>     >> http://blog.karssen.org
>     >> GPG key ID: A88F554A
>     >> -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
>     >>
>     >> _______________________________________________
>     >> genabel-devel mailing list
>     >> genabel-devel at lists.r-forge.r-project.org
>     <mailto:genabel-devel at lists.r-forge.r-project.org>
>     >>
>     https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
>     >
> 
>     --
>     *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
>     L.C. Karssen
>     Utrecht
>     The Netherlands
> 
>     lennart at karssen.org <mailto:lennart at karssen.org>
>     http://blog.karssen.org
>     GPG key ID: A88F554A
>     -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
> 
> 
> 
> 
> -- 
> -----------------------------------------------------
> Yurii S. Aulchenko
> 
> [ LinkedIn <http://nl.linkedin.com/in/yuriiaulchenko> ] [ Twitter
> <http://twitter.com/YuriiAulchenko> ] [ Blog
> <http://yurii-aulchenko.blogspot.nl/> ]

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140203/cd99bf0a/attachment.sig>

From yurii.aulchenko at gmail.com  Mon Feb  3 21:04:04 2014
From: yurii.aulchenko at gmail.com (Yurii Aulchenko)
Date: Mon, 3 Feb 2014 21:04:04 +0100
Subject: [GenABEL-dev] Submitted updated version of MetABEL to CRAN
In-Reply-To: <52EFF54B.7040205@karssen.org>
References: <52EFBD68.2000809@karssen.org>
 <43BF8E02-33B1-4E2E-85EC-7D73A2CC9808@gmail.com>
 <52EFF427.60704@karssen.org>
 <CAHX9t6JTENW5gBqJeS=RZEj4Z2O4yFFRWUmxv1abk5D5r=KTCg@mail.gmail.com>
 <52EFF54B.7040205@karssen.org>
Message-ID: <CAHX9t6LVG5Mfisc-P=xKHAxHkkfOUrc7=itNtzanTGBgXMWbkA@mail.gmail.com>

I get this as top entry in bug tracker :)

https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1187&group_id=505&atid=2058

it looks small but is a bit nasty - I remember I was looking at it,
thinking it is very easy one, and giving up soon (like 30 minutes) after
realizing it is not *that* simple.


On Mon, Feb 3, 2014 at 9:00 PM, L.C. Karssen <lennart at karssen.org> wrote:

> Hmm, I didn't see a bug report. Or didn't I look carefully enough?
>
>
> Lennart.
>
> On 03-02-14 20:56, Yurii Aulchenko wrote:
> > Good news! - thank you!
> >
> > Yurii
> >
> > PS we actually have a nasty bug in MetA :(
> >
> >
> > On Mon, Feb 3, 2014 at 8:55 PM, L.C. Karssen <lennart at karssen.org
> > <mailto:lennart at karssen.org>> wrote:
> >
> >     Hi Yurii,
> >
> >     Thanks. I'll keep that in mind. I decided to submit anyway, since I
> >     don't expect any other changes any time soon.
> >
> >     Anyway, Kurt mailed me that the package has been accepted.
> >
> >
> >     Best,
> >
> >     Lennart.
> >
> >     On 03-02-14 18:46, Yury Aulchenko wrote:
> >     > note that they do not like too frequent updates (or at least that
> >     was the case some time before)...
> >     >
> >     > Y
> >     >
> >     > On Feb 3, 2014, at 17:01, L.C. Karssen <lennart at karssen.org
> >     <mailto:lennart at karssen.org>> wrote:
> >     >
> >     >> Dear list,
> >     >>
> >     >> Today I fixed a small 'bug' in MetABEL (values show on the
> >     vertical axis
> >     >> of the forest plot [1]). While at it I also converted the
> >     documentation
> >     >> to roxygen format and made some other small improvements to the
> >     layout
> >     >> of the forest plots.
> >     >>
> >     >> I've just uploaded these changes to CRAN.
> >     >>
> >     >>
> >     >> Best,
> >     >>
> >     >> Lennart.
> >     >> --
> >     >> *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
> >     >> L.C. Karssen
> >     >> Utrecht
> >     >> The Netherlands
> >     >>
> >     >> lennart at karssen.org <mailto:lennart at karssen.org>
> >     >> http://blog.karssen.org
> >     >> GPG key ID: A88F554A
> >     >> -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
> >     >>
> >     >> _______________________________________________
> >     >> genabel-devel mailing list
> >     >> genabel-devel at lists.r-forge.r-project.org
> >     <mailto:genabel-devel at lists.r-forge.r-project.org>
> >     >>
> >
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
> >     >
> >
> >     --
> >     *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
> >     L.C. Karssen
> >     Utrecht
> >     The Netherlands
> >
> >     lennart at karssen.org <mailto:lennart at karssen.org>
> >     http://blog.karssen.org
> >     GPG key ID: A88F554A
> >     -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
> >
> >
> >
> >
> > --
> > -----------------------------------------------------
> > Yurii S. Aulchenko
> >
> > [ LinkedIn <http://nl.linkedin.com/in/yuriiaulchenko> ] [ Twitter
> > <http://twitter.com/YuriiAulchenko> ] [ Blog
> > <http://yurii-aulchenko.blogspot.nl/> ]
>
> --
> *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
> L.C. Karssen
> Utrecht
> The Netherlands
>
> lennart at karssen.org
> http://blog.karssen.org
> GPG key ID: A88F554A
> -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
>
>


-- 
-----------------------------------------------------
Yurii S. Aulchenko

[ LinkedIn <http://nl.linkedin.com/in/yuriiaulchenko> ] [
Twitter<http://twitter.com/YuriiAulchenko>] [
Blog <http://yurii-aulchenko.blogspot.nl/> ]
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140203/15591e40/attachment-0001.html>

From lennart at karssen.org  Mon Feb  3 22:32:35 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Mon, 03 Feb 2014 22:32:35 +0100
Subject: [GenABEL-dev] Submitted updated version of MetABEL to CRAN
In-Reply-To: <CAHX9t6LVG5Mfisc-P=xKHAxHkkfOUrc7=itNtzanTGBgXMWbkA@mail.gmail.com>
References: <52EFBD68.2000809@karssen.org>	<43BF8E02-33B1-4E2E-85EC-7D73A2CC9808@gmail.com>	<52EFF427.60704@karssen.org>	<CAHX9t6JTENW5gBqJeS=RZEj4Z2O4yFFRWUmxv1abk5D5r=KTCg@mail.gmail.com>	<52EFF54B.7040205@karssen.org>
 <CAHX9t6LVG5Mfisc-P=xKHAxHkkfOUrc7=itNtzanTGBgXMWbkA@mail.gmail.com>
Message-ID: <52F00AF3.3060202@karssen.org>

Oops, missed it somehow.


Lennart.

On 03-02-14 21:04, Yurii Aulchenko wrote:
> I get this as top entry in bug tracker :)
> 
> https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1187&group_id=505&atid=2058
> 
> it looks small but is a bit nasty - I remember I was looking at it,
> thinking it is very easy one, and giving up soon (like 30 minutes) after
> realizing it is not *that* simple. 
> 
> 
> On Mon, Feb 3, 2014 at 9:00 PM, L.C. Karssen <lennart at karssen.org
> <mailto:lennart at karssen.org>> wrote:
> 
>     Hmm, I didn't see a bug report. Or didn't I look carefully enough?
> 
> 
>     Lennart.
> 
>     On 03-02-14 20:56, Yurii Aulchenko wrote:
>     > Good news! - thank you!
>     >
>     > Yurii
>     >
>     > PS we actually have a nasty bug in MetA :(
>     >
>     >
>     > On Mon, Feb 3, 2014 at 8:55 PM, L.C. Karssen <lennart at karssen.org
>     <mailto:lennart at karssen.org>
>     > <mailto:lennart at karssen.org <mailto:lennart at karssen.org>>> wrote:
>     >
>     >     Hi Yurii,
>     >
>     >     Thanks. I'll keep that in mind. I decided to submit anyway,
>     since I
>     >     don't expect any other changes any time soon.
>     >
>     >     Anyway, Kurt mailed me that the package has been accepted.
>     >
>     >
>     >     Best,
>     >
>     >     Lennart.
>     >
>     >     On 03-02-14 18:46, Yury Aulchenko wrote:
>     >     > note that they do not like too frequent updates (or at least
>     that
>     >     was the case some time before)...
>     >     >
>     >     > Y
>     >     >
>     >     > On Feb 3, 2014, at 17:01, L.C. Karssen <lennart at karssen.org
>     <mailto:lennart at karssen.org>
>     >     <mailto:lennart at karssen.org <mailto:lennart at karssen.org>>> wrote:
>     >     >
>     >     >> Dear list,
>     >     >>
>     >     >> Today I fixed a small 'bug' in MetABEL (values show on the
>     >     vertical axis
>     >     >> of the forest plot [1]). While at it I also converted the
>     >     documentation
>     >     >> to roxygen format and made some other small improvements to the
>     >     layout
>     >     >> of the forest plots.
>     >     >>
>     >     >> I've just uploaded these changes to CRAN.
>     >     >>
>     >     >>
>     >     >> Best,
>     >     >>
>     >     >> Lennart.
>     >     >> --
>     >     >> *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
>     >     >> L.C. Karssen
>     >     >> Utrecht
>     >     >> The Netherlands
>     >     >>
>     >     >> lennart at karssen.org <mailto:lennart at karssen.org>
>     <mailto:lennart at karssen.org <mailto:lennart at karssen.org>>
>     >     >> http://blog.karssen.org
>     >     >> GPG key ID: A88F554A
>     >     >> -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
>     >     >>
>     >     >> _______________________________________________
>     >     >> genabel-devel mailing list
>     >     >> genabel-devel at lists.r-forge.r-project.org
>     <mailto:genabel-devel at lists.r-forge.r-project.org>
>     >     <mailto:genabel-devel at lists.r-forge.r-project.org
>     <mailto:genabel-devel at lists.r-forge.r-project.org>>
>     >     >>
>     >    
>     https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
>     >     >
>     >
>     >     --
>     >     *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
>     >     L.C. Karssen
>     >     Utrecht
>     >     The Netherlands
>     >
>     >     lennart at karssen.org <mailto:lennart at karssen.org>
>     <mailto:lennart at karssen.org <mailto:lennart at karssen.org>>
>     >     http://blog.karssen.org
>     >     GPG key ID: A88F554A
>     >     -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
>     >
>     >
>     >
>     >
>     > --
>     > -----------------------------------------------------
>     > Yurii S. Aulchenko
>     >
>     > [ LinkedIn <http://nl.linkedin.com/in/yuriiaulchenko> ] [ Twitter
>     > <http://twitter.com/YuriiAulchenko> ] [ Blog
>     > <http://yurii-aulchenko.blogspot.nl/> ]
> 
>     --
>     *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
>     L.C. Karssen
>     Utrecht
>     The Netherlands
> 
>     lennart at karssen.org <mailto:lennart at karssen.org>
>     http://blog.karssen.org
>     GPG key ID: A88F554A
>     -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
> 
> 
> 
> 
> -- 
> -----------------------------------------------------
> Yurii S. Aulchenko
> 
> [ LinkedIn <http://nl.linkedin.com/in/yuriiaulchenko> ] [ Twitter
> <http://twitter.com/YuriiAulchenko> ] [ Blog
> <http://yurii-aulchenko.blogspot.nl/> ]

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140203/7a3be368/attachment.sig>

From lennart at karssen.org  Wed Feb  5 22:13:20 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Wed, 05 Feb 2014 22:13:20 +0100
Subject: [GenABEL-dev] [Genabel-commits] r1594 -
	branches/ProbABEL-0.50/src
In-Reply-To: <20140204210043.DA5C9186CAD@r-forge.r-project.org>
References: <20140204210043.DA5C9186CAD@r-forge.r-project.org>
Message-ID: <52F2A970.4030101@karssen.org>

Hi Maarten,

Interesting commit! I didn't know about std::copy().

To help me understand: is this a bug in the current version ProbABEL as
well, or one that was introduced by one of your previous commits?


Thanks for the good work,

Lennart.

On 04-02-14 22:00, noreply at r-forge.r-project.org wrote:
> Author: maartenk
> Date: 2014-02-04 22:00:43 +0100 (Tue, 04 Feb 2014)
> New Revision: 1594
> 
> Modified:
>    branches/ProbABEL-0.50/src/maskedmatrix.cpp
>    branches/ProbABEL-0.50/src/reg1.cpp
>    branches/ProbABEL-0.50/src/regdata.cpp
> Log:
> -fixed incorrect multiplication with mmscore: other object was returned then expected.
> 
> 
> 
> Modified: branches/ProbABEL-0.50/src/maskedmatrix.cpp
> ===================================================================
> --- branches/ProbABEL-0.50/src/maskedmatrix.cpp	2014-02-04 11:27:35 UTC (rev 1593)
> +++ branches/ProbABEL-0.50/src/maskedmatrix.cpp	2014-02-04 21:00:43 UTC (rev 1594)
> @@ -67,7 +67,6 @@
>              nmeasured++;
>          }
>      }
> -
>      //Check update mask is the same as original matrix
>      if (nmeasured == length_of_mask)
>      {
> 
> Modified: branches/ProbABEL-0.50/src/reg1.cpp
> ===================================================================
> --- branches/ProbABEL-0.50/src/reg1.cpp	2014-02-04 11:27:35 UTC (rev 1593)
> +++ branches/ProbABEL-0.50/src/reg1.cpp	2014-02-04 21:00:43 UTC (rev 1594)
> @@ -164,9 +164,11 @@
>      // column with Prob(A1A1). Note the order is swapped cf the file!
>      int c2 = X.ncol - 1;
>  
> -    for (int i = 0; i < X.nrow; i++)
> -        for (int j = 0; j < (X.ncol - 2); j++)
> +    for (int i = 0; i < X.nrow; i++){
> +        for (int j = 0; j < (X.ncol - 2); j++){
>              nX[i * nX.ncol + j] = X[i * X.ncol + j];
> +        }
> +    }
>  
>      for (int i = 0; i < nX.nrow; i++)
>      {
> 
> Modified: branches/ProbABEL-0.50/src/regdata.cpp
> ===================================================================
> --- branches/ProbABEL-0.50/src/regdata.cpp	2014-02-04 11:27:35 UTC (rev 1593)
> +++ branches/ProbABEL-0.50/src/regdata.cpp	2014-02-04 21:00:43 UTC (rev 1594)
> @@ -14,7 +14,7 @@
>  #include "fvlib/frversion.h"
>  #include "fvlib/Logger.h"
>  #include "fvlib/Transposer.h"
> -
> +#include <algorithm> // STL algoritms
>  #include "regdata.h"
>  
>  regdata::regdata()
> @@ -203,7 +203,6 @@
>      int dim2X = X.ncol;
>      (to.X).reinit(to.nids, dim2X);
>      (to.Y).reinit(to.nids, dim2Y);
> -
>      int j = 0;
>      for (int i = 0; i < nids; i++)
>      {
> @@ -223,13 +222,10 @@
>      }
>  
>      // delete [] to.masked_data;
> -    to.masked_data = new unsigned short int[to.nids];
> -    for (int i = 0; i < to.nids; i++)
> -    {
> -        to.masked_data[i] = 0;
> -    }
> -    // std::cout << "get_unmasked: " << to.nids << " "
> -    //           << dim2X << " " << dim2Y << "\n";
> +    const int arr_size = nids;
> +    to.masked_data = new unsigned short int[arr_size];
> +    std::copy(masked_data, masked_data+arr_size,to.masked_data);
> +
>      return (to);
>  }
>  
> 
> _______________________________________________
> Genabel-commits mailing list
> Genabel-commits at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
> 

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140205/5a12956a/attachment.sig>

From lennart at karssen.org  Wed Feb  5 22:16:46 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Wed, 05 Feb 2014 22:16:46 +0100
Subject: [GenABEL-dev] [Genabel-commits] r1596 - branches/ProbABEL-0.50
In-Reply-To: <20140205202601.BAB3E186B9D@r-forge.r-project.org>
References: <20140205202601.BAB3E186B9D@r-forge.r-project.org>
Message-ID: <52F2AA3E.6080200@karssen.org>

Hi Maarten,

I haven't yet compiled this branch, but will be interested to see what
-Wextra gives in terms of warnings.

One more question below:

On 05-02-14 21:26, noreply at r-forge.r-project.org wrote:
> Author: maartenk
> Date: 2014-02-05 21:26:01 +0100 (Wed, 05 Feb 2014)
> New Revision: 1596
> 
> Modified:
>    branches/ProbABEL-0.50/configure.ac
> Log:
> added  CXXflags -DEIGENNO_DEBUG :removes checks for conformability (which gives a speedup in eigen)
> added  CXXflags -Wextra: adds adittional check (see http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html for more info)
> 
> 
> 
> 
> 
> Modified: branches/ProbABEL-0.50/configure.ac
> ===================================================================
> --- branches/ProbABEL-0.50/configure.ac	2014-02-05 16:35:59 UTC (rev 1595)
> +++ branches/ProbABEL-0.50/configure.ac	2014-02-05 20:26:01 UTC (rev 1596)
> @@ -21,9 +21,10 @@
>  AC_PROG_LN_S
>  if test -z "$CXXFLAGS"; then
>     # User did not set CXXFLAGS, so we can put in our own defaults
> -    CXXFLAGS="-g -O2"
> +    CXXFLAGS="-g -O2 -DEIGEN_NO_DEBUG -Wextra"
>  fi
> -if test -z "$CPPFLAGS"; then
> +if test -z "$
> +"; then
>     # User did not set CPPFLAGS, so we can put in our own defaults
>      CPPFLAGS="-Wall"
>  fi
> 

Is this correct? It seems that the name of the CPPFLAGS variable was
removed and
";then
moved to a new line.



Thanks,

Lennart.

> _______________________________________________
> Genabel-commits mailing list
> Genabel-commits at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
> 

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140205/699ef99d/attachment.sig>

From lennart at karssen.org  Thu Feb  6 14:11:09 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Thu, 06 Feb 2014 14:11:09 +0100
Subject: [GenABEL-dev] Abstract for the EMGM 2014 conference
In-Reply-To: <52EBE228.70503@karssen.org>
References: <52EAA7D1.5080106@karssen.org> <52EBE228.70503@karssen.org>
Message-ID: <52F389ED.4080909@karssen.org>

Dear all,

I have just received notification from the EMGM committee that my
abstract has been selected for a poster presentation.


Thank you for all you contributions!

Lennart.

On 31-01-14 18:49, L.C. Karssen wrote:
> Dear GenABEL devs,
> 
> Thank you for the various comments you sent. I have incorporated them,
> but also hd to shorten the abstract to 250 words. Please find below the
> abstract I submitted.
> 
> 
> 
> Best regards,
> 
> Lennart.
> 
> 
> Title: New software and developments in the GenABEL project
> 
> Authors: L.C. Karssen, on behalf of the GenABEL team
> 
> Abstract:
> In the last year the GenABEL project has seen a considerable number
> of improvements. These improvements do not only consist of updates
> to the packages in the GenABEL suite, but we also improved the
> development process and the way packages are made available to
> our users.
> 
> We will demonstrate newly implemented features in the GenABEL
> suite packages, as well as introduce OmicABEL, a package for rapid
> mixed-model based genome-wide association analysis of multiple
> traits (for example, metabolomics, glycomics, etc.).
> 
> Recently we started using the Jenkins Continuous Integration server
> to help us release software of higher quality. Jenkins is a
> framework that automatically runs several tests (e.g. static code
> analysis, checks for memory leaks) after a new commit to our version
> control system. This allows us to detect problems in the code at an
> early stage, before they bug the user.
> 
> After GenABEL, ProbABEL is the second package that is available as a
> Debian package, allowing users of upcoming Debian releases to
> install and upgrade ProbABEL with a single command. This also
> benefits users of Linux distributions derived from Debian, like
> Ubuntu.
> 
> In the coming year more packages are expected to be added the
> GenABEL suite as well as continued efforts to improve the existing
> ones. Moreover, we plan to increase both the ease of installation as
> well as the visibility of the GenABEL suite by adding more packages
> into both the Debian and Red Hat Linux repositories (as
> well as derivatives like CentOS and Scientific Linux).
> 
> 
> 
> On 30-01-14 20:28, L.C. Karssen wrote:
>> Dear list,
>>
>> I'm planning to go to the EMGM (European Mathematical Genetics Meeting)
>> in Cologne in April. I'd like to present a poster there and wrote the
>> abstract below.
>> Please let me know any comments or suggestions as soon possible as the
>> deadline for abstract submission is tomorrow (Fri 31 Jan).
>>
>> Thank you very much,
>>
>> Lennart.
>>
>>
>> --------------8<----------------8<------------------8<-------------
>>   Over the last year the GenABEL project has seen a considerable
>>   number of improvements. These improvements do not only consist of
>>   updates to the existing packages of the GenABEL suite, but are also
>>   manifest in the way the development process is being handled and
>>   the way the packages are made available to the users.
>>
>>   On our poster we will demonstrate the newly implemented features in
>>   the various packages of the GenABEL suite. We also welcome a new
>>   member to the GenABEL family: OmicABEL, a package for rapid
>>   mixed-model based genome-wide association analysis of multiple
>>   traits (think metabolomics, glycomics, etc.).
>>
>>   Recently we started using the open source Jenkins Continuous
>>   Integration server to help us release software of higher
>>   quality. Jenkins is a framework that automatically runs several
>>   tests (e.g. static code analysis, checks for memory leaks) for each
>>   of our packages. It builds and tests each project after a new commit
>>   in our version control system. This allows us to detect problems in
>>   the code at an early stage, before they bug the user.
>>
>>   After the GenABEL package, ProbABEL is the second package that is
>>   available as a Debian package. This means that users of upcoming
>>   Debian releases will be able to install ProbABEL with a simple click
>>   of a button or a single command. Moreover, since many other Linux
>>   distributions like Ubuntu and Linux Mint are derived from Debian,
>>   users of these distributions automatically benefit as well.
>>
>>   In the coming year more packages are expected to be added the
>>   GenABEL suite as well as continued efforts to improve the existing
>>   ones. Moreover, we plan to increase both the ease of installation as
>>   well as the visibility of the GenABEL suite by adding more packages
>>   into both the Debian and Red Hat Enterprise Linux repositories (as
>>   well as derivatives like CentOS and Scientific Linux).
>> --------------8<----------------8<------------------8<-------------
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> _______________________________________________
>> genabel-devel mailing list
>> genabel-devel at lists.r-forge.r-project.org
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
>>
> 
> --
> -----------------------------------------------------------------
> L.C. Karssen
> Utrecht
> The Netherlands
> 
> lennart at karssen.org
> http://blog.karssen.org
> 
> Stuur mij aub geen Word of Powerpoint bestanden!
> Zie http://www.gnu.org/philosophy/no-word-attachments.nl.html
> ------------------------------------------------------------------
> 
> 
> 
> _______________________________________________
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
> 

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140206/4b6da5c0/attachment.sig>

From kooyman at gmail.com  Thu Feb  6 17:46:03 2014
From: kooyman at gmail.com (Maarten Kooyman)
Date: Thu, 06 Feb 2014 17:46:03 +0100
Subject: [GenABEL-dev] [Genabel-commits] r1594 -
	branches/ProbABEL-0.50/src
In-Reply-To: <52F2A970.4030101@karssen.org>
References: <20140204210043.DA5C9186CAD@r-forge.r-project.org>
 <52F2A970.4030101@karssen.org>
Message-ID: <52F3BC4B.6030205@gmail.com>

Hi Lennart,

Thanks: there is also a std:count function which count the amount of a 
certain value in a array.

This bug is created when I removed multiple calling of get_masked_data 
for each snp.

KInd regards,

Maarten

On 05-02-14 22:13, L.C. Karssen wrote:
> Hi Maarten,
>
> Interesting commit! I didn't know about std::copy().
>
> To help me understand: is this a bug in the current version ProbABEL as
> well, or one that was introduced by one of your previous commits?
>
>
> Thanks for the good work,
>
> Lennart.
>
> On 04-02-14 22:00, noreply at r-forge.r-project.org wrote:
>> Author: maartenk
>> Date: 2014-02-04 22:00:43 +0100 (Tue, 04 Feb 2014)
>> New Revision: 1594
>>
>> Modified:
>>     branches/ProbABEL-0.50/src/maskedmatrix.cpp
>>     branches/ProbABEL-0.50/src/reg1.cpp
>>     branches/ProbABEL-0.50/src/regdata.cpp
>> Log:
>> -fixed incorrect multiplication with mmscore: other object was returned then expected.
>>
>>
>>
>> Modified: branches/ProbABEL-0.50/src/maskedmatrix.cpp
>> ===================================================================
>> --- branches/ProbABEL-0.50/src/maskedmatrix.cpp	2014-02-04 11:27:35 UTC (rev 1593)
>> +++ branches/ProbABEL-0.50/src/maskedmatrix.cpp	2014-02-04 21:00:43 UTC (rev 1594)
>> @@ -67,7 +67,6 @@
>>               nmeasured++;
>>           }
>>       }
>> -
>>       //Check update mask is the same as original matrix
>>       if (nmeasured == length_of_mask)
>>       {
>>
>> Modified: branches/ProbABEL-0.50/src/reg1.cpp
>> ===================================================================
>> --- branches/ProbABEL-0.50/src/reg1.cpp	2014-02-04 11:27:35 UTC (rev 1593)
>> +++ branches/ProbABEL-0.50/src/reg1.cpp	2014-02-04 21:00:43 UTC (rev 1594)
>> @@ -164,9 +164,11 @@
>>       // column with Prob(A1A1). Note the order is swapped cf the file!
>>       int c2 = X.ncol - 1;
>>   
>> -    for (int i = 0; i < X.nrow; i++)
>> -        for (int j = 0; j < (X.ncol - 2); j++)
>> +    for (int i = 0; i < X.nrow; i++){
>> +        for (int j = 0; j < (X.ncol - 2); j++){
>>               nX[i * nX.ncol + j] = X[i * X.ncol + j];
>> +        }
>> +    }
>>   
>>       for (int i = 0; i < nX.nrow; i++)
>>       {
>>
>> Modified: branches/ProbABEL-0.50/src/regdata.cpp
>> ===================================================================
>> --- branches/ProbABEL-0.50/src/regdata.cpp	2014-02-04 11:27:35 UTC (rev 1593)
>> +++ branches/ProbABEL-0.50/src/regdata.cpp	2014-02-04 21:00:43 UTC (rev 1594)
>> @@ -14,7 +14,7 @@
>>   #include "fvlib/frversion.h"
>>   #include "fvlib/Logger.h"
>>   #include "fvlib/Transposer.h"
>> -
>> +#include <algorithm> // STL algoritms
>>   #include "regdata.h"
>>   
>>   regdata::regdata()
>> @@ -203,7 +203,6 @@
>>       int dim2X = X.ncol;
>>       (to.X).reinit(to.nids, dim2X);
>>       (to.Y).reinit(to.nids, dim2Y);
>> -
>>       int j = 0;
>>       for (int i = 0; i < nids; i++)
>>       {
>> @@ -223,13 +222,10 @@
>>       }
>>   
>>       // delete [] to.masked_data;
>> -    to.masked_data = new unsigned short int[to.nids];
>> -    for (int i = 0; i < to.nids; i++)
>> -    {
>> -        to.masked_data[i] = 0;
>> -    }
>> -    // std::cout << "get_unmasked: " << to.nids << " "
>> -    //           << dim2X << " " << dim2Y << "\n";
>> +    const int arr_size = nids;
>> +    to.masked_data = new unsigned short int[arr_size];
>> +    std::copy(masked_data, masked_data+arr_size,to.masked_data);
>> +
>>       return (to);
>>   }
>>   
>>
>> _______________________________________________
>> Genabel-commits mailing list
>> Genabel-commits at lists.r-forge.r-project.org
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
>>
>
>
> _______________________________________________
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140206/d09c8764/attachment.html>

From kooyman at gmail.com  Thu Feb  6 18:00:42 2014
From: kooyman at gmail.com (Maarten Kooyman)
Date: Thu, 06 Feb 2014 18:00:42 +0100
Subject: [GenABEL-dev] [Genabel-commits] r1596 - branches/ProbABEL-0.50
In-Reply-To: <52F2AA3E.6080200@karssen.org>
References: <20140205202601.BAB3E186B9D@r-forge.r-project.org>
 <52F2AA3E.6080200@karssen.org>
Message-ID: <52F3BFBA.4030805@gmail.com>

Hi Lennart,

It does not look that -Wextra produce any extra warnings.

On 05-02-14 22:16, L.C. Karssen wrote:
> Hi Maarten,
>
> I haven't yet compiled this branch, but will be interested to see what
> -Wextra gives in terms of warnings.
>
> One more question below:
>
> On 05-02-14 21:26, noreply at r-forge.r-project.org wrote:
>> Author: maartenk
>> Date: 2014-02-05 21:26:01 +0100 (Wed, 05 Feb 2014)
>> New Revision: 1596
>>
>> Modified:
>>     branches/ProbABEL-0.50/configure.ac
>> Log:
>> added  CXXflags -DEIGENNO_DEBUG :removes checks for conformability (which gives a speedup in eigen)
>> added  CXXflags -Wextra: adds adittional check (see http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html for more info)
>>
>>
>>
>>
>>
>> Modified: branches/ProbABEL-0.50/configure.ac
>> ===================================================================
>> --- branches/ProbABEL-0.50/configure.ac	2014-02-05 16:35:59 UTC (rev 1595)
>> +++ branches/ProbABEL-0.50/configure.ac	2014-02-05 20:26:01 UTC (rev 1596)
>> @@ -21,9 +21,10 @@
>>   AC_PROG_LN_S
>>   if test -z "$CXXFLAGS"; then
>>      # User did not set CXXFLAGS, so we can put in our own defaults
>> -    CXXFLAGS="-g -O2"
>> +    CXXFLAGS="-g -O2 -DEIGEN_NO_DEBUG -Wextra"
>>   fi
>> -if test -z "$CPPFLAGS"; then
>> +if test -z "$
>> +"; then
>>      # User did not set CPPFLAGS, so we can put in our own defaults
>>       CPPFLAGS="-Wall"
>>   fi
>>
> Is this correct? It seems that the name of the CPPFLAGS variable was
> removed and
> ";then
> moved to a new line.

Well spotted! This was a copy paste error. I committed a fix.

