[FLR-list] ssb constraint with FLash

Iago MOSQUEIRA (JRC) iago.mosqueira at jrc.ec.europa.eu
Wed Oct 29 17:08:54 CET 2014 

Hi,

The issue has been fixed and we are running some extra tests before 
uploading a new version of Flash with this fix.


Iago

On 10/27/2014 03:54 PM, Iago MOSQUEIRA (JRC) wrote:
>
> Hi,
>
> We might have found the issue but we need to check it a bit more with a
> couple of different stocks.
>
>
> Iago
>
> On 27/10/14 15:53, Clara Ulrich wrote:
>> Hi Laurie,
>>
>> In your example below, you use ssb as a target, not as a minimum. This was working also for me in the previous example, but that is not really what we want? F should be the target, but SSB is just a minimum threshold, exactly like a TAC constraint?
>>
>> Clara
>>
>> From: flr-list-bounces at lists.r-forge.r-project.org [mailto:flr-list-bounces at lists.r-forge.r-project.org] On Behalf Of laurie
>> Sent: 27 October 2014 15:02
>> To: flr-list at lists.r-forge.r-project.org
>> Subject: Re: [FLR-list] ssb constraint with FLash
>>
>> There are a couple of issues neither of which should cause a problem but may point to a bug somewhere.
>> The ple4 object differs from the one Clara is using as it start at age 0 not age 1, plus it doesnt come from XSA so the F in the plusgroup is not the same as in plusgrpoup-1.
>> It looks like there is a bug in "[" i.e.
>>
>> range(ple4[-1])
>>        min       max plusgroup   minyear   maxyear   minfbar   maxfbar
>>          2        10        -1      1957      2008         2         6
>>
>>
>> The following code works as expected, it is a long winded way of constraining SSB. It points in a problem not in the estimation procedure (LU decomposition) but maybe how it is being set up.
>>
>>
>> plot(window(stk,start=2006))
>>
>> stk=window(iter(MyObj,1),end=2008)
>> stk =fwdWindow(stk,eql,end=2010)
>>
>> target <-fwdControl(data.frame(