  I think there should be a configure option to use or use not -DNDEBUG: 
disabling this is inconvenient when developing and debugging. I think 
something like ./configure --develop-flags. I don't have  any experience 
with autotools, so this might take a while.
>
>
>
> Thanks,
>
> Lennart.
>
>> _______________________________________________
>> Genabel-commits mailing list
>> Genabel-commits at lists.r-forge.r-project.org
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
>>
>
>
> _______________________________________________
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140206/dff2f5f6/attachment.html>

From lennart at karssen.org  Sun Feb  9 11:27:07 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Sun, 09 Feb 2014 11:27:07 +0100
Subject: [GenABEL-dev] Preparing for ProbABEL 0.4.3 release
Message-ID: <52F757FB.8070707@karssen.org>

Dear list,

I am currently preparing the 0.4.3 release of ProbABEL. If you have any
updates/fixes/etc. that you would like to have in this release, please
let me know. I aim to do the release somewhere in the coming week.


Best regards,

Lennart Karssen.

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140209/51b0abdb/attachment.sig>

From lennart at karssen.org  Sun Feb  9 14:13:29 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Sun, 09 Feb 2014 14:13:29 +0100
Subject: [GenABEL-dev] Design: Versions for filevector library,
	change symlinks
Message-ID: <52F77EF9.5010504@karssen.org>

Dear list,

As you may know I have been playing with the idea of creating filevector
(more specifically the fvlib directory) a separate library (see the
discussions at [1-3]). I also created a separate SVN branch to play
around with that idea. Basically this would mean that packages like
ProbABEL, DatABEL and VariABEL would get a dependency in the form of
what I call 'libfilevector'. Especially for the R packages in the
GenABEL suite this may not be very user-friendly because many of them
would not have admin rights (I assume) and therefore cannot simply
install operating system packages that contain libfilevector. Obviously
such a change to libfilevector would be a major one.

Apart from the discussions on the mailing list this topic came up again
a few weeks ago in a discussion with Maksim and Yurii. We discussed many
of the pros and cons (mostly revolving around user-friendliness vs.
technical advantages) and in the end we came up with the following:

Let's start to put versions on the filevector code. So far filevector
has never been released separately (always as part of the ABEL
packages). By putting a version number on the fv code and by making tags
of those releases in SVN we can solve several issues:
- It can be made clear which *ABEL package depends on which version of
the filevector code
- Improvements in the filevector code do not immediately affect the
'trunk' of the other packages
- This allows packages like ProbABEL to depend on a separate
libfilevector, whereas R packages can create a symlink to the tagged
code instead of filevector trunk (as is the present situation)

A final advantage may be that it becomes easier to treat filevector as
one of various file-format backends to the *ABELs. But this is something
I'll elaborate on in another e-mail to this list.


We would like to hear your opinions about this versioning of filevector
and tagging of these official filevector releases, as well as on the use
of symlinks to these tags by other packages (this may not be ideal, but
it solves the user-friendliness issue, while allowing the creation and
release of libfilevector).


Hope to hear from you!

Lennart.



[1]http://lists.r-forge.r-project.org/pipermail/genabel-devel/2013-July/000714.html
[2]
http://lists.r-forge.r-project.org/pipermail/genabel-devel/2013-November/000779.html
[2]
http://lists.r-forge.r-project.org/pipermail/genabel-devel/2013-November/000797.html

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140209/5ba08be1/attachment.sig>

From lennart at karssen.org  Sun Feb  9 22:31:40 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Sun, 09 Feb 2014 22:31:40 +0100
Subject: [GenABEL-dev] Score option in ProbABEL
Message-ID: <52F7F3BC.6070203@karssen.org>

Dear list,

I was wondering if one of the developers of the 'early days' knows why
the 'score' option has been disabled since v 0.1-6 of ProbABEL (see the
output of e.g. 'palinear -p p.phe -i i.info -d d.dose --score':

probabel v. 0.4.2
(C) Yurii Aulchenko, Lennart C. Karssen, Maksim Struchalin, EMCR

Using EIGEN version 3.2.0 for matrix operations
option --score suppressed from v 0.1-6


If this option is no longer used we could remove its code to reduce
clutter.

Looking forward to hearing from you!

Lennart.
-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140209/ca501c89/attachment.sig>

From lennart at karssen.org  Tue Feb 11 15:28:44 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Tue, 11 Feb 2014 15:28:44 +0100
Subject: [GenABEL-dev] Fwd: Re: [Genabel-commits] r1606 -
	pkg/VariABEL/src/VARlib
In-Reply-To: <52F9F5D8.3010501@karssen.org>
References: <52F9F5D8.3010501@karssen.org>
Message-ID: <52FA339C.5040200@karssen.org>

Dear Maksim, all,

The e-mail below should have gone to the devel list, not the commit list.


Best,

Lennart.


-------- Original Message --------
Subject: Re: [Genabel-commits] r1606 - pkg/VariABEL/src/VARlib
Date: Tue, 11 Feb 2014 11:05:12 +0100
From: L.C. Karssen <lennart at karssen.org>
Reply-To: genabel-devel at lists.r-forge.r-project.org
To: genabel-commits at lists.r-forge.r-project.org

Thanks Maksim, for adding the 2df test!

One question below:


On 11-02-14 09:29, noreply at r-forge.r-project.org wrote:
> Author: maksim
> Date: 2014-02-11 09:29:43 +0100 (Tue, 11 Feb 2014)
> New Revision: 1606
> 
> Modified:
>    pkg/VariABEL/src/VARlib/var_homogeneity_tests.h
> Log:
> Fixed
> 
> Modified: pkg/VariABEL/src/VARlib/var_homogeneity_tests.h
> ===================================================================
> --- pkg/VariABEL/src/VARlib/var_homogeneity_tests.h	2014-02-11 07:55:37 UTC (rev 1605)
> +++ pkg/VariABEL/src/VARlib/var_homogeneity_tests.h	2014-02-11 08:29:43 UTC (rev 1606)
> @@ -55,7 +55,7 @@
>  double var(my_small_vector vec);
>  
>  extern "C" {
> -void variance_homogeneity_test_C(double* snp, double *trait, double *design_matrix_geno_means, double *design_matrix, int *p, long unsigned* nids, double *betas, double *se, double * chi2, int * df, double * residuals, int *analys_type_, int * is_trait_na, int *testname, /*auxiliary variables:*/ double *qty, int *jpvt, double *qraux, double *work, double *v, double *x_for_ch2inv);
> +void variance_homogeneity_test_C(double* snp, double *trait, double *design_matrix_geno_means, double *design_matrix, int *p, long unsigned* nids, double *betas, double *se, double * chi2, int * df, double * residuals, int *analys_type_, int * is_trait_na, int *testname, /*auxiliary variables:*/ double *qty, int *jpvt, double *qraux, double *work, double *v, double *x_for_ch2inv,
> unsigned step);

I am wondering about this last argument: 'unsigned step'. Shouldn't that
be 'unsigned int step'?


Best,

Lennart.


>  }
>  
>  #endif
> 
> _______________________________________________
> Genabel-commits mailing list
> Genabel-commits at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
> 

-- 
-----------------------------------------------------------------
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org

Stuur mij aub geen Word of Powerpoint bestanden!
Zie http://www.gnu.org/philosophy/no-word-attachments.nl.html
------------------------------------------------------------------




-------------- next part --------------
_______________________________________________
Genabel-commits mailing list
Genabel-commits at lists.r-forge.r-project.org
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140211/57d22ae6/attachment.sig>

From m.v.struchalin at mail.ru  Tue Feb 11 16:06:09 2014
From: m.v.struchalin at mail.ru (Maksim Struchalin)
Date: Tue, 11 Feb 2014 22:06:09 +0700
Subject: [GenABEL-dev] Fwd: Re: [Genabel-commits] r1606 -
	pkg/VariABEL/src/VARlib
In-Reply-To: <52FA339C.5040200@karssen.org>
References: <52F9F5D8.3010501@karssen.org> <52FA339C.5040200@karssen.org>
Message-ID: <52FA3C61.2050204@mail.ru>

Dear Lennart,

'unsigned' stands for 'unsigned int' but I think you are right, it is 
better to write a full type name.

best,
Maksim

On 11/02/2014 21:28, L.C. Karssen wrote:
> Dear Maksim, all,
>
> The e-mail below should have gone to the devel list, not the commit list.
>
>
> Best,
>
> Lennart.
>
>
> -------- Original Message --------
> Subject: Re: [Genabel-commits] r1606 - pkg/VariABEL/src/VARlib
> Date: Tue, 11 Feb 2014 11:05:12 +0100
> From: L.C. Karssen <lennart at karssen.org>
> Reply-To: genabel-devel at lists.r-forge.r-project.org
> To: genabel-commits at lists.r-forge.r-project.org
>
> Thanks Maksim, for adding the 2df test!
>
> One question below:
>
>
> On 11-02-14 09:29, noreply at r-forge.r-project.org wrote:
>> Author: maksim
>> Date: 2014-02-11 09:29:43 +0100 (Tue, 11 Feb 2014)
>> New Revision: 1606
>>
>> Modified:
>>     pkg/VariABEL/src/VARlib/var_homogeneity_tests.h
>> Log:
>> Fixed
>>
>> Modified: pkg/VariABEL/src/VARlib/var_homogeneity_tests.h
>> ===================================================================
>> --- pkg/VariABEL/src/VARlib/var_homogeneity_tests.h	2014-02-11 07:55:37 UTC (rev 1605)
>> +++ pkg/VariABEL/src/VARlib/var_homogeneity_tests.h	2014-02-11 08:29:43 UTC (rev 1606)
>> @@ -55,7 +55,7 @@
>>   double var(my_small_vector vec);
>>   
>>   extern "C" {
>> -void variance_homogeneity_test_C(double* snp, double *trait, double *design_matrix_geno_means, double *design_matrix, int *p, long unsigned* nids, double *betas, double *se, double * chi2, int * df, double * residuals, int *analys_type_, int * is_trait_na, int *testname, /*auxiliary variables:*/ double *qty, int *jpvt, double *qraux, double *work, double *v, double *x_for_ch2inv);
>> +void variance_homogeneity_test_C(double* snp, double *trait, double *design_matrix_geno_means, double *design_matrix, int *p, long unsigned* nids, double *betas, double *se, double * chi2, int * df, double * residuals, int *analys_type_, int * is_trait_na, int *testname, /*auxiliary variables:*/ double *qty, int *jpvt, double *qraux, double *work, double *v, double *x_for_ch2inv,
>> unsigned step);
> I am wondering about this last argument: 'unsigned step'. Shouldn't that
> be 'unsigned int step'?
>
>
> Best,
>
> Lennart.
>
>
>>   }
>>   
>>   #endif
>>
>> _______________________________________________
>> Genabel-commits mailing list
>> Genabel-commits at lists.r-forge.r-project.org
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
>>
>
>
> _______________________________________________
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140211/8047829a/attachment.html>

From m.v.struchalin at mail.ru  Wed Feb 12 05:22:36 2014
From: m.v.struchalin at mail.ru (Maksim Struchalin)
Date: Wed, 12 Feb 2014 11:22:36 +0700
Subject: [GenABEL-dev] 'free' genetic model, test on significance
In-Reply-To: <D9430932-1A37-4DB8-AE6C-471DABE56BE7@gmail.com>
References: <633975060.12301.1391763223671.JavaMail.mobile-sync@veei10>	<8399649340900060896@unknownmsgid>	<52F4A3B8.3010409@mail.ru>	<FD553326-1CC3-48B1-BCCB-6E3326067D2C@gmail.com>
 <CAPF08KsLMQqZiv_muPFH+Bz1ANfc3_o0hQJmXX0LGESG_8J1LQ@mail.gmail.com>
 <52F9BBAB.4090806@mail.ru> <D9430932-1A37-4DB8-AE6C-471DABE56BE7@gmail.com>
Message-ID: <52FAF70C.1030204@mail.ru>

Hi All from the devel list,

Sodbo (cc) is going to test GxG and GxE interactions in his date using 
genotypic variance heterogeneity comparison (VariABEL package). 
Currently, VariABEL can test only genotyped date or imputed dosages and 
for only three genetic models (additive, dominant, recessive). But Sodbo 
would like to make a test which does not assume any genetic model. As I 
proposed, for this, we need to run the model 'trait_residuals2 ~ P_AB + 
P_BB', where trait_residuals2 is the squared residuals of a trait 
adjusted on covariates (environmental factors and others), P_AB and P_BB 
are probabilities of having AB and BB genotypes by a given individual 
(imputed SNPs from MACH). Now, I am in charge of implementing such a 
test in VariABEL and thinking about how to test the significance jointly 
for P_AB and P_BB. Because of its simIicity, I propose to implement a 
test expained in page 46 of 
http://academic.reed.edu/economics/parker/s10/312/notes/Notes5.pdf in a 
section 'Alternative calculation of F under classical assumption'. This 
test gives a 'liiiiiiitle' bit overestimated p-value if copmare to the 
test implemented in summary.lm() in R.

We would like to ask an advice on this issue from all of you. Does this 
regression model and the test on significance allows us to test a SNP 
without a prior assumption on genetic model?
Any comments are welcome.

best,
Maksim



On 12/02/2014 03:48, Yury Aulchenko wrote:
> Hi Maksim,
>
> It may be good if you could formulate your ideas concerning the 2df 
> test and send these to the devel list - then we will have a chance to 
> pick on the brains of people who are very knowledgable about 
> statistics (e.g. Xia and William)
>
> Let us know if you could do this.
>
> best wishes,
> Yurii
>
> On Feb 11, 2014, at 06:56, Maksim Struchalin <m.v.struchalin at mail.ru 
> <mailto:m.v.struchalin at mail.ru>> wrote:
>
>> ?????? ????? ? ???,
>>
>> ???????? ?? ?????? ????? ?????? ??? ????? ?? ??????. ? ???? ??????? 
>> ????? ?????????:
>> - beta_AB, sigma_AB - ?????? ? ?????? ??? AB
>> - beta_BB, sigma_BB - ?????? ? ?????? ??? BB
>> - chisq_AB, chisq_BB - ??-???????? ??? AB ? BB
>> - wald test ??? AB ? BB
>> - F ???? ?? ??????? (sigma_0 - sigma_1), ??? sigma_0 - ????????? 
>> '???????? ? ????????', ? sigma_1 - ????????? ???????? ????? 
>> ?????????. ???? ???? ??????? ?????????? ?????????????? ????????????? 
>> AB ? BB.
>> (??. ???. 46, ????? "Alternative calculation..." ? 
>> http://academic.reed.edu/economics/parker/s10/312/notes/Notes5.pdf).
>>
>> ? ???? ????????????? ???? ????? ??? AB ? BB, 2df ???? ???? ? ????? ?????.
>>
>> ??????
>>
>>
>> On 09/02/2014 20:12, Sodbo Sharapov wrote:
>>> ?????? ????!
>>>
>>> ?????? ???????? ?? ??????:https:
>>> https://www.dropbox.com/s/w6gayszobvgzr7o/Ara_2_Maksim.rdata.zip
>>>
>>> ??????? ???? - "X30.Mo98" - ???????????? ?????????. SNP ?????? ?? 
>>> 2?? ????????? (?? ??? ???? ????). ? ara_gwaa.rdata ???????? ??????? 
>>> ? ???????? ????????.
>>> ? prob_ara_2_Maksim.rdata ???????? ?????????????? ???????? - ? 
>>> ??????? ?????????????? ?????????. ????? ??? ???? 10 ???????? SNP - 
>>> ??? ????.
>>> ? prob_ara_2_Maksim.fvd/fvi - ???? ????? ? DatABEL ??????? - ??? 
>>> ?????????????? SNP ?????? ????? ?????
>>> ????? ??????? - ??????=)
>>>
>>>
>>> 7 ??????? 2014 ?., 20:03 ???????????? Yury Aulchenko 
>>> <yurii.aulchenko at gmail.com <mailto:yurii.aulchenko at gmail.com>> ???????:
>>>
>>>     Sodbo, togda navernoye stoit snachala sgenerit' danniye i
>>>     poslat' Maximu, a potom sozvonites', obsudite testy
>>>
>>>     nu i napominayu chto v principe eto horosho by sdelat' za
>>>     nedelyu-dve konechno do stadii chtob ty mog nachat' analiz
>>>     real'nih dannih :)
>>>
>>>     Y
>>>
>>>     On Feb 7, 2014, at 10:13, Maksim Struchalin
>>>     <m.v.struchalin at mail.ru <mailto:m.v.struchalin at mail.ru>> wrote:
>>>
>>>     > ??? ???? ?? ????????????. ????? ????? ? ????? ??????.
>>>     > ??????
>>>     >
>>>     > On 07/02/2014 16:06, Yurii Aulchenko wrote:
>>>     >> ??????, ?????, ? ?????? - ????? ????? ??????? ????????? ????
>>>     ?????? ??? ?????
>>>     >>
>>>     >> ----------------------
>>>     >> Yurii Aulchenko
>>>     >> (sent from mobile device)
>>>     >>
>>>     >>> On Feb 7, 2014, at 9:53 AM, Maksim Struchalin
>>>     <m.v.struchalin at mail.ru <mailto:m.v.struchalin at mail.ru>> wrote:
>>>     >>>
>>>     >>> ??????, ?????,
>>>     >>>
>>>     >>> ? ?????? ???????? 2df ??????. ?????? ??? ??????? ???????
>>>     ?????? ???
>>>     >>> ??????? ?? ??????????? ????? ???? ????. ???? ??????????
>>>     ?????? ?????? +
>>>     >>> ? ??? ???? ???????????? ??? ??????? ?????.
>>>     >>>
>>>     >>> ???? ??????? :-),
>>>     >>> ??????
>>>     >
>>>
>>>
>>>
>>>
>>> -- 
>>> ___________________________________
>>> _
>>> _With best regards
>>>
>>> Sodbo Zh. Sharapov
>>> Phone:  +79831347688
>>> Email: sharapovsodbo at gmail.com <mailto:sharapovsodbo at gmail.com>
>>> sharapov at bionet.nsc.ru <mailto:sharapov at bionet.nsc.ru>
>>> Skype:   sharapovsodbo
>>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140212/37dd604a/attachment-0001.html>

From lennart at karssen.org  Wed Feb 12 09:43:40 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Wed, 12 Feb 2014 09:43:40 +0100
Subject: [GenABEL-dev] Fwd: Re: [Genabel-commits] r1606 -
	pkg/VariABEL/src/VARlib
In-Reply-To: <52FA3C61.2050204@mail.ru>
References: <52F9F5D8.3010501@karssen.org> <52FA339C.5040200@karssen.org>
 <52FA3C61.2050204@mail.ru>
Message-ID: <52FB343C.80209@karssen.org>

Thanks Maksim!



Lennart.

On 11-02-14 16:06, Maksim Struchalin wrote:
> Dear Lennart,
> 
> 'unsigned' stands for 'unsigned int' but I think you are right, it is
> better to write a full type name.
> 
> best,
> Maksim
> 
> On 11/02/2014 21:28, L.C. Karssen wrote:
>> Dear Maksim, all,
>>
>> The e-mail below should have gone to the devel list, not the commit list.
>>
>>
>> Best,
>>
>> Lennart.
>>
>>
>> -------- Original Message --------
>> Subject: Re: [Genabel-commits] r1606 - pkg/VariABEL/src/VARlib
>> Date: Tue, 11 Feb 2014 11:05:12 +0100
>> From: L.C. Karssen <lennart at karssen.org>
>> Reply-To: genabel-devel at lists.r-forge.r-project.org
>> To: genabel-commits at lists.r-forge.r-project.org
>>
>> Thanks Maksim, for adding the 2df test!
>>
>> One question below:
>>
>>
>> On 11-02-14 09:29, noreply at r-forge.r-project.org wrote:
>>> Author: maksim
>>> Date: 2014-02-11 09:29:43 +0100 (Tue, 11 Feb 2014)
>>> New Revision: 1606
>>>
>>> Modified:
>>>    pkg/VariABEL/src/VARlib/var_homogeneity_tests.h
>>> Log:
>>> Fixed
>>>
>>> Modified: pkg/VariABEL/src/VARlib/var_homogeneity_tests.h
>>> ===================================================================
>>> --- pkg/VariABEL/src/VARlib/var_homogeneity_tests.h	2014-02-11 07:55:37 UTC (rev 1605)
>>> +++ pkg/VariABEL/src/VARlib/var_homogeneity_tests.h	2014-02-11 08:29:43 UTC (rev 1606)
>>> @@ -55,7 +55,7 @@
>>>  double var(my_small_vector vec);
>>>  
>>>  extern "C" {
>>> -void variance_homogeneity_test_C(double* snp, double *trait, double *design_matrix_geno_means, double *design_matrix, int *p, long unsigned* nids, double *betas, double *se, double * chi2, int * df, double * residuals, int *analys_type_, int * is_trait_na, int *testname, /*auxiliary variables:*/ double *qty, int *jpvt, double *qraux, double *work, double *v, double *x_for_ch2inv);
>>> +void variance_homogeneity_test_C(double* snp, double *trait, double *design_matrix_geno_means, double *design_matrix, int *p, long unsigned* nids, double *betas, double *se, double * chi2, int * df, double * residuals, int *analys_type_, int * is_trait_na, int *testname, /*auxiliary variables:*/ double *qty, int *jpvt, double *qraux, double *work, double *v, double *x_for_ch2inv,
>>> unsigned step);
>> I am wondering about this last argument: 'unsigned step'. Shouldn't that
>> be 'unsigned int step'?
>>
>>
>> Best,
>>
>> Lennart.
>>
>>
>>>  }
>>>  
>>>  #endif
>>>
>>> _______________________________________________
>>> Genabel-commits mailing list
>>> Genabel-commits at lists.r-forge.r-project.org
>>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
>>>
>>
>>
>> _______________________________________________
>> genabel-devel mailing list
>> genabel-devel at lists.r-forge.r-project.org
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
> 
> 
> 
> _______________________________________________
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
> 

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140212/d1ad0966/attachment.sig>

From 340071rs at student.eur.nl  Wed Feb 12 13:24:00 2014
From: 340071rs at student.eur.nl (Rebecca Steenaard)
Date: Wed, 12 Feb 2014 12:24:00 +0000
Subject: [GenABEL-dev] Bug in GenABEL-tutorial.pdf
Message-ID: <d34995d47afe41dca5102bc9eee3e2de@AM3PR02MB067.eurprd02.prod.outlook.com>



Dear,



Yurri Aulchenko told me to send this email to correct a bug in the GenABEL-tutorial pdf used during the GWAS course of NIHES. On page 241 the following line is used:

Thus the pooled estimate of Odds Ratio from association between type 2diabetes and "Ala" allele is

Given the rest of the exercise and the answers, the "Ala" should be "Pro", otherwise the answer is inverted.



With kind regards,

Rebecca Steenaard
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140212/f487647c/attachment.html>

From lennart at karssen.org  Wed Feb 12 13:57:44 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Wed, 12 Feb 2014 13:57:44 +0100
Subject: [GenABEL-dev] Bug in GenABEL-tutorial.pdf
In-Reply-To: <d34995d47afe41dca5102bc9eee3e2de@AM3PR02MB067.eurprd02.prod.outlook.com>
References: <d34995d47afe41dca5102bc9eee3e2de@AM3PR02MB067.eurprd02.prod.outlook.com>
Message-ID: <52FB6FC8.7050509@karssen.org>

Dear Rebecca,

Thank you for reporting this bug. I've reported it in our bug tracker at
https://r-forge.r-project.org/tracker/index.php?func=detail&aid=5367&group_id=505&atid=2058


Kind regards,

Lennart Karssen.

On 12-02-14 13:24, Rebecca Steenaard wrote:
>  
> 
> Dear,
> 
>  
> 
> Yurri Aulchenko told me to send this email to correct a bug in the
> GenABEL-tutorial pdf used during the GWAS course of NIHES. On page 241
> the following line is used:
> 
> Thus the pooled estimate of Odds Ratio from association between type
> 2diabetes and "Ala" allele is
> 
> Given the rest of the exercise and the answers, the "Ala" should be
> "Pro", otherwise the answer is inverted.
> 
>  
> 
> With kind regards,
> 
> Rebecca Steenaard
> 
> 
> 
> _______________________________________________
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
> 

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140212/b69e4ecc/attachment.sig>

From kooyman at gmail.com  Wed Feb 12 20:44:02 2014
From: kooyman at gmail.com (Maarten Kooyman)
Date: Wed, 12 Feb 2014 20:44:02 +0100
Subject: [GenABEL-dev] Fwd: Re: [Genabel-commits] r1606 -
	pkg/VariABEL/src/VARlib
In-Reply-To: <52FB343C.80209@karssen.org>
References: <52F9F5D8.3010501@karssen.org> <52FA339C.5040200@karssen.org>
 <52FA3C61.2050204@mail.ru> <52FB343C.80209@karssen.org>
Message-ID: <52FBCF02.4080008@gmail.com>

Dear All,

There is some nice information about (unsigned) integer in c++ on 
http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Integer_Types 
and there problems.

Kind regards,

Maarten Kooyman




On 12-02-14 09:43, L.C. Karssen wrote:
> Thanks Maksim!
>
>
>
> Lennart.
>
> On 11-02-14 16:06, Maksim Struchalin wrote:
>> Dear Lennart,
>>
>> 'unsigned' stands for 'unsigned int' but I think you are right, it is
>> better to write a full type name.
>>
>> best,
>> Maksim
>>
>> On 11/02/2014 21:28, L.C. Karssen wrote:
>>> Dear Maksim, all,
>>>
>>> The e-mail below should have gone to the devel list, not the commit list.
>>>
>>>
>>> Best,
>>>
>>> Lennart.
>>>
>>>
>>> -------- Original Message --------
>>> Subject: Re: [Genabel-commits] r1606 - pkg/VariABEL/src/VARlib
>>> Date: Tue, 11 Feb 2014 11:05:12 +0100
>>> From: L.C. Karssen <lennart at karssen.org>
>>> Reply-To: genabel-devel at lists.r-forge.r-project.org
>>> To: genabel-commits at lists.r-forge.r-project.org
>>>
>>> Thanks Maksim, for adding the 2df test!
>>>
>>> One question below:
>>>
>>>
>>> On 11-02-14 09:29, noreply at r-forge.r-project.org wrote:
>>>> Author: maksim
>>>> Date: 2014-02-11 09:29:43 +0100 (Tue, 11 Feb 2014)
>>>> New Revision: 1606
>>>>
>>>> Modified:
>>>>     pkg/VariABEL/src/VARlib/var_homogeneity_tests.h
>>>> Log:
>>>> Fixed
>>>>
>>>> Modified: pkg/VariABEL/src/VARlib/var_homogeneity_tests.h
>>>> ===================================================================
>>>> --- pkg/VariABEL/src/VARlib/var_homogeneity_tests.h	2014-02-11 07:55:37 UTC (rev 1605)
>>>> +++ pkg/VariABEL/src/VARlib/var_homogeneity_tests.h	2014-02-11 08:29:43 UTC (rev 1606)
>>>> @@ -55,7 +55,7 @@
>>>>   double var(my_small_vector vec);
>>>>   
>>>>   extern "C" {
>>>> -void variance_homogeneity_test_C(double* snp, double *trait, double *design_matrix_geno_means, double *design_matrix, int *p, long unsigned* nids, double *betas, double *se, double * chi2, int * df, double * residuals, int *analys_type_, int * is_trait_na, int *testname, /*auxiliary variables:*/ double *qty, int *jpvt, double *qraux, double *work, double *v, double *x_for_ch2inv);
>>>> +void variance_homogeneity_test_C(double* snp, double *trait, double *design_matrix_geno_means, double *design_matrix, int *p, long unsigned* nids, double *betas, double *se, double * chi2, int * df, double * residuals, int *analys_type_, int * is_trait_na, int *testname, /*auxiliary variables:*/ double *qty, int *jpvt, double *qraux, double *work, double *v, double *x_for_ch2inv,
>>>> unsigned step);
>>> I am wondering about this last argument: 'unsigned step'. Shouldn't that
>>> be 'unsigned int step'?
>>>
>>>
>>> Best,
>>>
>>> Lennart.
>>>
>>>
>>>>   }
>>>>   
>>>>   #endif
>>>>
>>>> _______________________________________________
>>>> Genabel-commits mailing list
>>>> Genabel-commits at lists.r-forge.r-project.org
>>>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
>>>>
>>>
>>> _______________________________________________
>>> genabel-devel mailing list
>>> genabel-devel at lists.r-forge.r-project.org
>>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
>>
>>
>> _______________________________________________
>> genabel-devel mailing list
>> genabel-devel at lists.r-forge.r-project.org
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
>>
>
>
> _______________________________________________
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140212/dcf7602b/attachment.html>

From lennart at karssen.org  Fri Feb 14 11:32:20 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Fri, 14 Feb 2014 11:32:20 +0100
Subject: [GenABEL-dev] [Genabel-commits] r1609 - pkg/OmicABELnoMM/src
In-Reply-To: <20140213204258.0DD561849B8@r-forge.r-project.org>
References: <20140213204258.0DD561849B8@r-forge.r-project.org>
Message-ID: <52FDF0B4.10008@karssen.org>

Hi Alvaro,

Thanks for your contribution and congratulations on your first commit in
the GenABEL project.

Glad to see that the number of has been reduced a lot! I still get a few
errors after running
 ./configure  LDFLAGS="-L/usr/lib/openblas-base"
 make -j4
but they seem to be minor and easy to fix. Have a look at
http://jenkins.genabel.org/jenkins/job/OmicABELnoMM/15/warnings23Result/
for the warning messages and corresponding lines in the code.
Also try to have a look at the CPPcheck results in Jenkins. It warns
about three (possible) memory leaks.


Thanks a lot and looking forward to your next commits.


Lennart.


On 13-02-14 21:42, noreply at r-forge.r-project.org wrote:
> Author: afrank
> Date: 2014-02-13 21:42:52 +0100 (Thu, 13 Feb 2014)
> New Revision: 1609
> 
> Modified:
>    pkg/OmicABELnoMM/src/AIOwrapper.cpp
>    pkg/OmicABELnoMM/src/Algorithm.cpp
>    pkg/OmicABELnoMM/src/Algorithm.h
>    pkg/OmicABELnoMM/src/Definitions.h
>    pkg/OmicABELnoMM/src/Utility.cpp
>    pkg/OmicABELnoMM/src/Utility.h
>    pkg/OmicABELnoMM/src/main.cpp
> Log:
> Warnings from c/c++ inconsistencies and -Wall fixed. Legacy code from unused functionality removed. A few bugs with indices fixed. Stable code.
> 
> Modified: pkg/OmicABELnoMM/src/AIOwrapper.cpp
> ===================================================================
> --- pkg/OmicABELnoMM/src/AIOwrapper.cpp	2014-02-13 09:24:32 UTC (rev 1608)
> +++ pkg/OmicABELnoMM/src/AIOwrapper.cpp	2014-02-13 20:42:52 UTC (rev 1609)
> @@ -29,9 +29,9 @@
>  
>      Fhandler->fakefiles = params.use_fake_files;
>  
> -    databel_fvi* Yfvi;
> -    databel_fvi* ALfvi;
> -    databel_fvi* ARfvi;
> +//    databel_fvi* Yfvi;
> +//    databel_fvi* ALfvi;
> +//    databel_fvi* ARfvi;
>  
>      if (!Fhandler->fakefiles)
>      {
> @@ -49,7 +49,7 @@
>          params.l = ALfvi->fvi_header.numVariables;
>  
>          //block size to keep mem under 1 gigabyte
> -        int opt_block = params.n / (4*1000^3) * (1/(2*params.r));
> +        //int opt_block = params.n / (4*1000^3) * (1/(2*params.r));
>          int opt_tb = 1000;
>          int opt_mb = 1000;
>  
> @@ -81,7 +81,7 @@
>  void AIOwrapper::finalize()
>  {
>      //cout << "f";
> -    void *status;
> +    //void *status;
>  
>      Fhandler->not_done = false;
>      pthread_mutex_lock(&(Fhandler->m_more));
> @@ -368,13 +368,15 @@
>      }
>  //
>  //            //!induce realistic fileread delay
> +
> +    return 0;
>  }
>  
>  
>  void AIOwrapper::load_ARblock(type_precision** Ar, int &Ar_blockSize)
>  {
> -    int status;
> -    int createstatus = 0;
> +    //int status;
> +    //int createstatus = 0;
>      //cout<<"^";
>  
>      while (Fhandler->ar_full_buffers.empty())
> @@ -428,8 +430,8 @@
>  
>  void AIOwrapper::load_Yblock(type_precision** Y, int &y_blockSize)
>  {
> -    int status;
> -    int createstatus = 0;
> +    //int status;
> +    //int createstatus = 0;
>  
>      while (Fhandler->full_buffers.empty())
>      {
> @@ -528,7 +530,7 @@
>  
>  void AIOwrapper::reset_Y()
>  {
> -    void *status;
> +    //void *status;
>  
>      Fhandler->seed = 1337;
>  
> @@ -569,7 +571,7 @@
>  
>  void AIOwrapper::reset_AR()
>  {
> -    void *status;
> +    //void *status;
>  
>  
>      //cout << "ra" << flush;
> @@ -697,7 +699,7 @@
>  FILE * AIOwrapper::fgls_fopen( const char *path, const char *mode )
>  {
>          FILE * f;
> -        char err[100];
> +        //char err[100];
>  
>          f = fopen(path, mode );
>          if ( f == NULL )
> 
> Modified: pkg/OmicABELnoMM/src/Algorithm.cpp
> ===================================================================
> --- pkg/OmicABELnoMM/src/Algorithm.cpp	2014-02-13 09:24:32 UTC (rev 1608)
> +++ pkg/OmicABELnoMM/src/Algorithm.cpp	2014-02-13 20:42:52 UTC (rev 1609)
> @@ -16,30 +16,12 @@
>  {
>      switch (type)
>      {
> -    case FULL_NEQ:
> -        fullNEQ(params, out);
> -        break;
> -    case P_NEQ:
> -        partialNEQ(params, out);
> -        break;
> -    case P_NEQ_B_OPT:
> -        partialNEQ_Blocked_STL(params, out);
> -        break;
> -    case FULL_QR:
> -        fullQR(params, out);
> -        break;
> -    case P_QR:
> -        partialQR(params, out);
> -        break;
> -    case P_QR_B_OPT:
> -        partialQR_Blocked_Rtl(params, out);
> -        break;
> -    case P_NEQ_B_OPT_MD:
> -        partialNEQ_Blocked_STL_MD(params, out);
> -        break;
> +        case P_NEQ_B_OPT_MD:
> +            partialNEQ_Blocked_STL_MD(params, out);
> +            break;
>  
> -    default:
> -        break;
> +        default:
> +            break;
>      }
>  }
>  
> @@ -74,7 +56,7 @@
>                                 int dim2_b, int dim1_b_bot)
>  {
>      // memcpy are faster version of the fors
> -    int i, k, w, top_idx, bot_idx, max = dim1_b*dim2_b;
> +    int i, k, w, top_idx, bot_idx;
>      int size;
>      top_idx = 0;
>      bot_idx = 0;
> @@ -111,7 +93,7 @@
>                             type_precision* bot, int dim1_QY,
>                             int dim2_QY, int dim1_qy_bot, int bot_blocks )
>  {
> -    int i, k, w, top_idx, bot_idx, max = dim1_QY*dim2_QY;
> +    int i, k, w, top_idx, bot_idx;
>      top_idx = 0;
>      bot_idx = 0;
>      for (k = 0; k < dim2_QY; k++)
> @@ -326,12 +308,9 @@
>      blas_set_num_threads(max_threads);
>  
>  
> -    type_precision *Ytemp;
> -    lapack_int info, n, lda, ldy, l, r, k, p;
> +    //type_precision *Ytemp;
> +    lapack_int info, n, lda, l, r, p;
>  
> -    int i, j, w;
> -    int m;
> -
>      cputime_type start_tick, start_tick2, end_tick;
>  
>      AIOfile.initialize(params);
> @@ -349,12 +328,12 @@
>      int y_iters = (y_amount + y_block_size - 1) / y_block_size;
>  
>  
> -    lda = n; ldy = n;
> -    k = p;
> +    lda = n;
>  
> +
>      for (int j = 0; j < y_iters; j++)
>      {
> -        if (y_iters >= 40 && (i%(y_iters/40)) == 0)
> +        if (y_iters >= 40 && (j%(y_iters/40)) == 0)
>          {
>              cout << "*" << flush;
>          }
> @@ -389,7 +368,9 @@
>  
>  
>      AIOfile.load_AL(&backupAL);
> -    int total_al_nans = replace_nans(0, backupAL, n, l);
> +    //int total_al_nans = replace_nans(0, backupAL, n, l);
> +    replace_nans(0, backupAL, n, l);
> +
>      copy_vec(backupAL, AL, n*l);
>  
>  
> @@ -402,7 +383,7 @@
>  
>      for (int j = 0; j < y_iters; j++)
>      {
> -        if (y_iters >= 40 && (i%(y_iters/40)) == 0)
> +        if (y_iters >= 40 && (j%(y_iters/40)) == 0)
>          {
>              cout << AIOfile.io_overhead << flush;
>              AIOfile.io_overhead = "*";
> @@ -412,7 +393,8 @@
>  
>          list<long int> y_nan_idxs[y_block_size];
>  
> -        int total_y_nans = replace_nans(&y_nan_idxs[0], Y, n, y_block_size);
> +        //int total_y_nans = replace_nans(&y_nan_idxs[0], Y, n, y_block_size);
> +        replace_nans(&y_nan_idxs[0], Y, n, y_block_size);
>  
>  
>          //! Ay_top = AL'*Y
> @@ -424,7 +406,8 @@
>          for (int i = 0; i < a_iters; i++)
>          {
>              AIOfile.load_ARblock(&backupAR, a_block_size);
> -            int total_ar_nans = replace_nans(0, backupAR, n, a_block_size * r);
> +            //int total_ar_nans = replace_nans(0, backupAR, n, a_block_size * r);
> +            replace_nans(0, backupAR, n, a_block_size * r);
>              copy_vec(backupAR, AR, n *  r * a_block_size);
>  
>              get_ticks(start_tick2);
> @@ -501,7 +484,7 @@
>                                   &Ay[l+ii*p], r);
>  
>  
> -                        type_precision* B = Ay;
> +                        //type_precision* B = Ay;
>                          type_precision S[p * p];
>  
>                          //! Rebuild S
> @@ -564,1194 +547,187 @@
>  
>  //!*************************************************!//
>  
> -
> +//for reference
>  void Algorithm::partialNEQ_Blocked_STL(struct Settings params,
>                                         struct Outputs &out)
>  {
> -    int max_threads = params.threads;
> -
> -    srand(time(NULL));
> -
> -    blas_set_num_threads(max_threads);
> -
> -    type_precision *Ytemp;
> -    lapack_int info, n, lda, ldy, l, r, k, p;
> -
> -    int i, j, w;
> -    int m;
> -
> -    cputime_type start_tick, start_tick2, end_tick;
> -
> -    AIOfile.initialize(params);
> -
> -    n = params.n; l = params.l; r = params.r; p = l+r;
> -
> -    int y_amount = params.m;
> -    int y_block_size = params.mb;  // kk
> -
> -    int a_amount = params.t;
> -    int a_block_size = params.tb;
> -
> -    int a_iters = (a_amount + a_block_size - 1) / a_block_size;
> -
> -    lda = n; ldy = n;
> -    k = p;
> -
> -
> -
> -    type_precision* backupAL;
> -
> -    AIOfile.load_AL(&backupAL);
> -
> -
> -    type_precision Stl[l * l];
> -    type_precision Str[l * r * a_block_size];
> -
> -
> -    type_precision* Ay_top = new type_precision[l * y_amount];
> -    type_precision* Ay_bot = new type_precision[y_block_size * a_block_size * r];
> -    type_precision* A = new type_precision[n * p];
> -    type_precision* AR = new type_precision[n * r * a_block_size];
> -    type_precision* AL = new type_precision[n * l];
> -
> -    copy_vec(backupAL, AL, n * l);
> -    copy_vec(backupAL, A, n * l);
> -
> -
> -    int y_iters = y_amount/y_block_size;
> -
> -    type_precision* Y;
> -    type_precision* backupAR;
> -
> -    // printf("\n\n%%Computations\n%%");
> -
> -
> -    get_ticks(start_tick);
> -
> -
> -    //! Generate S
> -    cblas_ssyrk(CblasColMajor, CblasUpper, CblasTrans,
> -                l, n, 1.0, backupAL, lda, 0.0, Stl, l);
> -
> -    for (j = 0; j < y_iters; j++)
> -    {
> -        AIOfile.load_Yblock(&Y, y_block_size);
> -
> -        //! Ay_top = AL'*Y
> -        cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
> -                    l, y_block_size, n, 1.0, AL, n, Y, n, 0.0,
> -                    &Ay_top[j * l * y_block_size], l);
> -    }
> -
> -
> -    for (i = 0; i < a_iters; i++)
> -    {
> -        if (a_iters > 10 && (i%(a_iters/10)) == 0)
> -        {
> -            cout << "%" << flush;
> -        }
> -
> -
> -        AIOfile.load_ARblock(&backupAR, a_block_size);
> -        AIOfile.reset_Y();
> -        copy_vec(backupAR, AR, n*r*a_block_size);
> -
> -        // matlab_print_matrix("A", n, p, A);
> -        //! Generate Str
> -        cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
> -                    l, r * a_block_size, n, 1.0, AL, n, AR, n, 0.0, Str, l);
> -
> -        type_precision Sbr[r*r*a_block_size];
> -
> -        #pragma omp parallel for default(shared) schedule(static)
> -        for (int ii= 0; ii < a_block_size; ii++)
> -        {
> -            //! Generate Sbr
> -            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
> -                        r, r, n, 1.0, &AR[ii*r*n], n, &AR[ii*r*n],
> -                        n, 0.0, &Sbr[ii*r*r], r);
> -        }
> -
> -        type_precision Ay[y_block_size*p];
> -        type_precision S[p*p];
> -
> -
> -        for (j = 0; j < y_iters; j++)
> -        {
> -            if (a_iters < 10 && (y_iters < 10 || (j%(y_iters/10)) == 0))
> -            {
> -                cout << "%" << flush;
> -            }
> -
> -
> -            AIOfile.load_Yblock(&Y, y_block_size);
> -
> -            //! Ay_bot = AR'*Y
> -            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
> -                        r * a_block_size, y_block_size, n, 1.0, AR, n, Y, n,
> -                        0.0, Ay_bot, r * a_block_size);
> -
> -            #pragma omp parallel for private(S, Ay) default(shared) schedule(static)
> -            for (int ii= 0; ii < a_block_size; ii++)
> -            {
> -                // matlab_print_matrix("Y", n, y_block_size, Y);
> -
> -                //! Rebuild AY
> -
> -                for (int jj = 0; jj < y_block_size; jj++)
> -                {
> -                    copy_vec(&Ay_top[j*l*y_block_size+jj*l], &Ay[jj*p], l);
> -                    copy_vec(&Ay_bot[ii*r + jj*r*a_block_size], &Ay[l+jj*p], r);
> -                }
> -
> -                type_precision* B = Ay;
> -                // matlab_print_matrix("Ay_top", l, y_block_size,&Ay_top[j*l*y_block_size]);
> -                // matlab_print_matrix("Ay_bot", r, y_block_size, Ay_bot);
> -                // matlab_print_matrix("Ay", p, y_block_size, Ay);
> -
> -                //! Rebuild S
> -                build_S(S, Stl, &Str[ii*r*l], &Sbr[ii*r*r], l, r);
> -                // matlab_print_matrix("S", p, p, S);
> -
> -                //! b = S\Ay
> -                info = LAPACKE_sposv(STORAGE_TYPE, 'U', p, y_block_size,
> -                                     S, p, Ay, p);
> -                // assert(info == 0,"POSV");
> -
> -
> -                if (ForceCheck)
> -                {
> -                    check_result(backupAL, &AR[ii * r * n], n, p,
> -                                 y_block_size, r, Y, B);
> -                }
> -            }
> -        }
> -    }
> -
> -    get_ticks(end_tick);
> -    out.duration = ticks2sec(end_tick - start_tick, cpu_freq);
> -    // out.gflops = a_amount/1000.0*(n*l*r+n*r*r+y_amount*(n*r+p*p*p)))/1000.0/1000.0;
> -    out.gflops = gemm_flops(l, n, l, 0) + gemm_flops(l, n, y_amount, 0) +
> -                 a_amount * (gemm_flops(l, n, r, 0) +
> -                             gemm_flops(r, n, r, 0) +
> -                             gemm_flops(r, n, y_amount, 0) +
> -                             (y_amount/1000.0) *
> -                             (p * p * p / 3.0) / 1000.0/1000.0);
> -
> -
> -    AIOfile.finalize();
> -
> -    delete []Ay_top;
> -    delete []Ay_bot;
> -    delete []A;
> -    delete []AL;
> -    delete []AR;
> -}
> -
> -
> -void Algorithm::partialNEQ(struct Settings params, struct Outputs &out)
> -{
> -    int max_threads = params.threads;
> -
> -
> -    srand(time(NULL));
> -
> -
> -    blas_set_num_threads(max_threads);
> -
> -
> -    type_precision *Ytemp;
> -    lapack_int info, n, lda, ldy, l, r, k, p;
> -
> -    int i, j, w;
> -    int m;
> -
> -    cputime_type start_tick, start_tick2, end_tick;
> -    AIOfile.initialize(params);
> -
> -    n = params.n; l = params.l; r = params.r; p = l+r;
> -
> -    int y_amount = params.m;
> -    int y_block_size = params.mb;// kk
> -
> -    int a_amount = params.t;
> -    int a_block_size = 1;
> -
> -    int a_iters = a_amount/a_block_size;
> -
> -    lda = n; ldy = n;
> -    k = p;
> -
> -
> -
> -    type_precision* backupAL;
> -
> -    AIOfile.load_AL(&backupAL);
> -
> -    type_precision* S = new type_precision[p * p];
> -    type_precision Stl[l * l];
> -    type_precision Str[l * r];
> -    type_precision Sbr[r * r];
> -    type_precision* A = new type_precision[n * p];
> -    type_precision* AL = A;
> -    type_precision* AR = &A[l * n];
> -
> -    copy_vec(backupAL, AL, n*l);
> -
> -
> -    int y_iters = y_amount / y_block_size;
> -
> -    type_precision* Y;
> -    type_precision* backupAR;
> -
> -    // printf("\n\n%%Computations\n%%");
> -
> -    get_ticks(start_tick);
> -
> -    //! Generate S
> -    cblas_ssyrk(CblasColMajor, CblasUpper, CblasTrans,
> -                l, n, 1.0, backupAL, lda, 0.0, Stl, l);
> -
> -    for (i = 0; i < a_iters; i++)
> -    {
> -        if (a_iters < 10 || (i%(a_iters/10)) == 0)
> -        {
> -            cout << "%" << flush;
> -        }
> -
> -        AIOfile.load_ARblock(&backupAR, a_block_size);
> -        AIOfile.reset_Y();
> -        copy_vec(backupAR, AR, n*r);
> -
> -        // matlab_print_matrix("A", n, p, A);
> -        //! Generate Str
> -        cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
> -                    l, r, n, 1.0, AL, n, AR, n, 0.0, Str, l);
> -
> -        //! Generate Sbr
> -        cblas_ssyrk(CblasColMajor, CblasUpper, CblasTrans,
> -                    r, n, 1.0, backupAR, lda, 0.0, Sbr, r);
> -
> -
> -
> -        // matlab_print_matrix("S", p, p, S);
> -
> -        for (j = 0; j < y_iters; j++)
> -        {
> -            build_S(S, Stl, Str, Sbr, l, r);
> -
> -            AIOfile.load_Yblock(&Y, y_block_size);
> -            type_precision Ay[y_block_size * p];
> -
> -            // matlab_print_matrix("Y", n, y_block_size, Y);
> -
> -            //! Ay = A'*Y
> -            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
> -                        p, y_block_size, n, 1.0, A, n, Y, n, 0.0, Ay, p);
> -
> -            type_precision* B = Ay;
> -            // matlab_print_matrix("Ay", p, y_block_size, Ay);
> -
> -            //! b = S\qy
> -            info = LAPACKE_sposv(STORAGE_TYPE, 'U', p, y_block_size,
> -                                 S, p, Ay, p);
> -            assert(info == 0, "POSV");
> -
> -            // matlab_print_matrix("B", p, y_block_size, B);
> -            // exit(0);
> -            if ( ForceCheck)
> -            {
> -                check_result(backupAL, backupAR, n, p, y_block_size, r, Y, B);
> -            }
> -        }  // END for over j < y_iters
> -    }  // END for over i < a_iters
> -
> -    get_ticks(end_tick);
> -    out.duration = ticks2sec(end_tick - start_tick, cpu_freq);
> -    out.gflops = 2.0 * (n * l * l / 1000.0 + a_amount / 1000.0 *
> -                        (n * l * r + n * r * r +
> -                         y_amount * (p * n + p * p * p))) /1000.0 /1000.0;
> -
> -
> -
> -    AIOfile.finalize();
> -
> -    delete []A;
> -    delete []S;
> -}
> -
> -
> -void Algorithm::fullNEQ(struct Settings params, struct Outputs &out)
> -{
> -    int max_threads = params.threads;
> -
> -
> -    srand(time(NULL));
> -
> -
> -    blas_set_num_threads(max_threads);
> -
> -
> -    type_precision *Ytemp;
> -    lapack_int info, n, lda, ldy, l, r, k, p;
> -
> -    int i, j, w;
> -    int m;
> -
> -    cputime_type start_tick, start_tick2, end_tick;
> -
> -    AIOfile.initialize(params);
> -
> -    n = params.n; l = params.l; r = params.r; p = l+r;
> -
> -    int y_amount = params.m;
> -    int y_block_size = params.mb;// kk
> -
> -    int a_amount = params.t;
> -    int a_block_size = 1;
> -
> -    int a_iters = a_amount/a_block_size;
> -
> -    lda = n; ldy = n;
> -    k = p;
> -
> -
> -
> -    type_precision* backupAL;
> -
> -    AIOfile.load_AL(&backupAL);
> -
> -    type_precision* S = new type_precision[p * p];
> -    type_precision* Stemp = new type_precision[p * p];
> -    type_precision* A = new type_precision[n * p];
> -    type_precision* AL = A;
> -    type_precision* AR = &A[l * n];
> -
> -    int y_iters = y_amount / y_block_size;
> -
> -    type_precision* Y;
> -    type_precision* backupAR;
> -
> -    // printf("\n\n%%Computations\n%%");
> -
> -    get_ticks(start_tick);
> -
> -    for (i = 0; i < a_iters; i++)
> -    {
> -        if (a_iters < 10 || (i%(a_iters/10)) == 0)
> -        {
> -            cout << "%" << flush;
> -        }
> -
> -        copy_vec(backupAL, AL, n * l);
> -
> -        AIOfile.load_ARblock(&backupAR, a_block_size);
> -        AIOfile.reset_Y();
> -        copy_vec(backupAR, AR, n*r);
> -
> -        // matlab_print_matrix("A", n, p, A);
> -        //! Generate S
> -        cblas_ssyrk(CblasColMajor, CblasUpper, CblasTrans,
> -                    p, n, 1.0, A, lda, 0.0, S, p);
> -        // matlab_print_matrix("S", p, p, S);
> -
> -        for (j = 0; j < y_iters; j++)
> -        {
> -            AIOfile.load_Yblock(&Y, y_block_size);
> -            type_precision Ay[y_block_size * p];
> -            copy_vec(S, Stemp, p*p);
> -
> -            // matlab_print_matrix("Y", n, y_block_size, Y);
> -
> -            //! Ay = A'*Y
> -            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
> -                        p, y_block_size, n, 1.0, A, n, Y, n, 0.0, Ay, p);
> -
> -            type_precision* B = Ay;
> -            // matlab_print_matrix("Ay", p, y_block_size, Ay);
> -
> -            //! b = S\qy
> -            info = LAPACKE_sposv(STORAGE_TYPE, 'U', p, y_block_size,
> -                                 Stemp, p, Ay, p);
> -            assert(info == 0, "POSV");
> -
> -            // matlab_print_matrix("B", p, y_block_size, B);
> -            // exit(0);
> -            if ( ForceCheck)
> -            {
> -                check_result(backupAL, backupAR, n, p, y_block_size, r, Y, B);
> -            }
> -        }
> -    }
> -
> -    get_ticks(end_tick);
> -    out.duration = ticks2sec(end_tick - start_tick, cpu_freq);
> -    out.gflops = 2.0 * a_amount * (n * p * p / 1000.0 + y_amount / 1000.0 *
> -                                   (p * n + p * p * p))/1000.0/1000.0;
> -
> -
> -    AIOfile.finalize();
> -
> -
> -    free(AL);
> -    delete []A;
> -    delete []S;
> -    delete []Stemp;
> -}
> -
> -
> -//!*************************************************!//
> -void Algorithm::fullQR(struct Settings params, struct Outputs &out)
> -{
> -    int max_threads = params.threads;
> -
> -
> -    srand(time(NULL));
> -
> -    blas_set_num_threads(max_threads);
> -
> -
> -
> -    cputime_type start_tick, start_tick2, end_tick;
> -    double acc_gemm = 0;
> -    double acc_trsm = 0;
> -    double acc_other = 0;
> -    double acc_atemp = 0;
> -    double acc_rtr = 0;
> -    double acc_rqr = 0;
> -
> -
> -    type_precision *Ytemp;
> -    lapack_int info, n, lda, ldy, l, r, k, p;
> -
> -    int i, j, w;
> -    int m;
> -
> -    AIOfile.initialize(params);
> -
> -    n = params.n; l = params.l; r = params.r; p = l+r;
> -
> -    int y_amount = params.m;
> -    int y_block_size = 1;// kk
> -
> -    int a_amount = params.t;
> -    int a_block_size = 1;
> -
> -    int a_iters = a_amount/a_block_size;
> -
> -    lda = n; ldy = n;
> -    k = p;
> -
> -
> -
> -    type_precision* backupAL;
> -
> -    AIOfile.load_AL(&backupAL);
> -
> -    type_precision* Q = new type_precision[n*p];
> -    type_precision* A = Q;
> -    type_precision* AL = A;
> -    type_precision* AR = &A[l*n];
> -
> -    type_precision tau[k];
> -
> -
> -    int y_iters = y_amount / y_block_size;
> -
> -
> -    type_precision* Y;
> -    type_precision* backupAR;
> -
> -
> -
> -    // printf("\n\n%%Computations\n%%");
> -
> -    get_ticks(start_tick);
> -    for (i = 0; i < a_iters; i++)
> -    {
> -        if (a_iters < 10 || (i%(a_iters/10)) == 0)
> -        {
> -            cout << "%" << flush;
> -        }
> -
> -        copy_vec(backupAL, AL, n*l);
> -
> -        AIOfile.load_ARblock(&backupAR, a_block_size);
> -        AIOfile.reset_Y();
> -        copy_vec(backupAR, AR, n*r);
> -
> -        //! Generate R
> -        info = LAPACKE_sgeqrf(STORAGE_TYPE, n, p, A, lda, tau);
> -        assert(info == 0, "QR decomp");
> -        type_precision* R = extract_R(A, n, p);
> -
> -        //! generate Q
> -        info = LAPACKE_sorgqr(STORAGE_TYPE, n, p, k, Q, lda, tau);
> -        assert(info == 0, "Q form");
> -
> -
> -
> -//        matlab_print_matrix("ALL", n, l, backupAL);
> -//        matlab_print_matrix("ARR", n, r, backupAR);
> -
> -        for (j = 0; j < y_iters; j++)
> -        {
> -            // matlab_print_matrix("RR", p, p, R);
> -            // matlab_print_matrix("QQ", n, p, Q);
> -
> -            AIOfile.load_Yblock(&Y, y_block_size);
> -            type_precision Qy[y_block_size*p];
> -
> -            //! qy = Q'*Y
> -            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
> -                        p, y_block_size, n, 1.0, Q, n, Y, n, 0.0, Qy, p);
> -
> -            type_precision* B = Qy;
> -
> -            //! b = R\qy
> -            cblas_strsm(CblasColMajor, CblasLeft, CblasUpper,
> -                        CblasNoTrans, CblasNonUnit,
> -                        p, y_block_size, 1.0, R, p, Qy, p);
> -
> -            if (ForceCheck)
> -            {
> -                check_result(backupAL, backupAR, n, p, y_block_size, r, Y, B);
> -            }
> -        }
> -
> -        free(R);
> -    }
> -
> -    get_ticks(end_tick);
> -    out.duration = ticks2sec(end_tick - start_tick, cpu_freq);
> -    out.gflops = 2.0 * a_amount *
> -        (2.0 * n * p * p / 1000.0 + y_amount / 1000.0 *
> -         (p * n + p * p * p)) / 1000.0 / 1000.0;
> -
> -
> -    AIOfile.finalize();
> -
> -
> -    free(AL);
> -    delete[] Q;
> -}
> -
> -
> -void Algorithm::partialQR(struct Settings params, struct Outputs &out)
> -{
> -    int max_threads = params.threads;
> -
> -
> -    srand(time(NULL));
> -
> -    blas_set_num_threads(max_threads);
> -
> -
> -
> -    cputime_type start_tick, start_tick2, end_tick;
> -    double acc_gemm = 0;
> -    double acc_trsm = 0;
> -    double acc_other = 0;
> -    double acc_atemp = 0;
> -    double acc_rtr = 0;
> -    double acc_rqr = 0;
> -
> -
> -    type_precision *RL, *Ytemp, *RtlRtr;
> -    lapack_int info, n, lda, ldy, l, r, k, p;
> -
> -    int i, j, w;
> -    int m;
> -
> -    AIOfile.initialize(params);
> -
> -    n = params.n; l = params.l; r = params.r; p = l+r;
> -
> -    int y_amount = params.m;
> -    int y_block_size = params.mb;// kk
> -
> -    int a_amount = params.t;
> -    int a_block_size = 1;
> -
> -    int a_iters = a_amount/a_block_size;
> -
> -    lda = n; ldy = n;
> -    k = l;
> -
> -
> -
> -    type_precision* backupAL;
> -
> -    AIOfile.load_AL(&backupAL);
> -
> -    type_precision* Q = new type_precision[n*p];
> -    type_precision* AL = Q;
> -    copy_vec(backupAL, AL, n*l);
> -
> -
> -    type_precision tau[k];
> -
> -    type_precision* QL;
> -    type_precision* QR = &Q[n*l];
> -
> -    int y_iters = y_amount / y_block_size;
> -
> -
> -    type_precision* Y;
> -
> -    type_precision Rtr[l*r];
> -    type_precision* Ar;
> -    type_precision Rbr[r*r];
> -
> -    get_ticks(start_tick);
> -    //! Generate RTL
> -    info = LAPACKE_sgeqrf(STORAGE_TYPE, n, l, AL, lda, tau);
> -    assert(info == 0, "QR decomp");
> -
> -    type_precision* R =prepare_R(AL, n, l, r);
> -    type_precision* Rtl = R;
> -
> -    QL = AL;// same as Q
> -    AL = backupAL;
> -    //! generate QL
> -    info = LAPACKE_sorgqr(STORAGE_TYPE, n, l, k, QL, lda, tau);
> -
> -    assert(info == 0, "Q form");
> -
> -    // printf("\n\n%%Computations\n%%");
> -
> -
> -    for (i = 0; i < a_iters; i++)
> -    {
> -        if (a_iters < 10 || (i%(a_iters/10)) == 0)
> -        {
> -            cout << "%" << flush;
> -        }
> -
> -        AIOfile.load_ARblock(&Ar, a_block_size);
> -        AIOfile.reset_Y();
> -
> -        //! Rtr = Q1'*AR
> -        cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
> -                    l, r*a_block_size, n, 1.0, QL, n, Ar, n, 0.0, Rtr, l);
> -
> -        type_precision* Atemp = replicate_vec(Ar, n*r*a_block_size);
> -
> -        //! Atemp = AR-Q1*Rtr
> -        cblas_sgemm(CblasColMajor, CblasNoTrans, CblasNoTrans,
> -                    n, r*a_block_size, l, -1.0, QL, n, Rtr, l, 1.0, Atemp, n);
> -
> -        type_precision* QrRbr = QR;
> -        copy_vec(Atemp, QrRbr, n*r);
> -        info = LAPACKE_sgeqrf(STORAGE_TYPE, n, r, QrRbr, lda, tau);
> -        assert(info == 0, "QR decomp of QrRbr");
> -
> -        // reuse of old function
> -        type_precision* Rbrtemp = prepare_R(QrRbr, n, r, 0);
> -
> -        copy_vec(Rbrtemp, Rbr, r*r);
> -        info = LAPACKE_sorgqr(STORAGE_TYPE, n, r, r, QrRbr, lda, tau);
> -        assert(info == 0, "QR form");
> -        free(Rbrtemp);
> -
> -        update_R(R, Rtr, Rbr, p, p, r);
> -
> -//        matlab_print_matrix("RL", p, l, Rtl);
> -//        matlab_print_matrix("Rtr", l, r, Rtr);
> -//        matlab_print_matrix("Rbr", r, r, Rbr);
> -//        matlab_print_matrix("RR", p, p, R);
> -//        matlab_print_matrix("QQ", n, p, Q);
> -        // matlab_print_matrix("ALL", n, l, AL);
> -        // matlab_print_matrix("ARR", n, r, Ar);
> -        // matlab_print_matrix("R", p, p, R);
> -        int top_idx = 0;
> -
> -        for (j = 0; j < y_iters; j++)
> -        {
> -            AIOfile.load_Yblock(&Y, y_block_size);
> -            type_precision Qy[y_block_size*p];
> -
> -
> -            //! qy = Q'*Y
> -            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
> -                        p, y_block_size, n, 1.0, Q, n, Y, n, 0.0, Qy, p);
> -
> -            type_precision* B = Qy;
> -
> -            //! b = R\qy
> -            cblas_strsm(CblasColMajor, CblasLeft, CblasUpper,
> -                        CblasNoTrans, CblasNonUnit,
> -                        p, y_block_size, 1.0, R, p, Qy, p);
> -
> -            // matlab_print_matrix("bcomp", p, y_block_size, B);
> -
> -
> -            if ( ForceCheck)
> -            {
> -                check_result(AL, Ar, n, p, y_block_size, r, Y, B);
> -            }
> -        }
> -
> -        free(Atemp);
> -    }
> -
> -
> -    get_ticks(end_tick);
> -    out.duration = ticks2sec(end_tick - start_tick, cpu_freq);
> -    out.gflops = 2.0*(n*l*l/1000.0 + a_amount *
> -                      (2.0 * n * l * r / 1000.0 +
> -                       n *  r * r / 1000.0 +
> -                       y_amount / 1000.0 * (p * n + p * p * p))) /
> -        1000.0 / 1000.0;
> -
> -
> -    AIOfile.finalize();
> -
> -
> -    free(Rtl);
> -    free(AL);
> -    delete[] Q;
> -}
> -
> -
> -void  Algorithm::partialQR_Blocked_Rtl(struct Settings params,
> -                                       struct Outputs &out)
> -{
> -    int max_threads = params.threads;
> -
> -    // srand (time(NULL));
> -
> -    int cpu_cores = max_threads;
> -
> -
> -    blas_set_num_threads(max_threads);
> -    omp_set_num_threads(max_threads);
> -
> -
> -
> -
> -    cputime_type start_tick, start_tick2, end_tick;
> -    double acc_pre     = 0;
> -    double acc_trsm    = 0;
> -    double acc_other   = 0;
> -    double acc_atemp   = 0;
> -    double acc_rtr     = 0;
> -    double acc_rqr     = 0;
> -    double acc_RTL_QLY = 0;
> -
> -
> -    type_precision *AL, *RL, *RQy_top, *Ytemp, *RtlRtr;
> -    lapack_int info, n, lda, ldy, l, r, k, p;
> -
> -    int i, j, w;
> -    int m;
> -
> -
> -    AIOfile.initialize(params);
> -
> -    n = params.n; l = params.l; r = params.r; p = l+r;
> -
> -
> -
> -    double lvl_3_cache_size = 6;
> -
> -
> -    int y_amount = params.t;
> -    int y_block_size = params.tb;// kk
> -
> -    int a_amount = params.m;
> -    int a_block_size = params.mb;
> -
> -    int a_iters = (a_amount + a_block_size - 1) / a_block_size;
> -
> -    int y_iters = (y_amount+ y_block_size - 1)/y_block_size;
> -
> -    int qy_idx = 0;
> -
> -    out.gflops = n/1000.0*l/1000.0*l/1000.0 +
> -        gemm_flops(l, n, y_amount, 0) +
> -        y_amount*l/1000.0*l/1000.0/1000.0 +
> -        a_amount * (gemm_flops(l, n, r, 0) +
> -                    l/1000.0*l/1000.0/1000.0 +
> -                    gemm_flops(n, l, r, 1) +
> -                    n/1000.0*r/1000.0*r/1000.0 +
> -                    gemm_flops(r, n, y_amount, 0) +
> -                    y_amount/1000.0*r/1000.0*r/1000.0 +
> -                    y_amount/1000.0*l/1000.0/1000.0);
> -
> -    int sch_block_size = max(1.0,
> -                             min((double)a_block_size / (double)max_threads,
> -                                 (1.0*1000*1000) /
> -                                 (double)((sizeof(type_precision) * n * r))));
> -
> -    // cout << endl<< "taskChunk "<< sch_block_size <<endl;
> -
> -    lda = n; ldy = n;
> -    k = l;
> -
> -    AIOfile.load_AL(&AL);
> -
> -    type_precision* backupAL = replicate_vec(AL, n*l);
> -
> -
> -
> -    RQy_top = new type_precision[l*y_amount];
> -
> -    type_precision* Bfinal = (type_precision*)malloc(max_threads * p *
> -                                                     y_block_size *
> -                                                     sizeof(type_precision));
> -
> -    type_precision tau[k];
> -
> -    type_precision* QL;
> -
> -    type_precision* Rtr = (type_precision*)malloc(l * r * a_block_size *
> -                                                  sizeof(type_precision));
> -    type_precision* Qy_bot = (type_precision*)malloc(a_block_size *
> -                                                     y_block_size * r *
> -                                                     sizeof(type_precision));
> -
> -    type_precision* B_top = new type_precision[max_threads*y_block_size*l];
> -    type_precision* B_bot = new type_precision[max_threads*y_block_size*r];
> -
> -    type_precision* Ar;
> -    type_precision Rbr[r*r*a_block_size];
> -
> -
> -
> -    // printf("\n\n%%Computations\n%%");
> -    for (i = 0; i < a_iters; i++)
> -    {
> -        if (a_iters >= 10 && (i%(a_iters/10)) == 0)
> -        {
> -            // cout << "*" << flush;
> -        }
> -
> -        for (j = 0; j < y_iters; j++)
> -        {
> -            if (a_iters < 10 && (y_iters < 10 || (j%(y_iters/10)) == 0))
> -            {
> -                // cout << "*" << flush;
> -            }
> -        }
> -    }
> -    cout << endl;
> -
> -
> -    //! Start of Computations
> -
> -    get_ticks(start_tick);
> -
> -
> -    info = LAPACKE_sgeqrf(STORAGE_TYPE, n, l, AL, lda, tau);
> -    assert(info == 0, "QR decomp");
> -
> -    type_precision* Rtl = prepare_R(AL, n, l, 0);
> -
> -    QL = AL;
> -    AL = backupAL;
> -
> -    info = LAPACKE_sorgqr(STORAGE_TYPE, n, l, k, QL, lda, tau);
> -
> -    assert(info == 0, "Q form");
> -
> -    matlab_print_matrix("QL", n, l, QL);
> -
> -    qy_idx = 0;
> -
> -    // printf("\n%%Preparing IO\n");
> -    //! Prepare File for the Y
> -
> -
> -    type_precision* Y;
> -
> -    // #pragma omp parallel default(shared)
> +//    int max_threads = params.threads;
> +//
> +//    srand(time(NULL));
> +//
> +//    blas_set_num_threads(max_threads);
> +//
> +//    type_precision *Ytemp;
> +//    lapack_int info, n, lda,ldy, l, r, p;
> +//
> +//    int i,k;
> +//
> +//    cputime_type start_tick, start_tick2, end_tick;
> +//
> +//    AIOfile.initialize(params);
> +//
> +//    n = params.n; l = params.l; r = params.r; p = l+r;
> +//
> +//    int y_amount = params.m;
> +//    int y_block_size = params.mb;  // kk
> +//
> +//    int a_amount = params.t;
> +//    int a_block_size = params.tb;
> +//
> +//    int a_iters = (a_amount + a_block_size - 1) / a_block_size;
> +//
> +//    lda = n; ldy = n;
> +//    k = p;
> +//
> +//
> +//
> +//    type_precision* backupAL;
> +//
> +//    AIOfile.load_AL(&backupAL);
> +//
> +//
> +//    type_precision Stl[l * l];
> +//    type_precision Str[l * r * a_block_size];
> +//
> +//
> +//    type_precision* Ay_top = new type_precision[l * y_amount];
> +//    type_precision* Ay_bot = new type_precision[y_block_size * a_block_size * r];
> +//    type_precision* A = new type_precision[n * p];
> +//    type_precision* AR = new type_precision[n * r * a_block_size];
> +//    type_precision* AL = new type_precision[n * l];
> +//
> +//    copy_vec(backupAL, AL, n * l);
> +//    copy_vec(backupAL, A, n * l);
> +//
> +//
> +//    int y_iters = y_amount/y_block_size;
> +//
> +//    type_precision* Y;
> +//    type_precision* backupAR;
> +//
> +//    // printf("\n\n%%Computations\n%%");
> +//
> +//
> +//    get_ticks(start_tick);
> +//
> +//
> +//    //! Generate S
> +//    cblas_ssyrk(CblasColMajor, CblasUpper, CblasTrans,
> +//                l, n, 1.0, backupAL, lda, 0.0, Stl, l);
> +//
> +//    for (int j = 0; j < y_iters; j++)
>  //    {
> -//        // #pragma omp for private(Y, y_block_size, qy_idx) nowait schedule(static, 1)
> +//        AIOfile.load_Yblock(&Y, y_block_size);
> +//
> +//        //! Ay_top = AL'*Y
> +//        cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
> +//                    l, y_block_size, n, 1.0, AL, n, Y, n, 0.0,
> +//                    &Ay_top[j * l * y_block_size], l);
> +//    }
> +//
> +//
> +//    for (int i = 0; i < a_iters; i++)
> +//    {
> +//        if (a_iters > 10 && (i%(a_iters/10)) == 0)
> +//        {
> +//            cout << "%" << flush;
> +//        }
> +//
> +//
> +//        AIOfile.load_ARblock(&backupAR, a_block_size);
> +//        AIOfile.reset_Y();
> +//        copy_vec(backupAR, AR, n*r*a_block_size);
> +//
> +//        // matlab_print_matrix("A", n, p, A);
> +//        //! Generate Str
> +//        cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
> +//                    l, r * a_block_size, n, 1.0, AL, n, AR, n, 0.0, Str, l);
> +//
> +//        type_precision Sbr[r*r*a_block_size];
> +//
> +//        #pragma omp parallel for default(shared) schedule(static)
> +//        for (int ii= 0; ii < a_block_size; ii++)
> +//        {
> +//            //! Generate Sbr
> +//            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
> +//                        r, r, n, 1.0, &AR[ii*r*n], n, &AR[ii*r*n],
> +//                        n, 0.0, &Sbr[ii*r*r], r);
> +//        }
> +//
> +//        type_precision Ay[y_block_size*p];
> +//        type_precision S[p*p];
> +//
> +//
>  //        for (int j = 0; j < y_iters; j++)
>  //        {
> -//            // qy_idx = j*y_block_size*l;
> -//            // #pragma omp critical
> -//            {AIOfile.load_Yblock(&Y, y_block_size);}
> +//            if (a_iters < 10 && (y_iters < 10 || (j%(y_iters/10)) == 0))
> +//            {
> +//                cout << "%" << flush;
> +//            }
>  //
> -//            //! qy_top = QL'*Y
> +//
> +//            AIOfile.load_Yblock(&Y, y_block_size);
> +//
> +//            //! Ay_bot = AR'*Y
>  //            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
> -//                            l, y_block_size, n, 1.0, QL, n, Y, n, 0.0,&RQy_top[qy_idx], l);
> +//                        r * a_block_size, y_block_size, n, 1.0, AR, n, Y, n,
> +//                        0.0, Ay_bot, r * a_block_size);
>  //
> -//            //! K | RtlQlY =RTL\qy_top
> -//            cblas_strsm(CblasColMajor, CblasLeft, CblasUpper, CblasNoTrans, CblasNonUnit,
> -//                                            l, y_block_size, 1.0, Rtl, l,&RQy_top[qy_idx], l);
> -//            qy_idx += y_block_size*l;
> +//            #pragma omp parallel for private(S, Ay) default(shared) schedule(static)
> +//            for (int ii= 0; ii < a_block_size; ii++)
> +//            {
> [TRUNCATED]
> 
> To get the complete diff run:
>     svnlook diff /svnroot/genabel -r 1609
> _______________________________________________
> Genabel-commits mailing list
> Genabel-commits at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
> 

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140214/e7a65565/attachment-0001.sig>

From lennart at karssen.org  Sat Feb 15 23:12:32 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Sat, 15 Feb 2014 23:12:32 +0100
Subject: [GenABEL-dev] [Genabel-commits] r1612 - in
 branches/ProbABEL-0.50: checks/R-tests src
In-Reply-To: <20140215213914.9A9001803B5@r-forge.r-project.org>
References: <20140215213914.9A9001803B5@r-forge.r-project.org>
Message-ID: <52FFE650.3010102@karssen.org>

Thanks Maarten.
Good to see you're not only improving the code, but also stick to the
coding guidelines.


Enjoy the weekend,

Lennart.

On 15-02-14 22:39, noreply at r-forge.r-project.org wrote:
> Author: maartenk
> Date: 2014-02-15 22:39:14 +0100 (Sat, 15 Feb 2014)
> New Revision: 1612
> 
> Modified:
>    branches/ProbABEL-0.50/checks/R-tests/run_models_in_R_palinear.R
>    branches/ProbABEL-0.50/src/command_line_settings.cpp
>    branches/ProbABEL-0.50/src/command_line_settings.h
>    branches/ProbABEL-0.50/src/data.cpp
>    branches/ProbABEL-0.50/src/data.h
>    branches/ProbABEL-0.50/src/main.cpp
>    branches/ProbABEL-0.50/src/main_functions_dump.cpp
>    branches/ProbABEL-0.50/src/maskedmatrix.cpp
>    branches/ProbABEL-0.50/src/maskedmatrix.h
>    branches/ProbABEL-0.50/src/reg1.cpp
>    branches/ProbABEL-0.50/src/reg1.h
> Log:
> -Fixed typo in run_models_in_R_palinear.R in a variable name (test will now succeed)
> -Comment out 2 unused functions (getProgramName andNmeasured ) made todo's to remove them in the future
> -Solved some cpplint and cppcheck warnings
> 
> 
> Modified: branches/ProbABEL-0.50/checks/R-tests/run_models_in_R_palinear.R
> ===================================================================
> --- branches/ProbABEL-0.50/checks/R-tests/run_models_in_R_palinear.R	2014-02-14 14:14:21 UTC (rev 1611)
> +++ branches/ProbABEL-0.50/checks/R-tests/run_models_in_R_palinear.R	2014-02-15 21:39:14 UTC (rev 1612)
> @@ -123,7 +123,7 @@
>  }
>  colnames(prob.2df.R) <- cols2df
>  rownames(prob.2df.R) <- NULL
> -#stopifnot( all.equal(prob.2df.PA1[1:5,], prob.2df.R[1:5,], tol=tol) )
> +stopifnot( all.equal(prob.2df.PA[1:5,], prob.2df.R[1:5,], tol=tol) )
>  cat("2df\n")
>  
>  cat("\t\t\t\t\t\tOK\n")
> 
> Modified: branches/ProbABEL-0.50/src/command_line_settings.cpp
> ===================================================================
> --- branches/ProbABEL-0.50/src/command_line_settings.cpp	2014-02-14 14:14:21 UTC (rev 1611)
> +++ branches/ProbABEL-0.50/src/command_line_settings.cpp	2014-02-15 21:39:14 UTC (rev 1612)
> @@ -120,12 +120,12 @@
>  {
>      return outfilename;
>  }
> +//TODO(unknown) This function is not used. Remove in near future
> +//char* cmdvars::getProgramName() const
> +//{
> +//    return program_name;
> +//}
>  
> -char* cmdvars::getProgramName() const
> -{
> -    return program_name;
> -}
> -
>  int cmdvars::getRobust() const
>  {
>      return robust;
> 
> Modified: branches/ProbABEL-0.50/src/command_line_settings.h
> ===================================================================
> --- branches/ProbABEL-0.50/src/command_line_settings.h	2014-02-14 14:14:21 UTC (rev 1611)
> +++ branches/ProbABEL-0.50/src/command_line_settings.h	2014-02-15 21:39:14 UTC (rev 1612)
> @@ -116,7 +116,8 @@
>      int getNoutcomes() const;
>      int getNpeople() const;
>      string getOutfilename() const;
> -    char* getProgramName() const;
> +//TODO(unknown) This function is not used. Remove in near future
> +//    char* getProgramName() const;
>      int getRobust() const;
>      int getScore() const;
>      string getSep() const;
> 
> Modified: branches/ProbABEL-0.50/src/data.cpp
> ===================================================================
> --- branches/ProbABEL-0.50/src/data.cpp	2014-02-14 14:14:21 UTC (rev 1611)
> +++ branches/ProbABEL-0.50/src/data.cpp	2014-02-15 21:39:14 UTC (rev 1612)
> @@ -47,49 +47,45 @@
>  #endif
>  #include "utilities.h"
>  
> +//TODO(unknown) This function is not used. Remove in near future
> +//unsigned int Nmeasured(char * fname, int nphenocols, int npeople)
> +//{
> +//// first pass -- find unmeasured people
> +//    std::ifstream infile(fname);
> +//    if (!infile)
> +//    {
> +//        std::cerr << "Nmeasured: cannot open file " << fname << endl;
> +//    }
> +//    char tmp[100];
> +//
> +//    for (int i = 0; i < nphenocols; i++)
> +//    {
> +//        infile >> tmp;
> +//    }
> +//
> +//    unsigned short int * allmeasured = new unsigned short int[npeople];
> +//    int nids = 0;
> +//    for (int i = 0; i < npeople; i++)
> +//    {
> +//        allmeasured[i] = 1;
> +//        infile >> tmp;
> +//        for (int j = 1; j < nphenocols; j++)
> +//        {
> +//            infile >> tmp;
> +//            if (tmp[0] == 'N' || tmp[0] == 'n')
> +//                allmeasured[i] = 0;
> +//        }
> +//        if (allmeasured[i] == 1)
> +//            nids++;
> +//    }
> +//    infile.close();
> +//
> +//    delete[] allmeasured;
> +//
> +//    return (nids);
> +//}
>  
> -unsigned int Nmeasured(char * fname, int nphenocols, int npeople)
> -{
> -//TODO: unused variables remove them for good if there is no reason to keep them
> -//int ncov = nphenocols - 2;
> -//int nids_all = npeople;
>  
> -// first pass -- find unmeasured people
> -    std::ifstream infile(fname);
> -    if (!infile)
> -    {
> -        std::cerr << "Nmeasured: cannot open file " << fname << endl;
> -    }
> -    char tmp[100];
> -
> -    for (int i = 0; i < nphenocols; i++)
> -    {
> -        infile >> tmp;
> -    }
> -
> -    unsigned short int * allmeasured = new unsigned short int[npeople];
> -    int nids = 0;
> -    for (int i = 0; i < npeople; i++)
> -    {
> -        allmeasured[i] = 1;
> -        infile >> tmp;
> -        for (int j = 1; j < nphenocols; j++)
> -        {
> -            infile >> tmp;
> -            if (tmp[0] == 'N' || tmp[0] == 'n')
> -                allmeasured[i] = 0;
> -        }
> -        if (allmeasured[i] == 1)
> -            nids++;
> -    }
> -    infile.close();
> -
> -    delete[] allmeasured;
> -
> -    return (nids);
> -}
> -
> -
>  /**
>   * Read SNP information from an mlinfo file generated by the
>   * imputation software.
> @@ -269,5 +265,3 @@
>  {
>      return matrix;
>  }
> -
> -//________________________________________Maksim_end
> 
> Modified: branches/ProbABEL-0.50/src/data.h
> ===================================================================
> --- branches/ProbABEL-0.50/src/data.h	2014-02-14 14:14:21 UTC (rev 1611)
> +++ branches/ProbABEL-0.50/src/data.h	2014-02-15 21:39:14 UTC (rev 1612)
> @@ -31,8 +31,8 @@
>  #include <string>
>  
>  extern bool is_interaction_excluded;
> -
> -unsigned int Nmeasured(char * fname, int nphenocols, int npeople);
> +//TODO(unknown) This function is not used. Remove in near future
> +//unsigned int Nmeasured(char * fname, int nphenocols, int npeople);
>  #include "phedata.h"
>  #include "gendata.h"
>  
> @@ -80,4 +80,4 @@
>      ~InvSigma();
>  };
>  
> -#endif /* DATA_H_ */
> +#endif//DATA_H_
> 
> Modified: branches/ProbABEL-0.50/src/main.cpp
> ===================================================================
> --- branches/ProbABEL-0.50/src/main.cpp	2014-02-14 14:14:21 UTC (rev 1611)
> +++ branches/ProbABEL-0.50/src/main.cpp	2014-02-15 21:39:14 UTC (rev 1612)
> @@ -151,7 +151,7 @@
>  #if LOGISTIC
>      logistic_reg nrd = logistic_reg(nrgd);
>  
> -    nrd.estimate( 0, MAXITER, EPS, 0,
> +    nrd.estimate(0, MAXITER, EPS, 0,
>                   input_var.getInteraction(),
>                   input_var.getNgpreds(),
>                   invvarmatrix,
> @@ -163,7 +163,7 @@
>  #if DEBUG
>      std::cout << "[DEBUG] linear_reg nrd = linear_reg(nrgd); DONE.";
>  #endif
> -    nrd.estimate( 0, CHOLTOL, 0, input_var.getInteraction(),
> +    nrd.estimate(0, CHOLTOL, 0, input_var.getInteraction(),
>                   input_var.getNgpreds(), invvarmatrix,
>                   input_var.getRobust(), 1);
>  #elif COXPH
> @@ -199,7 +199,8 @@
>          {
>              create_header(outfile, input_var, phd, interaction_cox);
>          }
> -    } else  // Dosage data: Only additive model => only one output file
> +    }
> +    else  // Dosage data: Only additive model => only one output file
>      {
>          outfile.push_back(
>              new std::ofstream((outfilename_str + "_add.out.txt").c_str()));
> @@ -278,8 +279,7 @@
>                  write_mlinfo(outfile, file, mli, csnp, input_var,
>                               rgd.gcount, rgd.freq);
>              }
> -        } else
> -        {
> +        } else{
>              // Dosage data: only additive model
>              int file = 0;
>              write_mlinfo(outfile, file, mli, csnp, input_var,
> @@ -304,7 +304,7 @@
>                  }
>                  else
>                  {
> -                    rd.estimate( 0, MAXITER, EPS, model,
> +                    rd.estimate(0, MAXITER, EPS, model,
>                                  input_var.getInteraction(),
>                                  input_var.getNgpreds(),
>                                  invvarmatrix,
> @@ -321,7 +321,7 @@
>                  }
>                  else
>                  {
> -                    rd.estimate( 0, CHOLTOL, model,
> +                    rd.estimate(0, CHOLTOL, model,
>                                  input_var.getInteraction(),
>                                  input_var.getNgpreds(),
>                                  invvarmatrix,
> @@ -405,7 +405,7 @@
>                              regdata new_rgd = rgd;
>                              new_rgd.remove_snp_from_X();
>                              linear_reg new_null_rd(new_rgd);
> -                            new_null_rd.estimate( 0,
> +                            new_null_rd.estimate(0,
>                                                   CHOLTOL, model,
>                                                   input_var.getInteraction(),
>                                                   input_var.getNgpreds(),
> @@ -416,7 +416,7 @@
>                              regdata new_rgd = rgd;
>                              new_rgd.remove_snp_from_X();
>                              logistic_reg new_null_rd(new_rgd);
> -                            new_null_rd.estimate( 0, MAXITER, EPS,
> +                            new_null_rd.estimate(0, MAXITER, EPS,
>                                                    model,
>                                                   input_var.getInteraction(),
>                                                   input_var.getNgpreds(),
> @@ -440,8 +440,7 @@
>                              // No missing SNP data, we can compute the LRT
>                              *chi2[model] << 2. * (loglik - null_loglik);
>                          }
> -                    } else
> -                    {
> +                    } else{
>                          // We want score test output
>                          *chi2[model] << rd.chi2_score;
>                      }
> @@ -481,8 +480,7 @@
>                  if (input_var.getNgpreds() == 0)
>                  {
>                      end_pos = rgd.X.ncol;
> -                } else
> -                {
> +                } else{
>                      end_pos = rgd.X.ncol - 1;
>                  }
>  
> @@ -511,16 +509,15 @@
>                              *covvalue[model] << "nan"
>                                               << input_var.getSep()
>                                               << "nan";
> -                        } else
> -                        {
> +                        } else{
>                              *covvalue[model] << "nan";
>                          }
>                      }
>  #endif
>                      // Oct 26, 2009
>                      *chi2[model] << "nan";
> -                } else
> -                { // ngpreds==1 (and SNP is rare)
> +                } else{
> +                    // ngpreds==1 (and SNP is rare)
>                      if (input_var.getInverseFilename() == NULL)
>                      {
>                          //                     Han Chen
> 
> Modified: branches/ProbABEL-0.50/src/main_functions_dump.cpp
> ===================================================================
> --- branches/ProbABEL-0.50/src/main_functions_dump.cpp	2014-02-14 14:14:21 UTC (rev 1611)
> +++ branches/ProbABEL-0.50/src/main_functions_dump.cpp	2014-02-15 21:39:14 UTC (rev 1612)
> @@ -417,8 +417,7 @@
>          if (input_var.getNgpreds() == 2)
>          {
>              start_pos = number_of_rows_or_columns - 2;
> -        } else
> -        {
> +        } else{
>              start_pos = number_of_rows_or_columns - 1;
>          }
>      } else if (!input_var.getAllcov() && model == 0
> @@ -427,8 +426,7 @@
>          if (input_var.getNgpreds() == 2)
>          {
>              start_pos = number_of_rows_or_columns - 4;
> -        } else
> -        {
> +        } else{
>              start_pos = number_of_rows_or_columns - 2;
>          }
>      } else if (!input_var.getAllcov() && model != 0
> @@ -439,8 +437,7 @@
>              && input_var.getInteraction() != 0)
>      {
>          start_pos = number_of_rows_or_columns - 2;
> -    } else
> -    {
> +    } else{
>          start_pos = 0;
>      }
>  
> 
> Modified: branches/ProbABEL-0.50/src/maskedmatrix.cpp
> ===================================================================
> --- branches/ProbABEL-0.50/src/maskedmatrix.cpp	2014-02-14 14:14:21 UTC (rev 1611)
> +++ branches/ProbABEL-0.50/src/maskedmatrix.cpp	2014-02-15 21:39:14 UTC (rev 1612)
> @@ -26,6 +26,9 @@
>   */
>  
>  
> +
> +
> +#include <algorithm>
>  #include "maskedmatrix.h"
>  #if EIGEN
>  #include "eigen_mematrix.h"
> @@ -47,11 +50,8 @@
>  //    matrix_original = M;
>      masked_data = &matrix_original;
>      mask_of_old = new unsigned short int[M.nrow];
> -    std::fill (mask_of_old,mask_of_old+M.nrow,0);
> -    //TODO:set length of mask for all types
> +    std::fill(mask_of_old, mask_of_old+M.nrow, 0);
>      length_of_mask = M.nrow;
> -    //TODO:set type (row,column,symmetric)
> -    //type="symmetric";
>  }
>  
>  void masked_matrix::set_matrix(const mematrix<double> &M)
> @@ -59,11 +59,8 @@
>      matrix_original = M;
>      masked_data = &matrix_original;
>      mask_of_old = new unsigned short int[M.nrow];
> -    std::fill (mask_of_old,mask_of_old+M.nrow,0);
> -    //TODO:set length of mask for all types
> +    std::fill(mask_of_old, mask_of_old+M.nrow, 0);
>      length_of_mask = M.nrow;
> -    //TODO:set type (row,column,symmetric)
> -    //type="symmetric";
>  }
>  
>  masked_matrix::~masked_matrix()
> @@ -85,7 +82,7 @@
>      else
>      {
>          //Check update mask is the same as old matrix
> -        if (std::equal (newmask, newmask+length_of_mask, mask_of_old))
> +        if (std::equal(newmask, newmask+length_of_mask, mask_of_old))
>          {
>              //new mask is the same as old matrix
>              masked_data = &matrix_masked_data;
> @@ -94,7 +91,7 @@
>          {
>              // new mask differs from old matrix and create new.
>              // mask_of_old = newmask;
> -            std::copy(newmask, newmask+length_of_mask,mask_of_old);
> +            std::copy(newmask, newmask+length_of_mask, mask_of_old);
>              mask_symmetric(nmeasured);
>              masked_data = &matrix_masked_data;
>          }
> 
> Modified: branches/ProbABEL-0.50/src/maskedmatrix.h
> ===================================================================
> --- branches/ProbABEL-0.50/src/maskedmatrix.h	2014-02-14 14:14:21 UTC (rev 1611)
> +++ branches/ProbABEL-0.50/src/maskedmatrix.h	2014-02-15 21:39:14 UTC (rev 1612)
> @@ -55,4 +55,4 @@
>      void mask_symmetric(int nmeasured);
>  };
>  
> -#endif /* MASKEDMATRIX_H_ */
> +#endif//MASKEDMATRIX_H_
> 
> Modified: branches/ProbABEL-0.50/src/reg1.cpp
> ===================================================================
> --- branches/ProbABEL-0.50/src/reg1.cpp	2014-02-14 14:14:21 UTC (rev 1611)
> +++ branches/ProbABEL-0.50/src/reg1.cpp	2014-02-15 21:39:14 UTC (rev 1612)
> @@ -91,14 +91,14 @@
>                              {
>                                  col_new++;
>                                  nX_without_interact_phe[row
> -                                        * nX_without_interact_phe.ncol + col_new] =
> +                                    * nX_without_interact_phe.ncol + col_new] =
>                                          nX[row * nX.ncol + col];
>                              }
>                              if (col != interaction - 1 && iscox)
>                              {
>                                  col_new++;
>                                  nX_without_interact_phe[row
> -                                        * nX_without_interact_phe.ncol + col_new] =
> +                                    * nX_without_interact_phe.ncol + col_new] =
>                                          nX[row * nX.ncol + col];
>                              }
>                          } // interaction_only, model==0, ngpreds==2
> @@ -148,14 +148,14 @@
>                              {
>                                  col_new++;
>                                  nX_without_interact_phe[row
> -                                        * nX_without_interact_phe.ncol + col_new] =
> +                                    * nX_without_interact_phe.ncol + col_new] =
>                                          nX[row * nX.ncol + col];
>                              }
>                              if (col != interaction - 1 && iscox)
>                              {
>                                  col_new++;
>                                  nX_without_interact_phe[row
> -                                        * nX_without_interact_phe.ncol + col_new] =
> +                                     * nX_without_interact_phe.ncol + col_new] =
>                                          nX[row * nX.ncol + col];
>                              }
>                          }
> @@ -311,7 +311,7 @@
>  }
>  
>  
> -void linear_reg::estimate( int verbose, double tol_chol,
> +void linear_reg::estimate(int verbose, double tol_chol,
>          int model, int interaction, int ngpreds, masked_matrix& invvarmatrixin,
>          int robust, int nullmodel) {
>      // suda interaction parameter
> @@ -357,7 +357,7 @@
>      double sigma2_internal;
>  
>  #if EIGEN
> -    LDLT <MatrixXd> Ch ;
> +    LDLT <MatrixXd> Ch;
>  #else
>      mematrix<double> tXX_i;
>  #endif
> @@ -373,13 +373,15 @@
>          //Oct 26, 2009
>  
>  #if EIGEN
> -        MatrixXd tXW = X.data.transpose() * invvarmatrixin.masked_data->data; // flops 5997000
> +        // next line is  5997000 flops
> +        MatrixXd tXW = X.data.transpose() * invvarmatrixin.masked_data->data;
>          Ch = LDLT <MatrixXd>(tXW * X.data); // 17991 flops
> -        beta.data = Ch.solve(tXW * reg_data.Y.data);
> -        sigma2 = (reg_data.Y.data - tXW.transpose() * beta.data).squaredNorm() ; //flops: 1000+5000+3000
> +        beta.data = Ch.solve(tXW * reg_data.Y.data);//5997 flops
> +        //next line is: 1000+5000+3000= 9000 flops
> +        sigma2 = (reg_data.Y.data - tXW.transpose() * beta.data).squaredNorm();
>  #else
> -
> -        mematrix<double> tXW = transpose(X) * invvarmatrixin.masked_data; // flops 5997000
> +        // next line is  5997000 flops
> +        mematrix<double> tXW = transpose(X) * invvarmatrixin.masked_data;
>          tXX_i = tXW * X;        // 17991 flops
>          // use cholesky to invert
>          cholesky2_mm(tXX_i, tol_chol);
> @@ -387,11 +389,12 @@
>          beta = tXX_i * (tXW * reg_data.Y);        // flops 15+5997
>          // now compute residual variance
>          sigma2 = 0.;
> +        //next line is: 1000+5000+= 6000 flops
>          mematrix<double> sigma2_matrix = reg_data.Y - (transpose(tXW) * beta); //flops: 1000+5000
>          for (int i = 0; i < sigma2_matrix.nrow; i++)
>          {
>              double val = sigma2_matrix.get(i, 0);
> -            sigma2 += val * val; // flops: 3000
> +            sigma2 += val * val; // flops: 3000 (iterations counted)
>          }
>  #endif
>          double N = X.nrow;
> @@ -409,9 +412,9 @@
>          int m = X.ncol;
>          MatrixXd txx = MatrixXd(m, m).setZero().selfadjointView<Lower>().\
>                  rankUpdate(X.data.adjoint());
> -        Ch=LDLT <MatrixXd>(txx.selfadjointView<Lower>());
> -        beta.data= Ch.solve(X.data.adjoint() * reg_data.Y.data);
> -        sigma2 = (reg_data.Y.data - (X.data * beta.data)).squaredNorm() ;
> +        Ch = LDLT <MatrixXd>(txx.selfadjointView<Lower>());
> +        beta.data = Ch.solve(X.data.adjoint() * reg_data.Y.data);
> +        sigma2 = (reg_data.Y.data - (X.data * beta.data)).squaredNorm();
>  
>  #else
>          mematrix<double> tX = transpose(X);
> @@ -431,7 +434,7 @@
>          }
>  #endif
>          double N = static_cast<double>(X.nrow);
> -        double P=static_cast<double>(length_beta);
> +        double P = static_cast<double>(length_beta);
>          sigma2_internal = sigma2 / (N - P);
>          sigma2 /= N;
>      }
> @@ -456,16 +459,16 @@
>  
>  #if EIGEN
>      double intercept = beta.get(0, 0);
> -    residuals.data= reg_data.Y.data.array()-intercept;
> +    residuals.data = reg_data.Y.data.array()-intercept;
>      //matrix.
> -    ArrayXXd betacol = beta.data.block(1,0,beta.data.rows()-1,1)\
> +    ArrayXXd betacol = beta.data.block(1, 0, beta.data.rows()-1, 1)\
>              .array().transpose();
> -    ArrayXXd resid_sub = (X.data.block(0,1,X.data.rows(),X.data.cols()-1)\
> -            *betacol.matrix().asDiagonal()).rowwise().sum() ;
> +    ArrayXXd resid_sub = (X.data.block(0, 1, X.data.rows(), X.data.cols()-1)\
> +            *betacol.matrix().asDiagonal()).rowwise().sum();
>      //std::cout << resid_sub << std::endl;
> -    residuals.data-=resid_sub.matrix();
> +    residuals.data -= resid_sub.matrix();
>      //residuals[i] -= resid_sub;
> -    loglik-=(residuals.data.array().square()*halfrecsig2).sum();
> +    loglik -= (residuals.data.array().square()*halfrecsig2).sum();
>  
>      //loglik -= halfrecsig2 * residuals[i] * residuals[i];
>  
> @@ -483,7 +486,7 @@
>  
>      loglik -= static_cast<double>(reg_data.nids) * log(sqrt(sigma2));
>  #if EIGEN
> -    MatrixXd tXX_inv=Ch.solve(MatrixXd(length_beta, length_beta).
> +    MatrixXd tXX_inv = Ch.solve(MatrixXd(length_beta, length_beta).
>              Identity(length_beta,length_beta));
>  #endif
>  
> @@ -512,7 +515,6 @@
>  
>  #endif
>  
> -
>      }
>      //cout << "estimate 0\n";
>  #if EIGEN
> @@ -526,18 +528,17 @@
>                  (sigma2_internal
>                          * tXX_inv.diagonal().array()).sqrt();
>      }
> -    int offset=X.ncol- 1;
> +    int offset = X.ncol- 1;
>      //if additive and interaction and 2 predictors and more then 2 betas
>  
>      if (model == 0 && interaction != 0 && ngpreds == 2 && length_beta > 2){
> -          offset=X.ncol - 2;
> +          offset = X.ncol - 2;
>      }
>  
>      if (robust)
>      {
>          covariance.data = robust_sigma2.data.bottomLeftCorner(
>                  offset, offset).diagonal();
> -
>      }
>      else
>      {
> @@ -603,13 +604,12 @@
>          }
>      }
>  #endif
> -
>  }
>  
>  void linear_reg::score(mematrix<double>& resid,
>          double tol_chol, int model, int interaction, int ngpreds,
>          const masked_matrix& invvarmatrix, int nullmodel) {
> -   // regdata rdata = rdatain.get_unmasked_data();
> +    //regdata rdata = rdatain.get_unmasked_data();
>      base_score(resid,  tol_chol, model, interaction, ngpreds,
>              invvarmatrix, nullmodel = 0);
>  }
> @@ -632,7 +632,7 @@
>      chi2_score = -1.;
>  }
>  
> -void logistic_reg::estimate( int verbose, int maxiter,
> +void logistic_reg::estimate(int verbose, int maxiter,
>          double eps, int model, int interaction, int ngpreds,
>          masked_matrix& invvarmatrixin, int robust, int nullmodel) {
>      // In contrast to the 'linear' case 'invvarmatrix' contains the
> @@ -656,7 +656,6 @@
>  
>      if (length_beta > 1)
>      {
> -
>          if (model == 0 && interaction != 0 && ngpreds == 2 && length_beta > 2)
>          {
>              covariance.reinit(length_beta - 2, 1);
> @@ -687,7 +686,7 @@
>  
>      if (invvarmatrix.nrow != 0 && invvarmatrix.ncol != 0)
>      {
> -        //TODO(maarten):invvarmatix is symmetric:is there an more effective way?
> +        //TODO(maarten) invvarmatix is symmetric:is there an more effective way?
>          tX = tX * invvarmatrix;
>      }
>      /*
> @@ -706,7 +705,6 @@
>       */
>      niter = 0;
>      double delta = 1.;
> -    double prevlik = 0.;
>      while (niter < maxiter && delta > eps)
>      {
>          mematrix<double> eMu = (X) * beta;
> @@ -776,7 +774,7 @@
>          }
>          // std::cout << "beta:\n"; beta.print();
>          // compute likelihood
> -        prevlik = loglik;
> +        double prevlik = loglik;
>          loglik = 0.;
>          for (int i = 0; i < eMu.nrow; i++)
>              loglik += reg_data.Y[i] * eMu_us[i] - log(1. + exp(eMu_us[i]));
> 
> Modified: branches/ProbABEL-0.50/src/reg1.h
> ===================================================================
> --- branches/ProbABEL-0.50/src/reg1.h	2014-02-14 14:14:21 UTC (rev 1611)
> +++ branches/ProbABEL-0.50/src/reg1.h	2014-02-15 21:39:14 UTC (rev 1612)
> @@ -85,13 +85,13 @@
>      linear_reg(regdata& rdatain);
>      ~linear_reg()
>      {
> -        delete [] reg_data.masked_data ;
> +        delete [] reg_data.masked_data;
>          //		delete beta;
>          //		delete sebeta;
>          //		delete residuals;
>      }
>  
> -    void estimate( int verbose, double tol_chol, int model,
> +    void estimate(int verbose, double tol_chol, int model,
>                    int interaction, int ngpreds,
>                    masked_matrix& invvarmatrixin,
>                    int robust, int nullmodel = 0);
> @@ -108,12 +108,12 @@
>      logistic_reg(regdata& rdatain);
>      ~logistic_reg()
>      {
> -        delete [] reg_data.masked_data ;
> +        delete [] reg_data.masked_data;
>          //		delete beta;
>          //		delete sebeta;
>      }
>  
> -    void estimate( int verbose, int maxiter, double eps,
> +    void estimate(int verbose, int maxiter, double eps,
>                    int model, int interaction, int ngpreds,
>                    masked_matrix& invvarmatrixin, int robust,
>                    int nullmodel = 0);
> @@ -123,4 +123,4 @@
>                 masked_matrix& invvarmatrix, int nullmodel = 0);
>  };
>  
> -#endif
> +#endif//REG1_H_
> 
> _______________________________________________
> Genabel-commits mailing list
> Genabel-commits at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
> 

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140215/a261e394/attachment-0001.sig>

From lennart at karssen.org  Mon Feb 17 12:03:54 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Mon, 17 Feb 2014 12:03:54 +0100
Subject: [GenABEL-dev] [Genabel-commits] r1609 - pkg/OmicABELnoMM/src
In-Reply-To: <52FDF0B4.10008@karssen.org>
References: <20140213204258.0DD561849B8@r-forge.r-project.org>
 <52FDF0B4.10008@karssen.org>
Message-ID: <5301EC9A.9020902@karssen.org>

Dear list,

As you may have noticed, I have given Alvaro Frank commit access to our
SVN repository. As announced in December [1], Alvaro is the main force
behind OmicABELnoMM. Alvaro will initially concentrate on removing the
compiler warnings and creating a test suite for OmicABELnoMM.
Welcome Alvaro, and looking forward to seeing more of your work/code!


Best regards,

Lennart.


[1]
http://lists.r-forge.r-project.org/pipermail/genabel-devel/2013-December/000891.html


On 14-02-14 11:32, L.C. Karssen wrote:
> Hi Alvaro,
>
> Thanks for your contribution and congratulations on your first commit in
> the GenABEL project.
>
> Glad to see that the number of has been reduced a lot! I still get a few
> errors after running
>  ./configure  LDFLAGS="-L/usr/lib/openblas-base"
>  make -j4
> but they seem to be minor and easy to fix. Have a look at
> http://jenkins.genabel.org/jenkins/job/OmicABELnoMM/15/warnings23Result/
> for the warning messages and corresponding lines in the code.
> Also try to have a look at the CPPcheck results in Jenkins. It warns
> about three (possible) memory leaks.
>
>
> Thanks a lot and looking forward to your next commits.
>
>
> Lennart.
>
>
> On 13-02-14 21:42, noreply at r-forge.r-project.org wrote:
>> Author: afrank
>> Date: 2014-02-13 21:42:52 +0100 (Thu, 13 Feb 2014)
>> New Revision: 1609
>>
>> Modified:
>>    pkg/OmicABELnoMM/src/AIOwrapper.cpp
>>    pkg/OmicABELnoMM/src/Algorithm.cpp
>>    pkg/OmicABELnoMM/src/Algorithm.h
>>    pkg/OmicABELnoMM/src/Definitions.h
>>    pkg/OmicABELnoMM/src/Utility.cpp
>>    pkg/OmicABELnoMM/src/Utility.h
>>    pkg/OmicABELnoMM/src/main.cpp
>> Log:
>> Warnings from c/c++ inconsistencies and -Wall fixed. Legacy code from unused functionality removed. A few bugs with indices fixed. Stable code.
>>
>> Modified: pkg/OmicABELnoMM/src/AIOwrapper.cpp
>> ===================================================================
>> --- pkg/OmicABELnoMM/src/AIOwrapper.cpp	2014-02-13 09:24:32 UTC (rev 1608)
>> +++ pkg/OmicABELnoMM/src/AIOwrapper.cpp	2014-02-13 20:42:52 UTC (rev 1609)
>> @@ -29,9 +29,9 @@
>>
>>      Fhandler->fakefiles = params.use_fake_files;
>>
>> -    databel_fvi* Yfvi;
>> -    databel_fvi* ALfvi;
>> -    databel_fvi* ARfvi;
>> +//    databel_fvi* Yfvi;
>> +//    databel_fvi* ALfvi;
>> +//    databel_fvi* ARfvi;
>>
>>      if (!Fhandler->fakefiles)
>>      {
>> @@ -49,7 +49,7 @@
>>          params.l = ALfvi->fvi_header.numVariables;
>>
>>          //block size to keep mem under 1 gigabyte
>> -        int opt_block = params.n / (4*1000^3) * (1/(2*params.r));
>> +        //int opt_block = params.n / (4*1000^3) * (1/(2*params.r));
>>          int opt_tb = 1000;
>>          int opt_mb = 1000;
>>
>> @@ -81,7 +81,7 @@
>>  void AIOwrapper::finalize()
>>  {
>>      //cout << "f";
>> -    void *status;
>> +    //void *status;
>>
>>      Fhandler->not_done = false;
>>      pthread_mutex_lock(&(Fhandler->m_more));
>> @@ -368,13 +368,15 @@
>>      }
>>  //
>>  //            //!induce realistic fileread delay
>> +
>> +    return 0;
>>  }
>>
>>
>>  void AIOwrapper::load_ARblock(type_precision** Ar, int &Ar_blockSize)
>>  {
>> -    int status;
>> -    int createstatus = 0;
>> +    //int status;
>> +    //int createstatus = 0;
>>      //cout<<"^";
>>
>>      while (Fhandler->ar_full_buffers.empty())
>> @@ -428,8 +430,8 @@
>>
>>  void AIOwrapper::load_Yblock(type_precision** Y, int &y_blockSize)
>>  {
>> -    int status;
>> -    int createstatus = 0;
>> +    //int status;
>> +    //int createstatus = 0;
>>
>>      while (Fhandler->full_buffers.empty())
>>      {
>> @@ -528,7 +530,7 @@
>>
>>  void AIOwrapper::reset_Y()
>>  {
>> -    void *status;
>> +    //void *status;
>>
>>      Fhandler->seed = 1337;
>>
>> @@ -569,7 +571,7 @@
>>
>>  void AIOwrapper::reset_AR()
>>  {
>> -    void *status;
>> +    //void *status;
>>
>>
>>      //cout << "ra" << flush;
>> @@ -697,7 +699,7 @@
>>  FILE * AIOwrapper::fgls_fopen( const char *path, const char *mode )
>>  {
>>          FILE * f;
>> -        char err[100];
>> +        //char err[100];
>>
>>          f = fopen(path, mode );
>>          if ( f == NULL )
>>
>> Modified: pkg/OmicABELnoMM/src/Algorithm.cpp
>> ===================================================================
>> --- pkg/OmicABELnoMM/src/Algorithm.cpp	2014-02-13 09:24:32 UTC (rev 1608)
>> +++ pkg/OmicABELnoMM/src/Algorithm.cpp	2014-02-13 20:42:52 UTC (rev 1609)
>> @@ -16,30 +16,12 @@
>>  {
>>      switch (type)
>>      {
>> -    case FULL_NEQ:
>> -        fullNEQ(params, out);
>> -        break;
>> -    case P_NEQ:
>> -        partialNEQ(params, out);
>> -        break;
>> -    case P_NEQ_B_OPT:
>> -        partialNEQ_Blocked_STL(params, out);
>> -        break;
>> -    case FULL_QR:
>> -        fullQR(params, out);
>> -        break;
>> -    case P_QR:
>> -        partialQR(params, out);
>> -        break;
>> -    case P_QR_B_OPT:
>> -        partialQR_Blocked_Rtl(params, out);
>> -        break;
>> -    case P_NEQ_B_OPT_MD:
>> -        partialNEQ_Blocked_STL_MD(params, out);
>> -        break;
>> +        case P_NEQ_B_OPT_MD:
>> +            partialNEQ_Blocked_STL_MD(params, out);
>> +            break;
>>
>> -    default:
>> -        break;
>> +        default:
>> +            break;
>>      }
>>  }
>>
>> @@ -74,7 +56,7 @@
>>                                 int dim2_b, int dim1_b_bot)
>>  {
>>      // memcpy are faster version of the fors
>> -    int i, k, w, top_idx, bot_idx, max = dim1_b*dim2_b;
>> +    int i, k, w, top_idx, bot_idx;
>>      int size;
>>      top_idx = 0;
>>      bot_idx = 0;
>> @@ -111,7 +93,7 @@
>>                             type_precision* bot, int dim1_QY,
>>                             int dim2_QY, int dim1_qy_bot, int bot_blocks )
>>  {
>> -    int i, k, w, top_idx, bot_idx, max = dim1_QY*dim2_QY;
>> +    int i, k, w, top_idx, bot_idx;
>>      top_idx = 0;
>>      bot_idx = 0;
>>      for (k = 0; k < dim2_QY; k++)
>> @@ -326,12 +308,9 @@
>>      blas_set_num_threads(max_threads);
>>
>>
>> -    type_precision *Ytemp;
>> -    lapack_int info, n, lda, ldy, l, r, k, p;
>> +    //type_precision *Ytemp;
>> +    lapack_int info, n, lda, l, r, p;
>>
>> -    int i, j, w;
>> -    int m;
>> -
>>      cputime_type start_tick, start_tick2, end_tick;
>>
>>      AIOfile.initialize(params);
>> @@ -349,12 +328,12 @@
>>      int y_iters = (y_amount + y_block_size - 1) / y_block_size;
>>
>>
>> -    lda = n; ldy = n;
>> -    k = p;
>> +    lda = n;
>>
>> +
>>      for (int j = 0; j < y_iters; j++)
>>      {
>> -        if (y_iters >= 40 && (i%(y_iters/40)) == 0)
>> +        if (y_iters >= 40 && (j%(y_iters/40)) == 0)
>>          {
>>              cout << "*" << flush;
>>          }
>> @@ -389,7 +368,9 @@
>>
>>
>>      AIOfile.load_AL(&backupAL);
>> -    int total_al_nans = replace_nans(0, backupAL, n, l);
>> +    //int total_al_nans = replace_nans(0, backupAL, n, l);
>> +    replace_nans(0, backupAL, n, l);
>> +
>>      copy_vec(backupAL, AL, n*l);
>>
>>
>> @@ -402,7 +383,7 @@
>>
>>      for (int j = 0; j < y_iters; j++)
>>      {
>> -        if (y_iters >= 40 && (i%(y_iters/40)) == 0)
>> +        if (y_iters >= 40 && (j%(y_iters/40)) == 0)
>>          {
>>              cout << AIOfile.io_overhead << flush;
>>              AIOfile.io_overhead = "*";
>> @@ -412,7 +393,8 @@
>>
>>          list<long int> y_nan_idxs[y_block_size];
>>
>> -        int total_y_nans = replace_nans(&y_nan_idxs[0], Y, n, y_block_size);
>> +        //int total_y_nans = replace_nans(&y_nan_idxs[0], Y, n, y_block_size);
>> +        replace_nans(&y_nan_idxs[0], Y, n, y_block_size);
>>
>>
>>          //! Ay_top = AL'*Y
>> @@ -424,7 +406,8 @@
>>          for (int i = 0; i < a_iters; i++)
>>          {
>>              AIOfile.load_ARblock(&backupAR, a_block_size);
>> -            int total_ar_nans = replace_nans(0, backupAR, n, a_block_size * r);
>> +            //int total_ar_nans = replace_nans(0, backupAR, n, a_block_size * r);
>> +            replace_nans(0, backupAR, n, a_block_size * r);
>>              copy_vec(backupAR, AR, n *  r * a_block_size);
>>
>>              get_ticks(start_tick2);
>> @@ -501,7 +484,7 @@
>>                                   &Ay[l+ii*p], r);
>>
>>
>> -                        type_precision* B = Ay;
>> +                        //type_precision* B = Ay;
>>                          type_precision S[p * p];
>>
>>                          //! Rebuild S
>> @@ -564,1194 +547,187 @@
>>
>>  //!*************************************************!//
>>
>> -
>> +//for reference
>>  void Algorithm::partialNEQ_Blocked_STL(struct Settings params,
>>                                         struct Outputs &out)
>>  {
>> -    int max_threads = params.threads;
>> -
>> -    srand(time(NULL));
>> -
>> -    blas_set_num_threads(max_threads);
>> -
>> -    type_precision *Ytemp;
>> -    lapack_int info, n, lda, ldy, l, r, k, p;
>> -
>> -    int i, j, w;
>> -    int m;
>> -
>> -    cputime_type start_tick, start_tick2, end_tick;
>> -
>> -    AIOfile.initialize(params);
>> -
>> -    n = params.n; l = params.l; r = params.r; p = l+r;
>> -
>> -    int y_amount = params.m;
>> -    int y_block_size = params.mb;  // kk
>> -
>> -    int a_amount = params.t;
>> -    int a_block_size = params.tb;
>> -
>> -    int a_iters = (a_amount + a_block_size - 1) / a_block_size;
>> -
>> -    lda = n; ldy = n;
>> -    k = p;
>> -
>> -
>> -
>> -    type_precision* backupAL;
>> -
>> -    AIOfile.load_AL(&backupAL);
>> -
>> -
>> -    type_precision Stl[l * l];
>> -    type_precision Str[l * r * a_block_size];
>> -
>> -
>> -    type_precision* Ay_top = new type_precision[l * y_amount];
>> -    type_precision* Ay_bot = new type_precision[y_block_size * a_block_size * r];
>> -    type_precision* A = new type_precision[n * p];
>> -    type_precision* AR = new type_precision[n * r * a_block_size];
>> -    type_precision* AL = new type_precision[n * l];
>> -
>> -    copy_vec(backupAL, AL, n * l);
>> -    copy_vec(backupAL, A, n * l);
>> -
>> -
>> -    int y_iters = y_amount/y_block_size;
>> -
>> -    type_precision* Y;
>> -    type_precision* backupAR;
>> -
>> -    // printf("\n\n%%Computations\n%%");
>> -
>> -
>> -    get_ticks(start_tick);
>> -
>> -
>> -    //! Generate S
>> -    cblas_ssyrk(CblasColMajor, CblasUpper, CblasTrans,
>> -                l, n, 1.0, backupAL, lda, 0.0, Stl, l);
>> -
>> -    for (j = 0; j < y_iters; j++)
>> -    {
>> -        AIOfile.load_Yblock(&Y, y_block_size);
>> -
>> -        //! Ay_top = AL'*Y
>> -        cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
>> -                    l, y_block_size, n, 1.0, AL, n, Y, n, 0.0,
>> -                    &Ay_top[j * l * y_block_size], l);
>> -    }
>> -
>> -
>> -    for (i = 0; i < a_iters; i++)
>> -    {
>> -        if (a_iters > 10 && (i%(a_iters/10)) == 0)
>> -        {
>> -            cout << "%" << flush;
>> -        }
>> -
>> -
>> -        AIOfile.load_ARblock(&backupAR, a_block_size);
>> -        AIOfile.reset_Y();
>> -        copy_vec(backupAR, AR, n*r*a_block_size);
>> -
>> -        // matlab_print_matrix("A", n, p, A);
>> -        //! Generate Str
>> -        cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
>> -                    l, r * a_block_size, n, 1.0, AL, n, AR, n, 0.0, Str, l);
>> -
>> -        type_precision Sbr[r*r*a_block_size];
>> -
>> -        #pragma omp parallel for default(shared) schedule(static)
>> -        for (int ii= 0; ii < a_block_size; ii++)
>> -        {
>> -            //! Generate Sbr
>> -            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
>> -                        r, r, n, 1.0, &AR[ii*r*n], n, &AR[ii*r*n],
>> -                        n, 0.0, &Sbr[ii*r*r], r);
>> -        }
>> -
>> -        type_precision Ay[y_block_size*p];
>> -        type_precision S[p*p];
>> -
>> -
>> -        for (j = 0; j < y_iters; j++)
>> -        {
>> -            if (a_iters < 10 && (y_iters < 10 || (j%(y_iters/10)) == 0))
>> -            {
>> -                cout << "%" << flush;
>> -            }
>> -
>> -
>> -            AIOfile.load_Yblock(&Y, y_block_size);
>> -
>> -            //! Ay_bot = AR'*Y
>> -            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
>> -                        r * a_block_size, y_block_size, n, 1.0, AR, n, Y, n,
>> -                        0.0, Ay_bot, r * a_block_size);
>> -
>> -            #pragma omp parallel for private(S, Ay) default(shared) schedule(static)
>> -            for (int ii= 0; ii < a_block_size; ii++)
>> -            {
>> -                // matlab_print_matrix("Y", n, y_block_size, Y);
>> -
>> -                //! Rebuild AY
>> -
>> -                for (int jj = 0; jj < y_block_size; jj++)
>> -                {
>> -                    copy_vec(&Ay_top[j*l*y_block_size+jj*l], &Ay[jj*p], l);
>> -                    copy_vec(&Ay_bot[ii*r + jj*r*a_block_size], &Ay[l+jj*p], r);
>> -                }
>> -
>> -                type_precision* B = Ay;
>> -                // matlab_print_matrix("Ay_top", l, y_block_size,&Ay_top[j*l*y_block_size]);
>> -                // matlab_print_matrix("Ay_bot", r, y_block_size, Ay_bot);
>> -                // matlab_print_matrix("Ay", p, y_block_size, Ay);
>> -
>> -                //! Rebuild S
>> -                build_S(S, Stl, &Str[ii*r*l], &Sbr[ii*r*r], l, r);
>> -                // matlab_print_matrix("S", p, p, S);
>> -
>> -                //! b = S\Ay
>> -                info = LAPACKE_sposv(STORAGE_TYPE, 'U', p, y_block_size,
>> -                                     S, p, Ay, p);
>> -                // assert(info == 0,"POSV");
>> -
>> -
>> -                if (ForceCheck)
>> -                {
>> -                    check_result(backupAL, &AR[ii * r * n], n, p,
>> -                                 y_block_size, r, Y, B);
>> -                }
>> -            }
>> -        }
>> -    }
>> -
>> -    get_ticks(end_tick);
>> -    out.duration = ticks2sec(end_tick - start_tick, cpu_freq);
>> -    // out.gflops = a_amount/1000.0*(n*l*r+n*r*r+y_amount*(n*r+p*p*p)))/1000.0/1000.0;
>> -    out.gflops = gemm_flops(l, n, l, 0) + gemm_flops(l, n, y_amount, 0) +
>> -                 a_amount * (gemm_flops(l, n, r, 0) +
>> -                             gemm_flops(r, n, r, 0) +
>> -                             gemm_flops(r, n, y_amount, 0) +
>> -                             (y_amount/1000.0) *
>> -                             (p * p * p / 3.0) / 1000.0/1000.0);
>> -
>> -
>> -    AIOfile.finalize();
>> -
>> -    delete []Ay_top;
>> -    delete []Ay_bot;
>> -    delete []A;
>> -    delete []AL;
>> -    delete []AR;
>> -}
>> -
>> -
>> -void Algorithm::partialNEQ(struct Settings params, struct Outputs &out)
>> -{
>> -    int max_threads = params.threads;
>> -
>> -
>> -    srand(time(NULL));
>> -
>> -
>> -    blas_set_num_threads(max_threads);
>> -
>> -
>> -    type_precision *Ytemp;
>> -    lapack_int info, n, lda, ldy, l, r, k, p;
>> -
>> -    int i, j, w;
>> -    int m;
>> -
>> -    cputime_type start_tick, start_tick2, end_tick;
>> -    AIOfile.initialize(params);
>> -
>> -    n = params.n; l = params.l; r = params.r; p = l+r;
>> -
>> -    int y_amount = params.m;
>> -    int y_block_size = params.mb;// kk
>> -
>> -    int a_amount = params.t;
>> -    int a_block_size = 1;
>> -
>> -    int a_iters = a_amount/a_block_size;
>> -
>> -    lda = n; ldy = n;
>> -    k = p;
>> -
>> -
>> -
>> -    type_precision* backupAL;
>> -
>> -    AIOfile.load_AL(&backupAL);
>> -
>> -    type_precision* S = new type_precision[p * p];
>> -    type_precision Stl[l * l];
>> -    type_precision Str[l * r];
>> -    type_precision Sbr[r * r];
>> -    type_precision* A = new type_precision[n * p];
>> -    type_precision* AL = A;
>> -    type_precision* AR = &A[l * n];
>> -
>> -    copy_vec(backupAL, AL, n*l);
>> -
>> -
>> -    int y_iters = y_amount / y_block_size;
>> -
>> -    type_precision* Y;
>> -    type_precision* backupAR;
>> -
>> -    // printf("\n\n%%Computations\n%%");
>> -
>> -    get_ticks(start_tick);
>> -
>> -    //! Generate S
>> -    cblas_ssyrk(CblasColMajor, CblasUpper, CblasTrans,
>> -                l, n, 1.0, backupAL, lda, 0.0, Stl, l);
>> -
>> -    for (i = 0; i < a_iters; i++)
>> -    {
>> -        if (a_iters < 10 || (i%(a_iters/10)) == 0)
>> -        {
>> -            cout << "%" << flush;
>> -        }
>> -
>> -        AIOfile.load_ARblock(&backupAR, a_block_size);
>> -        AIOfile.reset_Y();
>> -        copy_vec(backupAR, AR, n*r);
>> -
>> -        // matlab_print_matrix("A", n, p, A);
>> -        //! Generate Str
>> -        cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
>> -                    l, r, n, 1.0, AL, n, AR, n, 0.0, Str, l);
>> -
>> -        //! Generate Sbr
>> -        cblas_ssyrk(CblasColMajor, CblasUpper, CblasTrans,
>> -                    r, n, 1.0, backupAR, lda, 0.0, Sbr, r);
>> -
>> -
>> -
>> -        // matlab_print_matrix("S", p, p, S);
>> -
>> -        for (j = 0; j < y_iters; j++)
>> -        {
>> -            build_S(S, Stl, Str, Sbr, l, r);
>> -
>> -            AIOfile.load_Yblock(&Y, y_block_size);
>> -            type_precision Ay[y_block_size * p];
>> -
>> -            // matlab_print_matrix("Y", n, y_block_size, Y);
>> -
>> -            //! Ay = A'*Y
>> -            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
>> -                        p, y_block_size, n, 1.0, A, n, Y, n, 0.0, Ay, p);
>> -
>> -            type_precision* B = Ay;
>> -            // matlab_print_matrix("Ay", p, y_block_size, Ay);
>> -
>> -            //! b = S\qy
>> -            info = LAPACKE_sposv(STORAGE_TYPE, 'U', p, y_block_size,
>> -                                 S, p, Ay, p);
>> -            assert(info == 0, "POSV");
>> -
>> -            // matlab_print_matrix("B", p, y_block_size, B);
>> -            // exit(0);
>> -            if ( ForceCheck)
>> -            {
>> -                check_result(backupAL, backupAR, n, p, y_block_size, r, Y, B);
>> -            }
>> -        }  // END for over j < y_iters
>> -    }  // END for over i < a_iters
>> -
>> -    get_ticks(end_tick);
>> -    out.duration = ticks2sec(end_tick - start_tick, cpu_freq);
>> -    out.gflops = 2.0 * (n * l * l / 1000.0 + a_amount / 1000.0 *
>> -                        (n * l * r + n * r * r +
>> -                         y_amount * (p * n + p * p * p))) /1000.0 /1000.0;
>> -
>> -
>> -
>> -    AIOfile.finalize();
>> -
>> -    delete []A;
>> -    delete []S;
>> -}
>> -
>> -
>> -void Algorithm::fullNEQ(struct Settings params, struct Outputs &out)
>> -{
>> -    int max_threads = params.threads;
>> -
>> -
>> -    srand(time(NULL));
>> -
>> -
>> -    blas_set_num_threads(max_threads);
>> -
>> -
>> -    type_precision *Ytemp;
>> -    lapack_int info, n, lda, ldy, l, r, k, p;
>> -
>> -    int i, j, w;
>> -    int m;
>> -
>> -    cputime_type start_tick, start_tick2, end_tick;
>> -
>> -    AIOfile.initialize(params);
>> -
>> -    n = params.n; l = params.l; r = params.r; p = l+r;
>> -
>> -    int y_amount = params.m;
>> -    int y_block_size = params.mb;// kk
>> -
>> -    int a_amount = params.t;
>> -    int a_block_size = 1;
>> -
>> -    int a_iters = a_amount/a_block_size;
>> -
>> -    lda = n; ldy = n;
>> -    k = p;
>> -
>> -
>> -
>> -    type_precision* backupAL;
>> -
>> -    AIOfile.load_AL(&backupAL);
>> -
>> -    type_precision* S = new type_precision[p * p];
>> -    type_precision* Stemp = new type_precision[p * p];
>> -    type_precision* A = new type_precision[n * p];
>> -    type_precision* AL = A;
>> -    type_precision* AR = &A[l * n];
>> -
>> -    int y_iters = y_amount / y_block_size;
>> -
>> -    type_precision* Y;
>> -    type_precision* backupAR;
>> -
>> -    // printf("\n\n%%Computations\n%%");
>> -
>> -    get_ticks(start_tick);
>> -
>> -    for (i = 0; i < a_iters; i++)
>> -    {
>> -        if (a_iters < 10 || (i%(a_iters/10)) == 0)
>> -        {
>> -            cout << "%" << flush;
>> -        }
>> -
>> -        copy_vec(backupAL, AL, n * l);
>> -
>> -        AIOfile.load_ARblock(&backupAR, a_block_size);
>> -        AIOfile.reset_Y();
>> -        copy_vec(backupAR, AR, n*r);
>> -
>> -        // matlab_print_matrix("A", n, p, A);
>> -        //! Generate S
>> -        cblas_ssyrk(CblasColMajor, CblasUpper, CblasTrans,
>> -                    p, n, 1.0, A, lda, 0.0, S, p);
>> -        // matlab_print_matrix("S", p, p, S);
>> -
>> -        for (j = 0; j < y_iters; j++)
>> -        {
>> -            AIOfile.load_Yblock(&Y, y_block_size);
>> -            type_precision Ay[y_block_size * p];
>> -            copy_vec(S, Stemp, p*p);
>> -
>> -            // matlab_print_matrix("Y", n, y_block_size, Y);
>> -
>> -            //! Ay = A'*Y
>> -            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
>> -                        p, y_block_size, n, 1.0, A, n, Y, n, 0.0, Ay, p);
>> -
>> -            type_precision* B = Ay;
>> -            // matlab_print_matrix("Ay", p, y_block_size, Ay);
>> -
>> -            //! b = S\qy
>> -            info = LAPACKE_sposv(STORAGE_TYPE, 'U', p, y_block_size,
>> -                                 Stemp, p, Ay, p);
>> -            assert(info == 0, "POSV");
>> -
>> -            // matlab_print_matrix("B", p, y_block_size, B);
>> -            // exit(0);
>> -            if ( ForceCheck)
>> -            {
>> -                check_result(backupAL, backupAR, n, p, y_block_size, r, Y, B);
>> -            }
>> -        }
>> -    }
>> -
>> -    get_ticks(end_tick);
>> -    out.duration = ticks2sec(end_tick - start_tick, cpu_freq);
>> -    out.gflops = 2.0 * a_amount * (n * p * p / 1000.0 + y_amount / 1000.0 *
>> -                                   (p * n + p * p * p))/1000.0/1000.0;
>> -
>> -
>> -    AIOfile.finalize();
>> -
>> -
>> -    free(AL);
>> -    delete []A;
>> -    delete []S;
>> -    delete []Stemp;
>> -}
>> -
>> -
>> -//!*************************************************!//
>> -void Algorithm::fullQR(struct Settings params, struct Outputs &out)
>> -{
>> -    int max_threads = params.threads;
>> -
>> -
>> -    srand(time(NULL));
>> -
>> -    blas_set_num_threads(max_threads);
>> -
>> -
>> -
>> -    cputime_type start_tick, start_tick2, end_tick;
>> -    double acc_gemm = 0;
>> -    double acc_trsm = 0;
>> -    double acc_other = 0;
>> -    double acc_atemp = 0;
>> -    double acc_rtr = 0;
>> -    double acc_rqr = 0;
>> -
>> -
>> -    type_precision *Ytemp;
>> -    lapack_int info, n, lda, ldy, l, r, k, p;
>> -
>> -    int i, j, w;
>> -    int m;
>> -
>> -    AIOfile.initialize(params);
>> -
>> -    n = params.n; l = params.l; r = params.r; p = l+r;
>> -
>> -    int y_amount = params.m;
>> -    int y_block_size = 1;// kk
>> -
>> -    int a_amount = params.t;
>> -    int a_block_size = 1;
>> -
>> -    int a_iters = a_amount/a_block_size;
>> -
>> -    lda = n; ldy = n;
>> -    k = p;
>> -
>> -
>> -
>> -    type_precision* backupAL;
>> -
>> -    AIOfile.load_AL(&backupAL);
>> -
>> -    type_precision* Q = new type_precision[n*p];
>> -    type_precision* A = Q;
>> -    type_precision* AL = A;
>> -    type_precision* AR = &A[l*n];
>> -
>> -    type_precision tau[k];
>> -
>> -
>> -    int y_iters = y_amount / y_block_size;
>> -
>> -
>> -    type_precision* Y;
>> -    type_precision* backupAR;
>> -
>> -
>> -
>> -    // printf("\n\n%%Computations\n%%");
>> -
>> -    get_ticks(start_tick);
>> -    for (i = 0; i < a_iters; i++)
>> -    {
>> -        if (a_iters < 10 || (i%(a_iters/10)) == 0)
>> -        {
>> -            cout << "%" << flush;
>> -        }
>> -
>> -        copy_vec(backupAL, AL, n*l);
>> -
>> -        AIOfile.load_ARblock(&backupAR, a_block_size);
>> -        AIOfile.reset_Y();
>> -        copy_vec(backupAR, AR, n*r);
>> -
>> -        //! Generate R
>> -        info = LAPACKE_sgeqrf(STORAGE_TYPE, n, p, A, lda, tau);
>> -        assert(info == 0, "QR decomp");
>> -        type_precision* R = extract_R(A, n, p);
>> -
>> -        //! generate Q
>> -        info = LAPACKE_sorgqr(STORAGE_TYPE, n, p, k, Q, lda, tau);
>> -        assert(info == 0, "Q form");
>> -
>> -
>> -
>> -//        matlab_print_matrix("ALL", n, l, backupAL);
>> -//        matlab_print_matrix("ARR", n, r, backupAR);
>> -
>> -        for (j = 0; j < y_iters; j++)
>> -        {
>> -            // matlab_print_matrix("RR", p, p, R);
>> -            // matlab_print_matrix("QQ", n, p, Q);
>> -
>> -            AIOfile.load_Yblock(&Y, y_block_size);
>> -            type_precision Qy[y_block_size*p];
>> -
>> -            //! qy = Q'*Y
>> -            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
>> -                        p, y_block_size, n, 1.0, Q, n, Y, n, 0.0, Qy, p);
>> -
>> -            type_precision* B = Qy;
>> -
>> -            //! b = R\qy
>> -            cblas_strsm(CblasColMajor, CblasLeft, CblasUpper,
>> -                        CblasNoTrans, CblasNonUnit,
>> -                        p, y_block_size, 1.0, R, p, Qy, p);
>> -
>> -            if (ForceCheck)
>> -            {
>> -                check_result(backupAL, backupAR, n, p, y_block_size, r, Y, B);
>> -            }
>> -        }
>> -
>> -        free(R);
>> -    }
>> -
>> -    get_ticks(end_tick);
>> -    out.duration = ticks2sec(end_tick - start_tick, cpu_freq);
>> -    out.gflops = 2.0 * a_amount *
>> -        (2.0 * n * p * p / 1000.0 + y_amount / 1000.0 *
>> -         (p * n + p * p * p)) / 1000.0 / 1000.0;
>> -
>> -
>> -    AIOfile.finalize();
>> -
>> -
>> -    free(AL);
>> -    delete[] Q;
>> -}
>> -
>> -
>> -void Algorithm::partialQR(struct Settings params, struct Outputs &out)
>> -{
>> -    int max_threads = params.threads;
>> -
>> -
>> -    srand(time(NULL));
>> -
>> -    blas_set_num_threads(max_threads);
>> -
>> -
>> -
>> -    cputime_type start_tick, start_tick2, end_tick;
>> -    double acc_gemm = 0;
>> -    double acc_trsm = 0;
>> -    double acc_other = 0;
>> -    double acc_atemp = 0;
>> -    double acc_rtr = 0;
>> -    double acc_rqr = 0;
>> -
>> -
>> -    type_precision *RL, *Ytemp, *RtlRtr;
>> -    lapack_int info, n, lda, ldy, l, r, k, p;
>> -
>> -    int i, j, w;
>> -    int m;
>> -
>> -    AIOfile.initialize(params);
>> -
>> -    n = params.n; l = params.l; r = params.r; p = l+r;
>> -
>> -    int y_amount = params.m;
>> -    int y_block_size = params.mb;// kk
>> -
>> -    int a_amount = params.t;
>> -    int a_block_size = 1;
>> -
>> -    int a_iters = a_amount/a_block_size;
>> -
>> -    lda = n; ldy = n;
>> -    k = l;
>> -
>> -
>> -
>> -    type_precision* backupAL;
>> -
>> -    AIOfile.load_AL(&backupAL);
>> -
>> -    type_precision* Q = new type_precision[n*p];
>> -    type_precision* AL = Q;
>> -    copy_vec(backupAL, AL, n*l);
>> -
>> -
>> -    type_precision tau[k];
>> -
>> -    type_precision* QL;
>> -    type_precision* QR = &Q[n*l];
>> -
>> -    int y_iters = y_amount / y_block_size;
>> -
>> -
>> -    type_precision* Y;
>> -
>> -    type_precision Rtr[l*r];
>> -    type_precision* Ar;
>> -    type_precision Rbr[r*r];
>> -
>> -    get_ticks(start_tick);
>> -    //! Generate RTL
>> -    info = LAPACKE_sgeqrf(STORAGE_TYPE, n, l, AL, lda, tau);
>> -    assert(info == 0, "QR decomp");
>> -
>> -    type_precision* R =prepare_R(AL, n, l, r);
>> -    type_precision* Rtl = R;
>> -
>> -    QL = AL;// same as Q
>> -    AL = backupAL;
>> -    //! generate QL
>> -    info = LAPACKE_sorgqr(STORAGE_TYPE, n, l, k, QL, lda, tau);
>> -
>> -    assert(info == 0, "Q form");
>> -
>> -    // printf("\n\n%%Computations\n%%");
>> -
>> -
>> -    for (i = 0; i < a_iters; i++)
>> -    {
>> -        if (a_iters < 10 || (i%(a_iters/10)) == 0)
>> -        {
>> -            cout << "%" << flush;
>> -        }
>> -
>> -        AIOfile.load_ARblock(&Ar, a_block_size);
>> -        AIOfile.reset_Y();
>> -
>> -        //! Rtr = Q1'*AR
>> -        cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
>> -                    l, r*a_block_size, n, 1.0, QL, n, Ar, n, 0.0, Rtr, l);
>> -
>> -        type_precision* Atemp = replicate_vec(Ar, n*r*a_block_size);
>> -
>> -        //! Atemp = AR-Q1*Rtr
>> -        cblas_sgemm(CblasColMajor, CblasNoTrans, CblasNoTrans,
>> -                    n, r*a_block_size, l, -1.0, QL, n, Rtr, l, 1.0, Atemp, n);
>> -
>> -        type_precision* QrRbr = QR;
>> -        copy_vec(Atemp, QrRbr, n*r);
>> -        info = LAPACKE_sgeqrf(STORAGE_TYPE, n, r, QrRbr, lda, tau);
>> -        assert(info == 0, "QR decomp of QrRbr");
>> -
>> -        // reuse of old function
>> -        type_precision* Rbrtemp = prepare_R(QrRbr, n, r, 0);
>> -
>> -        copy_vec(Rbrtemp, Rbr, r*r);
>> -        info = LAPACKE_sorgqr(STORAGE_TYPE, n, r, r, QrRbr, lda, tau);
>> -        assert(info == 0, "QR form");
>> -        free(Rbrtemp);
>> -
>> -        update_R(R, Rtr, Rbr, p, p, r);
>> -
>> -//        matlab_print_matrix("RL", p, l, Rtl);
>> -//        matlab_print_matrix("Rtr", l, r, Rtr);
>> -//        matlab_print_matrix("Rbr", r, r, Rbr);
>> -//        matlab_print_matrix("RR", p, p, R);
>> -//        matlab_print_matrix("QQ", n, p, Q);
>> -        // matlab_print_matrix("ALL", n, l, AL);
>> -        // matlab_print_matrix("ARR", n, r, Ar);
>> -        // matlab_print_matrix("R", p, p, R);
>> -        int top_idx = 0;
>> -
>> -        for (j = 0; j < y_iters; j++)
>> -        {
>> -            AIOfile.load_Yblock(&Y, y_block_size);
>> -            type_precision Qy[y_block_size*p];
>> -
>> -
>> -            //! qy = Q'*Y
>> -            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
>> -                        p, y_block_size, n, 1.0, Q, n, Y, n, 0.0, Qy, p);
>> -
>> -            type_precision* B = Qy;
>> -
>> -            //! b = R\qy
>> -            cblas_strsm(CblasColMajor, CblasLeft, CblasUpper,
>> -                        CblasNoTrans, CblasNonUnit,
>> -                        p, y_block_size, 1.0, R, p, Qy, p);
>> -
>> -            // matlab_print_matrix("bcomp", p, y_block_size, B);
>> -
>> -
>> -            if ( ForceCheck)
>> -            {
>> -                check_result(AL, Ar, n, p, y_block_size, r, Y, B);
>> -            }
>> -        }
>> -
>> -        free(Atemp);
>> -    }
>> -
>> -
>> -    get_ticks(end_tick);
>> -    out.duration = ticks2sec(end_tick - start_tick, cpu_freq);
>> -    out.gflops = 2.0*(n*l*l/1000.0 + a_amount *
>> -                      (2.0 * n * l * r / 1000.0 +
>> -                       n *  r * r / 1000.0 +
>> -                       y_amount / 1000.0 * (p * n + p * p * p))) /
>> -        1000.0 / 1000.0;
>> -
>> -
>> -    AIOfile.finalize();
>> -
>> -
>> -    free(Rtl);
>> -    free(AL);
>> -    delete[] Q;
>> -}
>> -
>> -
>> -void  Algorithm::partialQR_Blocked_Rtl(struct Settings params,
>> -                                       struct Outputs &out)
>> -{
>> -    int max_threads = params.threads;
>> -
>> -    // srand (time(NULL));
>> -
>> -    int cpu_cores = max_threads;
>> -
>> -
>> -    blas_set_num_threads(max_threads);
>> -    omp_set_num_threads(max_threads);
>> -
>> -
>> -
>> -
>> -    cputime_type start_tick, start_tick2, end_tick;
>> -    double acc_pre     = 0;
>> -    double acc_trsm    = 0;
>> -    double acc_other   = 0;
>> -    double acc_atemp   = 0;
>> -    double acc_rtr     = 0;
>> -    double acc_rqr     = 0;
>> -    double acc_RTL_QLY = 0;
>> -
>> -
>> -    type_precision *AL, *RL, *RQy_top, *Ytemp, *RtlRtr;
>> -    lapack_int info, n, lda, ldy, l, r, k, p;
>> -
>> -    int i, j, w;
>> -    int m;
>> -
>> -
>> -    AIOfile.initialize(params);
>> -
>> -    n = params.n; l = params.l; r = params.r; p = l+r;
>> -
>> -
>> -
>> -    double lvl_3_cache_size = 6;
>> -
>> -
>> -    int y_amount = params.t;
>> -    int y_block_size = params.tb;// kk
>> -
>> -    int a_amount = params.m;
>> -    int a_block_size = params.mb;
>> -
>> -    int a_iters = (a_amount + a_block_size - 1) / a_block_size;
>> -
>> -    int y_iters = (y_amount+ y_block_size - 1)/y_block_size;
>> -
>> -    int qy_idx = 0;
>> -
>> -    out.gflops = n/1000.0*l/1000.0*l/1000.0 +
>> -        gemm_flops(l, n, y_amount, 0) +
>> -        y_amount*l/1000.0*l/1000.0/1000.0 +
>> -        a_amount * (gemm_flops(l, n, r, 0) +
>> -                    l/1000.0*l/1000.0/1000.0 +
>> -                    gemm_flops(n, l, r, 1) +
>> -                    n/1000.0*r/1000.0*r/1000.0 +
>> -                    gemm_flops(r, n, y_amount, 0) +
>> -                    y_amount/1000.0*r/1000.0*r/1000.0 +
>> -                    y_amount/1000.0*l/1000.0/1000.0);
>> -
>> -    int sch_block_size = max(1.0,
>> -                             min((double)a_block_size / (double)max_threads,
>> -                                 (1.0*1000*1000) /
>> -                                 (double)((sizeof(type_precision) * n * r))));
>> -
>> -    // cout << endl<< "taskChunk "<< sch_block_size <<endl;
>> -
>> -    lda = n; ldy = n;
>> -    k = l;
>> -
>> -    AIOfile.load_AL(&AL);
>> -
>> -    type_precision* backupAL = replicate_vec(AL, n*l);
>> -
>> -
>> -
>> -    RQy_top = new type_precision[l*y_amount];
>> -
>> -    type_precision* Bfinal = (type_precision*)malloc(max_threads * p *
>> -                                                     y_block_size *
>> -                                                     sizeof(type_precision));
>> -
>> -    type_precision tau[k];
>> -
>> -    type_precision* QL;
>> -
>> -    type_precision* Rtr = (type_precision*)malloc(l * r * a_block_size *
>> -                                                  sizeof(type_precision));
>> -    type_precision* Qy_bot = (type_precision*)malloc(a_block_size *
>> -                                                     y_block_size * r *
>> -                                                     sizeof(type_precision));
>> -
>> -    type_precision* B_top = new type_precision[max_threads*y_block_size*l];
>> -    type_precision* B_bot = new type_precision[max_threads*y_block_size*r];
>> -
>> -    type_precision* Ar;
>> -    type_precision Rbr[r*r*a_block_size];
>> -
>> -
>> -
>> -    // printf("\n\n%%Computations\n%%");
>> -    for (i = 0; i < a_iters; i++)
>> -    {
>> -        if (a_iters >= 10 && (i%(a_iters/10)) == 0)
>> -        {
>> -            // cout << "*" << flush;
>> -        }
>> -
>> -        for (j = 0; j < y_iters; j++)
>> -        {
>> -            if (a_iters < 10 && (y_iters < 10 || (j%(y_iters/10)) == 0))
>> -            {
>> -                // cout << "*" << flush;
>> -            }
>> -        }
>> -    }
>> -    cout << endl;
>> -
>> -
>> -    //! Start of Computations
>> -
>> -    get_ticks(start_tick);
>> -
>> -
>> -    info = LAPACKE_sgeqrf(STORAGE_TYPE, n, l, AL, lda, tau);
>> -    assert(info == 0, "QR decomp");
>> -
>> -    type_precision* Rtl = prepare_R(AL, n, l, 0);
>> -
>> -    QL = AL;
>> -    AL = backupAL;
>> -
>> -    info = LAPACKE_sorgqr(STORAGE_TYPE, n, l, k, QL, lda, tau);
>> -
>> -    assert(info == 0, "Q form");
>> -
>> -    matlab_print_matrix("QL", n, l, QL);
>> -
>> -    qy_idx = 0;
>> -
>> -    // printf("\n%%Preparing IO\n");
>> -    //! Prepare File for the Y
>> -
>> -
>> -    type_precision* Y;
>> -
>> -    // #pragma omp parallel default(shared)
>> +//    int max_threads = params.threads;
>> +//
>> +//    srand(time(NULL));
>> +//
>> +//    blas_set_num_threads(max_threads);
>> +//
>> +//    type_precision *Ytemp;
>> +//    lapack_int info, n, lda,ldy, l, r, p;
>> +//
>> +//    int i,k;
>> +//
>> +//    cputime_type start_tick, start_tick2, end_tick;
>> +//
>> +//    AIOfile.initialize(params);
>> +//
>> +//    n = params.n; l = params.l; r = params.r; p = l+r;
>> +//
>> +//    int y_amount = params.m;
>> +//    int y_block_size = params.mb;  // kk
>> +//
>> +//    int a_amount = params.t;
>> +//    int a_block_size = params.tb;
>> +//
>> +//    int a_iters = (a_amount + a_block_size - 1) / a_block_size;
>> +//
>> +//    lda = n; ldy = n;
>> +//    k = p;
>> +//
>> +//
>> +//
>> +//    type_precision* backupAL;
>> +//
>> +//    AIOfile.load_AL(&backupAL);
>> +//
>> +//
>> +//    type_precision Stl[l * l];
>> +//    type_precision Str[l * r * a_block_size];
>> +//
>> +//
>> +//    type_precision* Ay_top = new type_precision[l * y_amount];
>> +//    type_precision* Ay_bot = new type_precision[y_block_size * a_block_size * r];
>> +//    type_precision* A = new type_precision[n * p];
>> +//    type_precision* AR = new type_precision[n * r * a_block_size];
>> +//    type_precision* AL = new type_precision[n * l];
>> +//
>> +//    copy_vec(backupAL, AL, n * l);
>> +//    copy_vec(backupAL, A, n * l);
>> +//
>> +//
>> +//    int y_iters = y_amount/y_block_size;
>> +//
>> +//    type_precision* Y;
>> +//    type_precision* backupAR;
>> +//
>> +//    // printf("\n\n%%Computations\n%%");
>> +//
>> +//
>> +//    get_ticks(start_tick);
>> +//
>> +//
>> +//    //! Generate S
>> +//    cblas_ssyrk(CblasColMajor, CblasUpper, CblasTrans,
>> +//                l, n, 1.0, backupAL, lda, 0.0, Stl, l);
>> +//
>> +//    for (int j = 0; j < y_iters; j++)
>>  //    {
>> -//        // #pragma omp for private(Y, y_block_size, qy_idx) nowait schedule(static, 1)
>> +//        AIOfile.load_Yblock(&Y, y_block_size);
>> +//
>> +//        //! Ay_top = AL'*Y
>> +//        cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
>> +//                    l, y_block_size, n, 1.0, AL, n, Y, n, 0.0,
>> +//                    &Ay_top[j * l * y_block_size], l);
>> +//    }
>> +//
>> +//
>> +//    for (int i = 0; i < a_iters; i++)
>> +//    {
>> +//        if (a_iters > 10 && (i%(a_iters/10)) == 0)
>> +//        {
>> +//            cout << "%" << flush;
>> +//        }
>> +//
>> +//
>> +//        AIOfile.load_ARblock(&backupAR, a_block_size);
>> +//        AIOfile.reset_Y();
>> +//        copy_vec(backupAR, AR, n*r*a_block_size);
>> +//
>> +//        // matlab_print_matrix("A", n, p, A);
>> +//        //! Generate Str
>> +//        cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
>> +//                    l, r * a_block_size, n, 1.0, AL, n, AR, n, 0.0, Str, l);
>> +//
>> +//        type_precision Sbr[r*r*a_block_size];
>> +//
>> +//        #pragma omp parallel for default(shared) schedule(static)
>> +//        for (int ii= 0; ii < a_block_size; ii++)
>> +//        {
>> +//            //! Generate Sbr
>> +//            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
>> +//                        r, r, n, 1.0, &AR[ii*r*n], n, &AR[ii*r*n],
>> +//                        n, 0.0, &Sbr[ii*r*r], r);
>> +//        }
>> +//
>> +//        type_precision Ay[y_block_size*p];
>> +//        type_precision S[p*p];
>> +//
>> +//
>>  //        for (int j = 0; j < y_iters; j++)
>>  //        {
>> -//            // qy_idx = j*y_block_size*l;
>> -//            // #pragma omp critical
>> -//            {AIOfile.load_Yblock(&Y, y_block_size);}
>> +//            if (a_iters < 10 && (y_iters < 10 || (j%(y_iters/10)) == 0))
>> +//            {
>> +//                cout << "%" << flush;
>> +//            }
>>  //
>> -//            //! qy_top = QL'*Y
>> +//
>> +//            AIOfile.load_Yblock(&Y, y_block_size);
>> +//
>> +//            //! Ay_bot = AR'*Y
>>  //            cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
>> -//                            l, y_block_size, n, 1.0, QL, n, Y, n, 0.0,&RQy_top[qy_idx], l);
>> +//                        r * a_block_size, y_block_size, n, 1.0, AR, n, Y, n,
>> +//                        0.0, Ay_bot, r * a_block_size);
>>  //
>> -//            //! K | RtlQlY =RTL\qy_top
>> -//            cblas_strsm(CblasColMajor, CblasLeft, CblasUpper, CblasNoTrans, CblasNonUnit,
>> -//                                            l, y_block_size, 1.0, Rtl, l,&RQy_top[qy_idx], l);
>> -//            qy_idx += y_block_size*l;
>> +//            #pragma omp parallel for private(S, Ay) default(shared) schedule(static)
>> +//            for (int ii= 0; ii < a_block_size; ii++)
>> +//            {
>> [TRUNCATED]
>>
>> To get the complete diff run:
>>     svnlook diff /svnroot/genabel -r 1609
>> _______________________________________________
>> Genabel-commits mailing list
>> Genabel-commits at lists.r-forge.r-project.org
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
>>
>
>
>
> _______________________________________________
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
>

--
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140217/1c5dfbc2/attachment-0001.sig>

From lennart at karssen.org  Tue Feb 18 13:56:15 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Tue, 18 Feb 2014 13:56:15 +0100
Subject: [GenABEL-dev] [Genabel-commits] r1611 - in
 branches/ProbABEL-0.50: checks/R-tests examples src
In-Reply-To: <20140214141421.B62EB185DF0@r-forge.r-project.org>
References: <20140214141421.B62EB185DF0@r-forge.r-project.org>
Message-ID: <5303586F.4000801@karssen.org>

Hi Maarten,

I found this commit e-mail in my inbox still waiting for review.

Please have a look at the comments below.


Thanks a lot for your contributions,

Lennart.

On 14-02-14 15:14, noreply at r-forge.r-project.org wrote:
> Author: maartenk
> Date: 2014-02-14 15:14:21 +0100 (Fri, 14 Feb 2014)
> New Revision: 1611
> 
> Modified:
>    branches/ProbABEL-0.50/checks/R-tests/run_models_in_R_palinear.R
>    branches/ProbABEL-0.50/examples/mmscore.R
>    branches/ProbABEL-0.50/src/reg1.cpp
> Log:
> -removed unnecessary lines from examples/mmscore.R

Nice! Having this lying around in the examples would only confuse a user.


> -Disabled in run_models_in_R_palinear.R last snp in 2df model because 2 variables were completely anti correlated. There should be a check on Rank of Cholesky decompositions for this kind of problem

I agree that such a check would be appropriate. Doesn't EIGEN give a
warning? Maybe someone else knows how we can tell R's lm() or glm() to
do these kinds of checks?

> -rewrote mmscore weighted regression to EIGEN form for palinear
> 
> Modified: branches/ProbABEL-0.50/checks/R-tests/run_models_in_R_palinear.R
> ===================================================================
> --- branches/ProbABEL-0.50/checks/R-tests/run_models_in_R_palinear.R	2014-02-14 07:58:43 UTC (rev 1610)
> +++ branches/ProbABEL-0.50/checks/R-tests/run_models_in_R_palinear.R	2014-02-14 14:14:21 UTC (rev 1611)
> @@ -123,7 +123,7 @@
>  }
>  colnames(prob.2df.R) <- cols2df
>  rownames(prob.2df.R) <- NULL
> -stopifnot( all.equal(prob.2df.PA, prob.2df.R, tol=tol) )
> +#stopifnot( all.equal(prob.2df.PA1[1:5,], prob.2df.R[1:5,], tol=tol) )
>  cat("2df\n")
>  
>  cat("\t\t\t\t\t\tOK\n")
> 
> Modified: branches/ProbABEL-0.50/examples/mmscore.R
> ===================================================================
> --- branches/ProbABEL-0.50/examples/mmscore.R	2014-02-14 07:58:43 UTC (rev 1610)
> +++ branches/ProbABEL-0.50/examples/mmscore.R	2014-02-14 14:14:21 UTC (rev 1611)
> @@ -90,83 +90,3 @@
>  
>  ## Mow, go to ProbABEL and start analysis
>  
> -
> -
> -
> -
> -##____________________________________________________
> -## The following part is for historic purposes only. It is not
> -## necessary for using the --mmscore option of ProbABEL's palinear.
> -
> -## Create test file with genotypes
> -
> -data_cut   <- data[, snps]
> -gen_table  <- as.numeric(data_cut)
> -prob_table <- matrix()
> -
> -#Replace NA by mean for each snp. NA is forbiden in genotypes in ProbABEL input (!).
> -#for(snpnum in 1:dim(gen_table)[2])
> -#	{
> -#	mean <- mean(gen_table[,snpnum], na.rm=T)
> -#	gen_table[is.na(gen_table[,snpnum]),snpnum]	<- mean
> -#	}
> -
> -
> -gen_table_df        <- data.frame(gen_table)
> -gen_table_df$MLDOSE <- gen_table_df[, 1]
> -gen_table_df[,1]    <- "MLDOSE"
> -colnam <- colnames(gen_table_df)
> -
> -#colnam[-c(1:length(colnam)-1)] <- colnam[1]
> -#colnam[1] <- "MLDOSE"
> -
> -colnam[length(colnam)] <- colnam[1]
> -colnam[1]              <- "MLDOSE"
> -colnames(gen_table_df) <- colnam
> -
> -
> -rownames <- rownames(gen_table_df)
> -rownames <- paste("1->", rownames, sep="")
> -rownames(gen_table_df) <- rownames
> -
> -write.table(file="mmscore_gen.mldose",
> -            gen_table_df,
> -            row.names=TRUE,
> -            col.names=FALSE,
> -            quote=FALSE,
> -            na="NaN")
> -
> -mlinfo <- data.frame(SNP=colnam[2:length(colnam)])
> -mlinfo$Al1 <- "A"
> -mlinfo$Al2 <- "B"
> -mlinfo$Freq1 <- 0.5847
> -mlinfo$MAF <- 0.5847
> -mlinfo$Quality <- 0.5847
> -mlinfo$Rsq <- 0.5847
> -
> -write.table(mlinfo, "mmscore_gen.mlinfo", row.names=FALSE, quote=FALSE)
> -
> -## arrange probability-file
> -prob_table <- matrix(NA,
> -                     ncol=(dim(gen_table_df)[2]-1) * 2,
> -                     nrow=dim(gen_table_df)[1])
> -j <- 1
> -for (i in (1:(dim(gen_table_df)[2]-1))) {
> -    prob_table[,j] <- rep(0, dim(prob_table)[1])
> -    prob_table[gen_table_df[,i+1]==2, j] <- 1
> -    prob_table[is.na(gen_table_df[, i+1]), j] <- NA
> -    j <- j + 1
> -    prob_table[, j] <- rep(0,dim(prob_table)[1])
> -    prob_table[gen_table_df[, i+1]==1, j] <- 1
> -    prob_table[is.na(gen_table_df[, i+1]),j] <- NA
> -    j <- j + 1
> -}
> -prob_table_df <- data.frame(MLPROB="MLPROB", prob_table)
> -rownames(prob_table_df) <- rownames(gen_table_df)
> -
> -write.table(file="mmscore_gen.mlprob",
> -            prob_table_df,
> -            row.names=TRUE,
> -            col.names=FILE,
> -            quote=FILE,
> -            na="NaN")
> 
> Modified: branches/ProbABEL-0.50/src/reg1.cpp
> ===================================================================
> --- branches/ProbABEL-0.50/src/reg1.cpp	2014-02-14 07:58:43 UTC (rev 1610)
> +++ branches/ProbABEL-0.50/src/reg1.cpp	2014-02-14 14:14:21 UTC (rev 1611)
> @@ -355,9 +355,11 @@
>      }
>  
>      double sigma2_internal;
> -    mematrix<double> tXX_i;
> +
>  #if EIGEN
>      LDLT <MatrixXd> Ch ;
> +#else
> +    mematrix<double> tXX_i;
>  #endif
>      if (invvarmatrixin.length_of_mask != 0)
>      {
> @@ -369,12 +371,16 @@
>          //C=AB (m X n matrix A and n x P matrix B)
>          //flops=mp(2n-1) (when n is big enough flops=mpn2)
>          //Oct 26, 2009
> +
> +#if EIGEN
> +        MatrixXd tXW = X.data.transpose() * invvarmatrixin.masked_data->data; // flops 5997000
> +        Ch = LDLT <MatrixXd>(tXW * X.data); // 17991 flops
> +        beta.data = Ch.solve(tXW * reg_data.Y.data);
> +        sigma2 = (reg_data.Y.data - tXW.transpose() * beta.data).squaredNorm() ; //flops: 1000+5000+3000
> +#else
> +
>  

Are you sure you want to keep these flops values here? I guess they are
no longer accurate when using EIGEN.


>        mematrix<double> tXW = transpose(X) * invvarmatrixin.masked_data; // flops 5997000
>          tXX_i = tXW * X;        // 17991 flops
> -#if EIGEN
> -        Ch=LDLT <MatrixXd>(tXX_i.data.selfadjointView<Lower>());
> -#endif
> -
>          // use cholesky to invert
>          cholesky2_mm(tXX_i, tol_chol);
>          chinv2_mm(tXX_i);
> @@ -387,6 +393,7 @@
>              double val = sigma2_matrix.get(i, 0);
>              sigma2 += val * val; // flops: 3000
>          }
> +#endif
>          double N = X.nrow;
>          //sigma2_internal = sigma2 / (N - static_cast<double>(length_beta));
>          // Ugly fix to the fact that if we do mmscore, sigma2 is already
> @@ -400,15 +407,12 @@
>      {
>  #if EIGEN
>          int m = X.ncol;
> -        MatrixXd txx = MatrixXd(m, m).setZero().selfadjointView<Lower>().rankUpdate(X.data.adjoint());
> +        MatrixXd txx = MatrixXd(m, m).setZero().selfadjointView<Lower>().\
> +                rankUpdate(X.data.adjoint());
>          Ch=LDLT <MatrixXd>(txx.selfadjointView<Lower>());
>          beta.data= Ch.solve(X.data.adjoint() * reg_data.Y.data);
> +        sigma2 = (reg_data.Y.data - (X.data * beta.data)).squaredNorm() ;
>  
> -       tXX_i.data=Ch.solve(MatrixXd(m, m).Identity(m,m));
> -       tXX_i.nrow=tXX_i.data.rows();
> -       tXX_i.ncol=tXX_i.data.cols();
> -       tXX_i.nelements=tXX_i.ncol*tXX_i.nrow;
> -
>  #else
>          mematrix<double> tX = transpose(X);
>          // use cholesky to invert
> @@ -416,14 +420,10 @@
>                  cholesky2_mm(tXX_i, tol_chol);
>                  chinv2_mm(tXX_i);
>                  beta = tXX_i * (tX * (reg_data.Y));
> -#endif
>  
>          // now compute residual variance
>          sigma2 = 0.;
>          mematrix<double> sigma2_matrix = reg_data.Y - (X * beta);
> -#if EIGEN
> -        sigma2 = sigma2_matrix.data.squaredNorm() ;
> -#else
>          for (int i = 0; i < sigma2_matrix.nrow; i++)
>          {
>              double val = sigma2_matrix.get(i, 0);
> @@ -458,8 +458,10 @@
>      double intercept = beta.get(0, 0);
>      residuals.data= reg_data.Y.data.array()-intercept;
>      //matrix.
> -    ArrayXXd betacol = beta.data.block(1,0,beta.data.rows()-1,1).array().transpose();
> -    ArrayXXd resid_sub = (X.data.block(0,1,X.data.rows(),X.data.cols()-1)*betacol.matrix().asDiagonal()).rowwise().sum() ;
> +    ArrayXXd betacol = beta.data.block(1,0,beta.data.rows()-1,1)\
> +            .array().transpose();
> +    ArrayXXd resid_sub = (X.data.block(0,1,X.data.rows(),X.data.cols()-1)\
> +            *betacol.matrix().asDiagonal()).rowwise().sum() ;
>      //std::cout << resid_sub << std::endl;
>      residuals.data-=resid_sub.matrix();
>      //residuals[i] -= resid_sub;
> @@ -481,15 +483,18 @@
>  
>      loglik -= static_cast<double>(reg_data.nids) * log(sqrt(sigma2));
>  #if EIGEN
> -    MatrixXd tXX_inv=Ch.solve(MatrixXd(length_beta, length_beta).Identity(length_beta,length_beta));
> +    MatrixXd tXX_inv=Ch.solve(MatrixXd(length_beta, length_beta).
> +            Identity(length_beta,length_beta));
>  #endif
>  
>      mematrix<double> robust_sigma2(X.ncol, X.ncol);
>      if (robust)
>      {
>  #if EIGEN
> -        MatrixXd Xresiduals = X.data.array().colwise()*residuals.data.col(0).array();
> -        MatrixXd  XbyR = MatrixXd(X.ncol, X.ncol).setZero().selfadjointView<Lower>().rankUpdate(Xresiduals.adjoint());
> +        MatrixXd Xresiduals = X.data.array().colwise()\
> +                *residuals.data.col(0).array();
> +        MatrixXd  XbyR = MatrixXd(X.ncol, X.ncol).setZero()\
> +                .selfadjointView<Lower>().rankUpdate(Xresiduals.adjoint());
>          robust_sigma2.data= tXX_inv*XbyR *tXX_inv;
>  #else
>  
> 
> _______________________________________________
> Genabel-commits mailing list
> Genabel-commits at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
> 


Best regards,

Lennart.

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140218/4c2656fe/attachment.sig>

From lennart at karssen.org  Tue Feb 18 13:59:05 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Tue, 18 Feb 2014 13:59:05 +0100
Subject: [GenABEL-dev] [Genabel-commits] r1589 -
	branches/ProbABEL-0.50/src
In-Reply-To: <20140203222918.2BF0018663E@r-forge.r-project.org>
References: <20140203222918.2BF0018663E@r-forge.r-project.org>
Message-ID: <53035919.2060307@karssen.org>

Hi Maarten,

This is another even older commit waiting for review.
One question: In the log message you mention this leads to minor memory
leaks. Any idea why? I assume they have been detected by valgrind? Or
was it cppcheck that warns about them?

Or have they been resolved in the mean time?


Thanks,

Lennart.


On 03-02-14 23:29, noreply at r-forge.r-project.org wrote:
> Author: maartenk
> Date: 2014-02-03 23:29:17 +0100 (Mon, 03 Feb 2014)
> New Revision: 1589
> 
> Modified:
>    branches/ProbABEL-0.50/src/main.cpp
>    branches/ProbABEL-0.50/src/reg1.cpp
>    branches/ProbABEL-0.50/src/reg1.h
>    branches/ProbABEL-0.50/src/regdata.cpp
>    branches/ProbABEL-0.50/src/regdata.h
> Log:
> Removes mutiple masking the data per snp in linear and logistic regression. This commit speeds up palinear by ~11%. This introduces minor memory leaks.
> 
> Modified: branches/ProbABEL-0.50/src/main.cpp
> ===================================================================
> --- branches/ProbABEL-0.50/src/main.cpp	2014-02-03 22:05:56 UTC (rev 1588)
> +++ branches/ProbABEL-0.50/src/main.cpp	2014-02-03 22:29:17 UTC (rev 1589)
> @@ -126,7 +126,8 @@
>      std::cout << " loaded null data..." << std::flush;
>  #if LOGISTIC
>      logistic_reg nrd = logistic_reg(nrgd);
> -    nrd.estimate(nrgd, 0, MAXITER, EPS, CHOLTOL, 0,
> +
> +    nrd.estimate( 0, MAXITER, EPS, CHOLTOL, 0,
>                   input_var.getInteraction(),
>                   input_var.getNgpreds(),
>                   invvarmatrix,
> @@ -138,7 +139,7 @@
>  #if DEBUG
>      std::cout << "[DEBUG] linear_reg nrd = linear_reg(nrgd); DONE.";
>  #endif
> -    nrd.estimate(nrgd, 0, CHOLTOL, 0, input_var.getInteraction(),
> +    nrd.estimate( 0, CHOLTOL, 0, input_var.getInteraction(),
>                   input_var.getNgpreds(), invvarmatrix,
>                   input_var.getRobust(), 1);
>  #elif COXPH
> @@ -272,14 +273,14 @@
>                  logistic_reg rd(rgd);
>                  if (input_var.getScore())
>                  {
> -                    rd.score(nrd.residuals, rgd, 0, CHOLTOL, model,
> +                    rd.score(nrd.residuals,  0, CHOLTOL, model,
>                               input_var.getInteraction(),
>                               input_var.getNgpreds(),
>                               invvarmatrix);
>                  }
>                  else
>                  {
> -                    rd.estimate(rgd, 0, MAXITER, EPS, CHOLTOL, model,
> +                    rd.estimate( 0, MAXITER, EPS, CHOLTOL, model,
>                                  input_var.getInteraction(),
>                                  input_var.getNgpreds(),
>                                  invvarmatrix,
> @@ -289,14 +290,14 @@
>                  linear_reg rd(rgd);
>                  if (input_var.getScore())
>                  {
> -                    rd.score(nrd.residuals, rgd, 0, CHOLTOL, model,
> +                    rd.score(nrd.residuals, 0, CHOLTOL, model,
>                               input_var.getInteraction(),
>                               input_var.getNgpreds(),
>                               invvarmatrix);
>                  }
>                  else
>                  {
> -                    rd.estimate(rgd, 0, CHOLTOL, model,
> +                    rd.estimate( 0, CHOLTOL, model,
>                                  input_var.getInteraction(),
>                                  input_var.getNgpreds(),
>                                  invvarmatrix,
> @@ -380,7 +381,7 @@
>                              regdata new_rgd = rgd;
>                              new_rgd.remove_snp_from_X();
>                              linear_reg new_null_rd(new_rgd);
> -                            new_null_rd.estimate(new_rgd, 0,
> +                            new_null_rd.estimate( 0,
>                                                   CHOLTOL, model,
>                                                   input_var.getInteraction(),
>                                                   input_var.getNgpreds(),
> @@ -391,7 +392,7 @@
>                              regdata new_rgd = rgd;
>                              new_rgd.remove_snp_from_X();
>                              logistic_reg new_null_rd(new_rgd);
> -                            new_null_rd.estimate(new_rgd, 0, MAXITER, EPS,
> +                            new_null_rd.estimate( 0, MAXITER, EPS,
>                                                   CHOLTOL, model,
>                                                   input_var.getInteraction(),
>                                                   input_var.getNgpreds(),
> 
> Modified: branches/ProbABEL-0.50/src/reg1.cpp
> ===================================================================
> --- branches/ProbABEL-0.50/src/reg1.cpp	2014-02-03 22:05:56 UTC (rev 1588)
> +++ branches/ProbABEL-0.50/src/reg1.cpp	2014-02-03 22:29:17 UTC (rev 1589)
> @@ -224,10 +224,10 @@
>  }
>  
>  linear_reg::linear_reg(regdata& rdatain) {
> -    regdata rdata = rdatain.get_unmasked_data();
> +    reg_data = rdatain.get_unmasked_data();
>      // std::cout << "linear_reg: " << rdata.nids << " " << (rdata.X).ncol
>      //           << " " << (rdata.Y).ncol << "\n";
> -    int length_beta = (rdata.X).ncol;
> +    int length_beta = (reg_data.X).ncol;
>      beta.reinit(length_beta, 1);
>      sebeta.reinit(length_beta, 1);
>      //Han Chen
> @@ -236,18 +236,18 @@
>          covariance.reinit(length_beta - 1, 1);
>      }
>      //Oct 26, 2009
> -    residuals.reinit(rdata.nids, 1);
> +    residuals.reinit(reg_data.nids, 1);
>      sigma2 = -1.;
>      loglik = -9.999e+32;
>      chi2_score = -1.;
>  }
>  
> -void base_reg::base_score(mematrix<double>& resid, regdata& rdata, int verbose,
> +void base_reg::base_score(mematrix<double>& resid, int verbose,
>          double tol_chol, int model, int interaction, int ngpreds,
>          const masked_matrix& invvarmatrix, int nullmodel) {
> -    mematrix<double> oX = rdata.extract_genotypes();
> +    mematrix<double> oX = reg_data.extract_genotypes();
>      mematrix<double> X = apply_model(oX, model, interaction, ngpreds,
> -            rdata.is_interaction_excluded, false, nullmodel);
> +            reg_data.is_interaction_excluded, false, nullmodel);
>      beta.reinit(X.ncol, 1);
>      sebeta.reinit(X.ncol, 1);
>      double N = static_cast<double>(resid.nrow);
> @@ -286,31 +286,31 @@
>  }
>  
>  
> -void linear_reg::estimate(regdata& rdatain, int verbose, double tol_chol,
> +void linear_reg::estimate( int verbose, double tol_chol,
>          int model, int interaction, int ngpreds, masked_matrix& invvarmatrixin,
>          int robust, int nullmodel) {
>      // suda interaction parameter
>      // model should come here
> -    regdata rdata = rdatain.get_unmasked_data();
> +    //regdata rdata = rdatain.get_unmasked_data();
>  
>      if (verbose)
>      {
> -        cout << rdata.is_interaction_excluded
> +        cout << reg_data.is_interaction_excluded
>                  << " <-rdata.is_interaction_excluded\n";
>          // std::cout << "invvarmatrix:\n";
>          // invvarmatrixin.masked_data->print();
>          std::cout << "rdata.X:\n";
> -        rdata.X.print();
> +        reg_data.X.print();
>      }
>  
> -    mematrix<double> X = apply_model(rdata.X, model, interaction, ngpreds,
> -            rdata.is_interaction_excluded, false, nullmodel);
> +    mematrix<double> X = apply_model(reg_data.X, model, interaction, ngpreds,
> +            reg_data.is_interaction_excluded, false, nullmodel);
>      if (verbose)
>      {
>          std::cout << "X:\n";
>          X.print();
>          std::cout << "Y:\n";
> -        rdata.Y.print();
> +        reg_data.Y.print();
>      }
>  
>      int length_beta = X.ncol;
> @@ -337,7 +337,7 @@
>      if (invvarmatrixin.length_of_mask != 0)
>      {
>          //retrieve masked data W
> -        invvarmatrixin.update_mask(rdatain.masked_data);
> +        invvarmatrixin.update_mask(reg_data.masked_data);
>  
>          // This regression is Weighted Least Square: used for mmscore :
>          // FLOPS count are calculated for 3*1000 matrix as follow:
> @@ -353,10 +353,10 @@
>          // use cholesky to invert
>          cholesky2_mm(tXX_i, tol_chol);
>          chinv2_mm(tXX_i);
> -        beta = tXX_i * (tXW * rdata.Y);        // flops 15+5997
> +        beta = tXX_i * (tXW * reg_data.Y);        // flops 15+5997
>          // now compute residual variance
>          sigma2 = 0.;
> -        mematrix<double> sigma2_matrix = rdata.Y - (transpose(tXW) * beta); //flops: 1000+5000
> +        mematrix<double> sigma2_matrix = reg_data.Y - (transpose(tXW) * beta); //flops: 1000+5000
>          for (int i = 0; i < sigma2_matrix.nrow; i++)
>          {
>              double val = sigma2_matrix.get(i, 0);
> @@ -377,7 +377,7 @@
>          int m = X.ncol;
>          MatrixXd txx = MatrixXd(m, m).setZero().selfadjointView<Lower>().rankUpdate(X.data.adjoint());
>          Ch=LDLT <MatrixXd>(txx.selfadjointView<Lower>());
> -        beta.data= Ch.solve(X.data.adjoint() * rdata.Y.data);
> +        beta.data= Ch.solve(X.data.adjoint() * reg_data.Y.data);
>  
>         tXX_i.data=Ch.solve(MatrixXd(m, m).Identity(m,m));
>         tXX_i.nrow=tXX_i.data.rows();
> @@ -390,12 +390,12 @@
>                  tXX_i = tX * X;
>                  cholesky2_mm(tXX_i, tol_chol);
>                  chinv2_mm(tXX_i);
> -                beta = tXX_i * (tX * (rdata.Y));
> +                beta = tXX_i * (tX * (reg_data.Y));
>  #endif
>  
>          // now compute residual variance
>          sigma2 = 0.;
> -        mematrix<double> sigma2_matrix = rdata.Y - (X * beta);
> +        mematrix<double> sigma2_matrix = reg_data.Y - (X * beta);
>  #if EIGEN
>          sigma2 = sigma2_matrix.data.squaredNorm() ;
>  #else
> @@ -431,7 +431,7 @@
>  
>  #if EIGEN
>      double intercept = beta.get(0, 0);
> -    residuals.data= rdata.Y.data.array()-intercept;
> +    residuals.data= reg_data.Y.data.array()-intercept;
>      //matrix.
>      ArrayXXd betacol = beta.data.block(1,0,beta.data.rows()-1,1).array().transpose();
>      ArrayXXd resid_sub = (X.data.block(0,1,X.data.rows(),X.data.cols()-1)*betacol.matrix().asDiagonal()).rowwise().sum() ;
> @@ -443,9 +443,9 @@
>      //loglik -= halfrecsig2 * residuals[i] * residuals[i];
>  
>  #else
> -    for (int i = 0; i < rdata.nids; i++)
> +    for (int i = 0; i < reg_data.nids; i++)
>       {
> -         double resid = rdata.Y[i] - beta.get(0, 0); // intercept
> +         double resid = reg_data.Y[i] - beta.get(0, 0); // intercept
>           for (int j = 1; j < beta.nrow; j++){
>               resid -= beta.get(j, 0) * X.get(i, j);
>           }
> @@ -454,7 +454,7 @@
>       }
>  #endif
>  
> -    loglik -= static_cast<double>(rdata.nids) * log(sqrt(sigma2));
> +    loglik -= static_cast<double>(reg_data.nids) * log(sqrt(sigma2));
>  #if EIGEN
>      MatrixXd tXX_inv=Ch.solve(MatrixXd(length_beta, length_beta).Identity(length_beta,length_beta));
>  #endif
> @@ -576,17 +576,17 @@
>  
>  }
>  
> -void linear_reg::score(mematrix<double>& resid, regdata& rdatain, int verbose,
> +void linear_reg::score(mematrix<double>& resid, int verbose,
>          double tol_chol, int model, int interaction, int ngpreds,
>          const masked_matrix& invvarmatrix, int nullmodel) {
> -    regdata rdata = rdatain.get_unmasked_data();
> -    base_score(resid, rdata, verbose, tol_chol, model, interaction, ngpreds,
> +   // regdata rdata = rdatain.get_unmasked_data();
> +    base_score(resid, verbose, tol_chol, model, interaction, ngpreds,
>              invvarmatrix, nullmodel = 0);
>  }
>  
>  logistic_reg::logistic_reg(regdata& rdatain) {
> -    regdata rdata = rdatain.get_unmasked_data();
> -    int length_beta = (rdata.X).ncol;
> +    reg_data = rdatain.get_unmasked_data();
> +    int length_beta = reg_data.X.ncol;
>      beta.reinit(length_beta, 1);
>      sebeta.reinit(length_beta, 1);
>      //Han Chen
> @@ -595,37 +595,38 @@
>          covariance.reinit(length_beta - 1, 1);
>      }
>      //Oct 26, 2009
> -    residuals.reinit((rdata.X).nrow, 1);
> +    residuals.reinit(reg_data.X.nrow, 1);
>      sigma2 = -1.;
>      loglik = -9.999e+32; // should actually be MAX of the corresponding type
>      niter = -1;
>      chi2_score = -1.;
>  }
>  
> -void logistic_reg::estimate(regdata& rdatain, int verbose, int maxiter,
> +void logistic_reg::estimate( int verbose, int maxiter,
>          double eps, double tol_chol, int model, int interaction, int ngpreds,
>          masked_matrix& invvarmatrixin, int robust, int nullmodel) {
>      // In contrast to the 'linear' case 'invvarmatrix' contains the
>      // inverse of correlation matrix (not the inverse of var-cov matrix)
>      // h2.object$InvSigma * h.object2$h2an$estimate[length(h2$h2an$estimate)]
>      // the inverse of var-cov matrix scaled by total variance
> -    regdata rdata = rdatain.get_unmasked_data();
> +    //regdata rdata = rdatain.get_unmasked_data();
>      // a lot of code duplicated between linear and logistic...
>      // e.g. a piece below...
>      mematrix<double> invvarmatrix;
>      if (invvarmatrixin.length_of_mask != 0)
>      {
> -        invvarmatrixin.update_mask(rdatain.masked_data);
> +        invvarmatrixin.update_mask(reg_data.masked_data);
>      }
> -
> -    mematrix<double> X = apply_model(rdata.X, model, interaction, ngpreds,
> -            rdata.is_interaction_excluded, false, nullmodel);
> +    mematrix<double> X = apply_model(reg_data.X, model, interaction, ngpreds,
> +            reg_data.is_interaction_excluded, false, nullmodel);
>      int length_beta = X.ncol;
>      beta.reinit(length_beta, 1);
>      sebeta.reinit(length_beta, 1);
>      //Han Chen
> +
>      if (length_beta > 1)
>      {
> +
>          if (model == 0 && interaction != 0 && ngpreds == 2 && length_beta > 2)
>          {
>              covariance.reinit(length_beta - 2, 1);
> @@ -635,13 +636,14 @@
>              covariance.reinit(length_beta - 1, 1);
>          }
>      }
> +
>      //Oct 26, 2009
>      mematrix<double> W((X).nrow, 1);
>      mematrix<double> z((X).nrow, 1);
>      mematrix<double> tXWX(length_beta, length_beta);
>      mematrix<double> tXWX_i(length_beta, length_beta);
>      mematrix<double> tXWz(length_beta, 1);
> -    double prev = (rdata.Y).column_mean(0);
> +    double prev = (reg_data.Y).column_mean(0);
>      if (prev >= 1. || prev <= 0.)
>      {
>          std::cerr << "prevalence not within (0,1)\n";
> @@ -652,6 +654,7 @@
>  
>      beta.put(log(prev / (1. - prev)), 0, 0);
>      mematrix<double> tX = transpose(X);
> +
>      if (invvarmatrix.nrow != 0 && invvarmatrix.ncol != 0)
>      {
>          //TODO(maarten):invvarmatix is symmetric:is there an more effective way?
> @@ -684,12 +687,12 @@
>              double value = emu;
>              double zval;
>              value = exp(value) / (1. + exp(value));
> -            residuals[i] = (rdata.Y).get(i, 0) - value;
> +            residuals[i] = (reg_data.Y).get(i, 0) - value;
>              eMu.put(value, i, 0);
>              W.put(value * (1. - value), i, 0);
>              zval = emu
>                      + (1. / (value * (1. - value)))
> -                            * (((rdata.Y).get(i, 0)) - value);
> +                            * (((reg_data.Y).get(i, 0)) - value);
>              z.put(zval, i, 0);
>          }
>  
> @@ -746,7 +749,7 @@
>          prevlik = loglik;
>          loglik = 0.;
>          for (int i = 0; i < eMu.nrow; i++)
> -            loglik += rdata.Y[i] * eMu_us[i] - log(1. + exp(eMu_us[i]));
> +            loglik += reg_data.Y[i] * eMu_us[i] - log(1. + exp(eMu_us[i]));
>  
>          delta = fabs(1. - (prevlik / loglik));
>          niter++;
> @@ -829,9 +832,9 @@
>      // exit(1);
>  }
>  
> -void logistic_reg::score(mematrix<double>& resid, regdata& rdata, int verbose,
> +void logistic_reg::score(mematrix<double>& resid, int verbose,
>          double tol_chol, int model, int interaction, int ngpreds,
>          masked_matrix& invvarmatrix, int nullmodel) {
> -    base_score(resid, rdata, verbose, tol_chol, model, interaction, ngpreds,
> +    base_score(resid,  verbose, tol_chol, model, interaction, ngpreds,
>              invvarmatrix, nullmodel = 0);
>  }
> 
> Modified: branches/ProbABEL-0.50/src/reg1.h
> ===================================================================
> --- branches/ProbABEL-0.50/src/reg1.h	2014-02-03 22:05:56 UTC (rev 1588)
> +++ branches/ProbABEL-0.50/src/reg1.h	2014-02-03 22:29:17 UTC (rev 1589)
> @@ -49,8 +49,9 @@
>      double sigma2;
>      double loglik;
>      double chi2_score;
> +    regdata reg_data;
>  
> -    void base_score(mematrix<double>& resid, regdata& rdata, int verbose,
> +    void base_score(mematrix<double>& resid,  int verbose,
>              double tol_chol, int model, int interaction, int ngpreds,
>              const masked_matrix& invvarmatrix, int nullmodel);
>  };
> @@ -60,17 +61,18 @@
>      linear_reg(regdata& rdatain);
>      ~linear_reg()
>      {
> +        delete [] reg_data.masked_data ;
>          //		delete beta;
>          //		delete sebeta;
>          //		delete residuals;
>      }
>  
> -    void estimate(regdata& rdatain, int verbose, double tol_chol, int model,
> +    void estimate( int verbose, double tol_chol, int model,
>                    int interaction, int ngpreds,
>                    masked_matrix& invvarmatrixin,
>                    int robust, int nullmodel = 0);
>  
> -    void score(mematrix<double>& resid, regdata& rdatain, int verbose,
> +    void score(mematrix<double>& resid,  int verbose,
>                 double tol_chol, int model, int interaction, int ngpreds,
>                 const masked_matrix& invvarmatrix, int nullmodel = 0);
>  };
> @@ -82,16 +84,17 @@
>      logistic_reg(regdata& rdatain);
>      ~logistic_reg()
>      {
> +        delete [] reg_data.masked_data ;
>          //		delete beta;
>          //		delete sebeta;
>      }
>  
> -    void estimate(regdata& rdatain, int verbose, int maxiter, double eps,
> +    void estimate( int verbose, int maxiter, double eps,
>                    double tol_chol, int model, int interaction, int ngpreds,
>                    masked_matrix& invvarmatrixin, int robust,
>                    int nullmodel = 0);
>      // just a stupid copy from linear_reg
> -    void score(mematrix<double>& resid, regdata& rdata, int verbose,
> +    void score(mematrix<double>& resid,  int verbose,
>                 double tol_chol, int model, int interaction, int ngpreds,
>                 masked_matrix& invvarmatrix, int nullmodel = 0);
>  };
> 
> Modified: branches/ProbABEL-0.50/src/regdata.cpp
> ===================================================================
> --- branches/ProbABEL-0.50/src/regdata.cpp	2014-02-03 22:05:56 UTC (rev 1588)
> +++ branches/ProbABEL-0.50/src/regdata.cpp	2014-02-03 22:29:17 UTC (rev 1589)
> @@ -175,15 +175,15 @@
>      }
>  }
>  
> +//
> +//regdata::~regdata()
> +//{
> +//    //delete[] regdata::masked_data;
> +//    // delete X;
> +//    // delete Y;
> +//}
>  
> -regdata::~regdata()
> -{
> -    delete[] regdata::masked_data;
> -    // delete X;
> -    // delete Y;
> -}
>  
> -
>  regdata regdata::get_unmasked_data()
>  {
>      regdata to;  // = regdata(*this);
> 
> Modified: branches/ProbABEL-0.50/src/regdata.h
> ===================================================================
> --- branches/ProbABEL-0.50/src/regdata.h	2014-02-03 22:05:56 UTC (rev 1588)
> +++ branches/ProbABEL-0.50/src/regdata.h	2014-02-03 22:29:17 UTC (rev 1589)
> @@ -38,7 +38,7 @@
>      void update_snp(gendata *gend, const int snpnum);
>      void remove_snp_from_X();
>      regdata get_unmasked_data();
> -    ~regdata();
> +    //~regdata();
>  
>   private:
>  };
> 
> _______________________________________________
> Genabel-commits mailing list
> Genabel-commits at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
> 

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140218/f409478a/attachment-0001.sig>

From yurii.aulchenko at gmail.com  Tue Feb 18 16:43:35 2014
From: yurii.aulchenko at gmail.com (Yurii Aulchenko)
Date: Tue, 18 Feb 2014 16:43:35 +0100
Subject: [GenABEL-dev] Score option in ProbABEL
In-Reply-To: <52F7F3BC.6070203@karssen.org>
References: <52F7F3BC.6070203@karssen.org>
Message-ID: <CAHX9t6La64==5TQ9pqd31qwL-yMUvVFSR4_Q6befxjOvMFMFqg@mail.gmail.com>

I think it was "too much trouble" to support this (meaning to parse command
line correctly; to add code :)) while it is of relatively limited use (even
in theory it does not combine with all other analysis options) and
usefulness (decrease run time by a factor of ~2 - some times).

Y


On Sun, Feb 9, 2014 at 10:31 PM, L.C. Karssen <lennart at karssen.org> wrote:

> Dear list,
>
> I was wondering if one of the developers of the 'early days' knows why
> the 'score' option has been disabled since v 0.1-6 of ProbABEL (see the
> output of e.g. 'palinear -p p.phe -i i.info -d d.dose --score':
>
> probabel v. 0.4.2
> (C) Yurii Aulchenko, Lennart C. Karssen, Maksim Struchalin, EMCR
>
> Using EIGEN version 3.2.0 for matrix operations
> option --score suppressed from v 0.1-6
>
>
> If this option is no longer used we could remove its code to reduce
> clutter.
>
> Looking forward to hearing from you!
>
> Lennart.
> --
> *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
> L.C. Karssen
> Utrecht
> The Netherlands
>
> lennart at karssen.org
> http://blog.karssen.org
> GPG key ID: A88F554A
> -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
>
>
> _______________________________________________
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
>



-- 
-----------------------------------------------------
Yurii S. Aulchenko

[ LinkedIn <http://nl.linkedin.com/in/yuriiaulchenko> ] [
Twitter<http://twitter.com/YuriiAulchenko>] [
Blog <http://yurii-aulchenko.blogspot.nl/> ]
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140218/7733de42/attachment.html>

From alvaro.frank at rwth-aachen.de  Tue Feb 18 17:16:53 2014
From: alvaro.frank at rwth-aachen.de (Frank, Alvaro Jesus)
Date: Tue, 18 Feb 2014 16:16:53 +0000
Subject: [GenABEL-dev] jenkins gcc to g++
Message-ID: <244CF001646FF74FB34F372310A332C57A9C45@MBX2.rwth-ad.de>

Hi all,

I have been trying to replicate Jenkins behavior locally to avoid unnecessary numbers commit>bugfix that could be committed at once. but the project locally ahs to be compiled using g++ and not gcc. Jenkins uses gcc for the unused variables warnings.
Is there a way to perhaps change https://jenkins.genabel.org/jenkins/job/OmicABELnoMM/ to check using g++ for more consistent results?

Thank you for the help.

-Alvaro
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140218/b7acb402/attachment.html>

From lennart at karssen.org  Tue Feb 18 17:49:49 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Tue, 18 Feb 2014 17:49:49 +0100
Subject: [GenABEL-dev] jenkins gcc to g++
In-Reply-To: <244CF001646FF74FB34F372310A332C57A9C45@MBX2.rwth-ad.de>
References: <244CF001646FF74FB34F372310A332C57A9C45@MBX2.rwth-ad.de>
Message-ID: <53038F2D.8060807@karssen.org>

Hi Alvaro,

Thank you for looking into this!

On 18-02-14 17:16, Frank, Alvaro Jesus wrote:
> Hi all,
> 
> I have been trying to replicate Jenkins behavior locally to avoid
> unnecessary numbers commit>bugfix that could be committed at once. but
> the project locally ahs to be compiled using g++ and not gcc. Jenkins
> uses gcc for the unused variables warnings.

If I understand you correctly, you mean that the output of the "GNU Make
+ GNU C compiler warnings" is run by gcc instead of g++. Right?
I didn't check in detail which compiler is used by this check, but in
the Jenkins config it says that it uses the console log (see URL in the
next sentence).

If I look at the console output of the last build
(http://www.karssen.org/jenkins/job/OmicABELnoMM/18/console) I see the
same warnings pop up. For example:

g++ -DHAVE_CONFIG_H -I. -I./src   -Wall -fopenmp -g -O2 -MT
src/Algorithm.o -MD -MP -MF $depbase.Tpo -c -o src/Algorithm.o
src/Algorithm.cpp &&\
	mv -f $depbase.Tpo $depbase.Po
src/AIOwrapper.cpp: In static member function ?static void*
AIOwrapper::async_io(void*)?:
src/AIOwrapper.cpp:227:82: warning: ignoring return value of ?size_t
fread(void*, size_t, size_t, FILE*)?, declared with attribute
warn_unused_result [-Wunused-result]
                 fread (tobeFilled->buff, sizeof(type_precision),
size_buff, fp_Y);

So I'm not sure if there is a problem with the "Gnu Make + ..." check.

What the Jenkins build does is the following (Jenkins runs on an Ubuntu
13.10 machine, so look there for the exact package versions of automake,
gcc, etc.):

- Check SVN every 15 minutes for changes

- If there are changes, run the following:

- Remove files not in SVN (e.g. Makefile.in and .o files from previous
builds)

- Count lines of code (sloccount) and run cppcheck:
sloccount --duplicates --wide --details src/ | grep -vP
".svn|cppcheck-result.xml|sloccount.sc" > sloccount.sc
cppcheck --quiet --enable=all --xml src/ 2> cppcheck-result.xml

- Run cpplint.py:
#!/bin/bash -x
cpplint.py
--filter=-whitespace/tab,-whitespace/comments,-whitespace/newline,-whitespace/braces,-build/header_guard,-build/include
src/*.{cpp,h} 2> src/cpplint.log
exit 0

- Do the actual build:
autoreconf -i
if [ -f Makefile ]; then
   make distclean
fi
./configure --disable-silent-rules LDFLAGS="-L/usr/lib/openblas-base"
make -j2

- Scan for compiler warnings ("Gnu Make + ...").
- Publish cppcheck results
- Publish sloccount results
- Report violations from cpplint.log


Best,

Lennart.

> Is there a way to perhaps change
> https://jenkins.genabel.org/jenkins/job/OmicABELnoMM/ to check using g++
> for more consistent results?
> 
> Thank you for the help.
> 
> -Alvaro
> 
> 
> _______________________________________________
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
> 

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140218/42a42a39/attachment.sig>

From alvaro.frank at rwth-aachen.de  Tue Feb 18 17:56:34 2014
From: alvaro.frank at rwth-aachen.de (Frank, Alvaro Jesus)
Date: Tue, 18 Feb 2014 16:56:34 +0000
Subject: [GenABEL-dev] jenkins gcc to g++
In-Reply-To: <53038F2D.8060807@karssen.org>
References: <244CF001646FF74FB34F372310A332C57A9C45@MBX2.rwth-ad.de>,
 <53038F2D.8060807@karssen.org>
Message-ID: <244CF001646FF74FB34F372310A332C57A9C67@MBX2.rwth-ad.de>

Hi Lennart, 

thank you for the quick response, it is interesting to note that using

g++ *.cpp -L/usr/lib/openblas-base -lpthread -lopenblas -llapacke -fopenmp -Wall -pedantic -Wunused-result -Wmaybe-uninitialized

wont yield the warnings from the "-Wmaybe-uninitialized" flag, but adding the optimizations "-02" will:

g++ *.cpp -L/usr/lib/openblas-base -lpthread -lopenblas -llapacke -fopenmp -O2 -Wall -pedantic -Wunused-result -Wmaybe-uninitialized

O2 seems to trigger the warnings only when present.

-Alvaro


________________________________________
From: genabel-devel-bounces at lists.r-forge.r-project.org [genabel-devel-bounces at lists.r-forge.r-project.org] on behalf of L.C. Karssen [lennart at karssen.org]
Sent: Tuesday, February 18, 2014 5:49 PM
To: genabel-devel at lists.r-forge.r-project.org
Subject: Re: [GenABEL-dev] jenkins gcc to g++

Hi Alvaro,

Thank you for looking into this!

On 18-02-14 17:16, Frank, Alvaro Jesus wrote:
> Hi all,
>
> I have been trying to replicate Jenkins behavior locally to avoid
> unnecessary numbers commit>bugfix that could be committed at once. but
> the project locally ahs to be compiled using g++ and not gcc. Jenkins
> uses gcc for the unused variables warnings.

If I understand you correctly, you mean that the output of the "GNU Make
+ GNU C compiler warnings" is run by gcc instead of g++. Right?
I didn't check in detail which compiler is used by this check, but in
the Jenkins config it says that it uses the console log (see URL in the
next sentence).

If I look at the console output of the last build
(http://www.karssen.org/jenkins/job/OmicABELnoMM/18/console) I see the
same warnings pop up. For example:

g++ -DHAVE_CONFIG_H -I. -I./src   -Wall -fopenmp -g -O2 -MT
src/Algorithm.o -MD -MP -MF $depbase.Tpo -c -o src/Algorithm.o
src/Algorithm.cpp &&\
        mv -f $depbase.Tpo $depbase.Po
src/AIOwrapper.cpp: In static member function ?static void*
AIOwrapper::async_io(void*)?:
src/AIOwrapper.cpp:227:82: warning: ignoring return value of ?size_t
fread(void*, size_t, size_t, FILE*)?, declared with attribute
warn_unused_result [-Wunused-result]
                 fread (tobeFilled->buff, sizeof(type_precision),
size_buff, fp_Y);

So I'm not sure if there is a problem with the "Gnu Make + ..." check.

What the Jenkins build does is the following (Jenkins runs on an Ubuntu
13.10 machine, so look there for the exact package versions of automake,
gcc, etc.):

- Check SVN every 15 minutes for changes

- If there are changes, run the following:

- Remove files not in SVN (e.g. Makefile.in and .o files from previous
builds)

- Count lines of code (sloccount) and run cppcheck:
sloccount --duplicates --wide --details src/ | grep -vP
".svn|cppcheck-result.xml|sloccount.sc" > sloccount.sc
cppcheck --quiet --enable=all --xml src/ 2> cppcheck-result.xml

- Run cpplint.py:
#!/bin/bash -x
cpplint.py
--filter=-whitespace/tab,-whitespace/comments,-whitespace/newline,-whitespace/braces,-build/header_guard,-build/include
src/*.{cpp,h} 2> src/cpplint.log
exit 0

- Do the actual build:
autoreconf -i
if [ -f Makefile ]; then
   make distclean
fi
./configure --disable-silent-rules LDFLAGS="-L/usr/lib/openblas-base"
make -j2

- Scan for compiler warnings ("Gnu Make + ...").
- Publish cppcheck results
- Publish sloccount results
- Report violations from cpplint.log


Best,

Lennart.

> Is there a way to perhaps change
> https://jenkins.genabel.org/jenkins/job/OmicABELnoMM/ to check using g++
> for more consistent results?
>
> Thank you for the help.
>
> -Alvaro
>
>
> _______________________________________________
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
>

--
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-


From lennart at karssen.org  Tue Feb 18 18:08:41 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Tue, 18 Feb 2014 18:08:41 +0100
Subject: [GenABEL-dev] jenkins gcc to g++
In-Reply-To: <244CF001646FF74FB34F372310A332C57A9C67@MBX2.rwth-ad.de>
References: <244CF001646FF74FB34F372310A332C57A9C45@MBX2.rwth-ad.de>,
 <53038F2D.8060807@karssen.org>
 <244CF001646FF74FB34F372310A332C57A9C67@MBX2.rwth-ad.de>
Message-ID: <53039399.2020302@karssen.org>

Hi Alvaro,


On 18-02-14 17:56, Frank, Alvaro Jesus wrote:
> Hi Lennart, 
> 
> thank you for the quick response, it is interesting to note that using
> 
> g++ *.cpp -L/usr/lib/openblas-base -lpthread -lopenblas -llapacke -fopenmp -Wall -pedantic -Wunused-result -Wmaybe-uninitialized
> 
> wont yield the warnings from the "-Wmaybe-uninitialized" flag, but adding the optimizations "-02" will:
> 
> g++ *.cpp -L/usr/lib/openblas-base -lpthread -lopenblas -llapacke -fopenmp -O2 -Wall -pedantic -Wunused-result -Wmaybe-uninitialized
> 
> O2 seems to trigger the warnings only when present.

That's interesting! Learned something new. I guess that when using -O2
the optimisations may be such that it cannot guarantee certain
initialisations.
You probably know more about compiler optimisations, but my
understanding is that -O2 is sort of the least -O flag you should use if
you want to make use of the capabilities of current hardware.
Maybe it's worth trying to profile your code with and without the -O2
option to see what the difference in speed is.

By the way, the -O2 is added in configure.ac (if the user didn't specify
CPPFLAGS when running ./configure.


Best,

Lennart.

> 
> -Alvaro
> 
> 
> ________________________________________
> From: genabel-devel-bounces at lists.r-forge.r-project.org [genabel-devel-bounces at lists.r-forge.r-project.org] on behalf of L.C. Karssen [lennart at karssen.org]
> Sent: Tuesday, February 18, 2014 5:49 PM
> To: genabel-devel at lists.r-forge.r-project.org
> Subject: Re: [GenABEL-dev] jenkins gcc to g++
> 
> Hi Alvaro,
> 
> Thank you for looking into this!
> 
> On 18-02-14 17:16, Frank, Alvaro Jesus wrote:
>> Hi all,
>>
>> I have been trying to replicate Jenkins behavior locally to avoid
>> unnecessary numbers commit>bugfix that could be committed at once. but
>> the project locally ahs to be compiled using g++ and not gcc. Jenkins
>> uses gcc for the unused variables warnings.
> 
> If I understand you correctly, you mean that the output of the "GNU Make
> + GNU C compiler warnings" is run by gcc instead of g++. Right?
> I didn't check in detail which compiler is used by this check, but in
> the Jenkins config it says that it uses the console log (see URL in the
> next sentence).
> 
> If I look at the console output of the last build
> (http://www.karssen.org/jenkins/job/OmicABELnoMM/18/console) I see the
> same warnings pop up. For example:
> 
> g++ -DHAVE_CONFIG_H -I. -I./src   -Wall -fopenmp -g -O2 -MT
> src/Algorithm.o -MD -MP -MF $depbase.Tpo -c -o src/Algorithm.o
> src/Algorithm.cpp &&\
>         mv -f $depbase.Tpo $depbase.Po
> src/AIOwrapper.cpp: In static member function ?static void*
> AIOwrapper::async_io(void*)?:
> src/AIOwrapper.cpp:227:82: warning: ignoring return value of ?size_t
> fread(void*, size_t, size_t, FILE*)?, declared with attribute
> warn_unused_result [-Wunused-result]
>                  fread (tobeFilled->buff, sizeof(type_precision),
> size_buff, fp_Y);
> 
> So I'm not sure if there is a problem with the "Gnu Make + ..." check.
> 
> What the Jenkins build does is the following (Jenkins runs on an Ubuntu
> 13.10 machine, so look there for the exact package versions of automake,
> gcc, etc.):
> 
> - Check SVN every 15 minutes for changes
> 
> - If there are changes, run the following:
> 
> - Remove files not in SVN (e.g. Makefile.in and .o files from previous
> builds)
> 
> - Count lines of code (sloccount) and run cppcheck:
> sloccount --duplicates --wide --details src/ | grep -vP
> ".svn|cppcheck-result.xml|sloccount.sc" > sloccount.sc
> cppcheck --quiet --enable=all --xml src/ 2> cppcheck-result.xml
> 
> - Run cpplint.py:
> #!/bin/bash -x
> cpplint.py
> --filter=-whitespace/tab,-whitespace/comments,-whitespace/newline,-whitespace/braces,-build/header_guard,-build/include
> src/*.{cpp,h} 2> src/cpplint.log
> exit 0
> 
> - Do the actual build:
> autoreconf -i
> if [ -f Makefile ]; then
>    make distclean
> fi
> ./configure --disable-silent-rules LDFLAGS="-L/usr/lib/openblas-base"
> make -j2
> 
> - Scan for compiler warnings ("Gnu Make + ...").
> - Publish cppcheck results
> - Publish sloccount results
> - Report violations from cpplint.log
> 
> 
> Best,
> 
> Lennart.
> 
>> Is there a way to perhaps change
>> https://jenkins.genabel.org/jenkins/job/OmicABELnoMM/ to check using g++
>> for more consistent results?
>>
>> Thank you for the help.
>>
>> -Alvaro
>>
>>
>> _______________________________________________
>> genabel-devel mailing list
>> genabel-devel at lists.r-forge.r-project.org
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
>>
> 
> --
> *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
> L.C. Karssen
> Utrecht
> The Netherlands
> 
> lennart at karssen.org
> http://blog.karssen.org
> GPG key ID: A88F554A
> -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
> 

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140218/a1f2c1e2/attachment-0001.sig>

From lennart at karssen.org  Sun Feb 23 23:33:38 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Sun, 23 Feb 2014 23:33:38 +0100
Subject: [GenABEL-dev] Problem when loading filevector files in current
 DatABEL on CRAN (?)
Message-ID: <530A7742.3020006@karssen.org>

Dear all,

I was trying to load a filevector/DatABEL file in R today using R 3.0.2
and DatABEL 0.9-4 (from CRAN). I tried several files (e.g. the ones
included in the ProbABEL examples directory) and always got the
following error:

library(DatABEL)
DatABEL v.0.9-4 (March 12, 2013) loaded

> databel("./mach1.dose.large")
uninames$unique.names = TRUE
uninames$unique.rownames = TRUE
uninames$unique.colnames = TRUE
backingfilename = ./mach1.dose.large
cachesizeMb = 64
number of columns (variables) =  5000
number of rows (observations) =  1320
usedRowIndex: 1  2  3  4  5  ...
usedColIndex: 1  2  3  4  5  6  7  8  9  10  ...
Upper-left 10 columns and  5 rows:
file contains data of unknown type 5
Error in !.Call("assignDoubleMatrix", from at data, newi, newj, ret,
as.integer(1)) :
  invalid argument type


After I checked out the latest version from SVN, built it (using the
makedistrib_DatABEL.sh script) and installed it (I gave it a fake 0.9-5
version number) I was able to read the file:

> library(DatABEL)
DatABEL v.0.9-5 (March 12, 2013) loaded

Installed DatABEL version (0.9-5) is not the same as stable
version available from CRAN (0.9-4). Unless used intentionally,
consider updating to the latest CRAN version. For that, use
'install.packages("DatABEL")', or ask your system administrator
to update the package.


> databel("./mach1.dose.large")
uninames$unique.names = TRUE
uninames$unique.rownames = TRUE
uninames$unique.colnames = TRUE
backingfilename = ./mach1.dose.large
cachesizeMb = 64
number of columns (variables) =  5000
number of rows (observations) =  1320
usedRowIndex: 1  2  3  4  5  ...
usedColIndex: 1  2  3  4  5  6  7  8  9  10  ...
Upper-left 10 columns and  5 rows:
    rs48146833818 rs60765062039 rs61390741602 rs14182581817 rs72744993535
id1         0.472         1.369         0.001         0.555         1.955
id2         1.983         1.992         1.984         1.985         1.999
id3         1.112         1.897         0.991         1.115         1.981
id4         0.307         1.523         0.001         0.330         1.960
id5         0.519         0.988         0.003         0.533         1.727
    rs61166101522 rs130430004482 rs60542571458 rs60866164126 rs60394033150
id1             2              2         0.064         0.789         0.427
id2             2              2         1.969         1.203         0.545
id3             2              2         1.002         0.868         0.442
id4             2              2         0.048         0.568         0.308
id5             2              2         0.161         1.092         0.939
>


Any reason why this happened? As far as I traced back in SVN I can see
that both Maksim and I made some changes in the DatABEL code in the past
few months, but they don't seem to be in relevant sections of the code.
Maybe this has to do with changes we made in filevector (fvlib)...

Anyway, can somebody else verify this? If so, we should upload a new
version to CRAN ASAP.


Thanks,

Lennart.


P.S. If this is indeed related to changes in fvlib it makes the changes
to the way we use fvlib Maksim, Yurii and I proposed earlier even more
urgent (see
http://lists.r-forge.r-project.org/pipermail/genabel-devel/2014-February/000958.html).
Looking forward to any comments you may have on that proposal as well!


-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140223/84893c39/attachment.sig>

From lennart at karssen.org  Thu Feb 27 23:46:57 2014
From: lennart at karssen.org (L.C. Karssen)
Date: Thu, 27 Feb 2014 23:46:57 +0100
Subject: [GenABEL-dev] [Genabel-commits] r1625 -
	branches/ProbABEL-0.50/src
In-Reply-To: <20140227215638.BC3BA186EA1@r-forge.r-project.org>
References: <20140227215638.BC3BA186EA1@r-forge.r-project.org>
Message-ID: <530FC061.6050002@karssen.org>

Hi Maarten,

On 27-02-14 22:56, noreply at r-forge.r-project.org wrote:
> Author: maartenk
> Date: 2014-02-27 22:56:38 +0100 (Thu, 27 Feb 2014)
> New Revision: 1625
> 
> Modified:
>    branches/ProbABEL-0.50/src/reg1.cpp
> Log:
> using predefined matrices in mmscore method. Initial testing showed a 35% speedup in the regression part with(in a separate toy program to test regression)

So you're saying that using predefined Matrix sizes speeds up the
computation by roughly a third? Wow! SO if I understand it correctly,
the allocation of an fixed/known sized matrix followed by matrix
multiplication is much faster than in the case that dimensions of the
matrix are not fixed... Right?

> 
> Modified: branches/ProbABEL-0.50/src/reg1.cpp
> ===================================================================
> --- branches/ProbABEL-0.50/src/reg1.cpp	2014-02-26 21:43:42 UTC (rev 1624)
> +++ branches/ProbABEL-0.50/src/reg1.cpp	2014-02-27 21:56:38 UTC (rev 1625)
> @@ -357,6 +357,7 @@
>      double sigma2_internal;
>  
>  #if EIGEN
> +
>      LDLT <MatrixXd> Ch;
>  #else
>      mematrix<double> tXX_i;
> @@ -373,12 +374,30 @@
>          //Oct 26, 2009
>  
>  #if EIGEN
> -        // next line is  5997000 flops
> -        MatrixXd tXW = X.data.transpose() * invvarmatrixin.masked_data->data;
> -        Ch = LDLT <MatrixXd>(tXW * X.data); // 17991 flops
> -        beta.data = Ch.solve(tXW * reg_data.Y.data);//5997 flops
> -        //next line is: 1000+5000+3000= 9000 flops
> -        sigma2 = (reg_data.Y.data - tXW.transpose() * beta.data).squaredNorm();
> +        cout << "BB"<<X.data.cols()<<"AAAAAAa"<<endl;
> +        if (X.data.cols()== 3){
> +            Matrix<double,3,Dynamic> tXW = X.data.transpose()*invvarmatrixin.masked_data->data;
> +            Matrix3d xWx = tXW * X.data;
> +            Ch = LDLT <MatrixXd>  (xWx );
> +            Vector3d beta_3f = Ch.solve(tXW * reg_data.Y.data);
> +            sigma2 = (reg_data.Y.data - tXW.transpose() * beta_3f).squaredNorm();
> +            beta.data = beta_3f;
	
	The two lines above are repeated almost identically in the cases below.
Wouldn't it make more sense to simple call the Vector{3d,2d,beta.data}
in each of the cases 'betaf'. Then you could take the sigma2 calculation
and assignment to beta.data out of the ifs. Like this:
         Vector3d betaf = Ch.solve(tXW * reg_data.Y.data);
      }
	... the else if and else here...
      }
         sigma2 = (reg_data.Y.data - tXW.transpose() *
                    betaf).squaredNorm();
         beta.data = betaf;
  #else


Thanks!

Lennart.


> +        }
> +       else if(X.data.cols()== 2){
> +            Matrix<double,2,Dynamic> tXW = X.data.transpose()*invvarmatrixin.masked_data->data;
> +            Matrix2d xWx = tXW * X.data;
> +            Ch = LDLT <MatrixXd>  (xWx );
> +            Vector2d beta_2f = Ch.solve(tXW * reg_data.Y.data);
> +            sigma2 = (reg_data.Y.data - tXW.transpose() * beta_2f).squaredNorm();
> +            beta.data = beta_2f;
> +        }else{
> +            // next line is  5997000 flops
> +            MatrixXd tXW = X.data.transpose() * invvarmatrixin.masked_data->data;
> +            Ch = LDLT <MatrixXd>(tXW * X.data); // 17991 flops
> +            beta.data = Ch.solve(tXW * reg_data.Y.data);//5997 flops
> +            //next line is: 1000+5000+3000= 9000 flops
> +            sigma2 = (reg_data.Y.data - tXW.transpose() * beta.data).squaredNorm();
> +        }
>  #else
>          // next line is  5997000 flops
>          mematrix<double> tXW = transpose(X) * invvarmatrixin.masked_data;
> 
> _______________________________________________
> Genabel-commits mailing list
> Genabel-commits at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits
> 

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140227/120e565b/attachment.sig>

From shenxia911 at gmail.com  Fri Feb 28 09:39:12 2014
From: shenxia911 at gmail.com (Xia Shen)
Date: Fri, 28 Feb 2014 09:39:12 +0100
Subject: [GenABEL-dev] Suggested change of default weight in GenABEL/ibs()
Message-ID: <768CB804-C3EF-4E6A-B524-937A1B54331C@gmail.com>

Hi,

I suggest to change the default weight argument in the GenABEL/ibs() function to be set to ?freq? instead of ?no?. I have colleagues constantly forget to set to ?freq? and produce, what I would call, ?wrong? kinship matrix.

Xia Shen
PhD

Division of Computational Genetics
Department of Clinical Science
Swedish University of Agricultural Sciences
Uppsala, Sweden

www.shen.se



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140228/9830b857/attachment.html>

From yurii.aulchenko at gmail.com  Fri Feb 28 14:15:36 2014
From: yurii.aulchenko at gmail.com (Yurii Aulchenko)
Date: Fri, 28 Feb 2014 20:15:36 +0700
Subject: [GenABEL-dev] Suggested change of default weight in
	GenABEL/ibs()
In-Reply-To: <768CB804-C3EF-4E6A-B524-937A1B54331C@gmail.com>
References: <768CB804-C3EF-4E6A-B524-937A1B54331C@gmail.com>
Message-ID: <-317756408431031488@unknownmsgid>

In principle I agree this is good idea, rarely "no" (default) option is
used; but I am always worried to change defaults as this may destroy people
pipelines.

We need third opinion :)

Best,
Y

----------------------
Yurii Aulchenko
(sent from mobile device)

On Feb 28, 2014, at 3:45 PM, Xia Shen <shenxia911 at gmail.com> wrote:

Hi,

I suggest to change the default weight argument in the GenABEL/ibs()
function to be set to "freq" instead of "no". I have colleagues constantly
forget to set to "freq" and produce, what I would call, "wrong" kinship
matrix.

*Xia Shen*
PhD

Division of Computational Genetics
Department of Clinical Science
*Swedish University of Agricultural Sciences*
Uppsala, Sweden

www.shen.se



_______________________________________________
genabel-devel mailing list
genabel-devel at lists.r-forge.r-project.org
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140228/3c72aa69/attachment.html>

From maoj at mail.nih.gov  Fri Feb 28 19:51:15 2014
From: maoj at mail.nih.gov (Jean Mao)
Date: Fri, 28 Feb 2014 13:51:15 -0500
Subject: [GenABEL-dev] Error wil testing ProbABEL 0.4.2
Message-ID: <5310DAA3.4020405@mail.nih.gov>

Hi,

Thank you very much for providing this software to scientific community. 
We have installed it in our cluster for NIH scientists.

Recently, while trying to update to 0.4.2 version, I ran into error 
message when testing the installation. Attached is the log file. Any 
help will be appreciated. Thank you.

Jean Mao
Helix Staff
CIT, NIH
-------------- next part --------------
===================================================
   ProbABEL 0.4.2: checks/R-tests/test-suite.log
===================================================

# TOTAL: 3
# PASS:  1
# SKIP:  0
# XFAIL: 1
# FAIL:  1
# XPASS: 0
# ERROR: 0

.. contents:: :depth: 2

XFAIL: run_R_test_palogist.sh
=============================

Checking logistic regression...
Dose: check consistency of names		OK
Prob: check consistency of names		OK
Check consistency dose <-> prob gtdata		OK
Running ProbABEL...				OK
Comparing R output with ProbABEL output		Error: all.equal(dose.add.PA, dose.add.R, tol = tol) are not all TRUE
Execution halted

FAIL: run_R_test_pacox.sh
=========================

Checking Cox PH regression...
Loading required package: splines
Dose: check consistency of names		OK
Prob: check consistency of names		OK
Check consistency dose <-> prob gtdata		OK
Running ProbABEL...				OK
Comparing R output with ProbABEL output		Error: all.equal(dose.add.PA, dose.add.R, tol = tol) are not all TRUE
Execution halted


From kooyman at gmail.com  Fri Feb 28 22:21:28 2014
From: kooyman at gmail.com (Maarten Kooyman)
Date: Fri, 28 Feb 2014 22:21:28 +0100
Subject: [GenABEL-dev] Error wil testing ProbABEL 0.4.2
In-Reply-To: <5310DAA3.4020405@mail.nih.gov>
References: <5310DAA3.4020405@mail.nih.gov>
Message-ID: <5310FDD8.6000705@gmail.com>

Dear Jean,

Can you please send the output of ./configure and make. This helps to 
recreate the error and inspect the source of the problem.

Kind regards,

Maarten

On 28-02-14 19:51, Jean Mao wrote:
> Hi,
>
> Thank you very much for providing this software to scientific 
> community. We have installed it in our cluster for NIH scientists.
>
> Recently, while trying to update to 0.4.2 version, I ran into error 
> message when testing the installation. Attached is the log file. Any 
> help will be appreciated. Thank you.
>
> Jean Mao
> Helix Staff
> CIT, NIH
>
>
> _______________________________________________
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140228/062641e2/attachment.html>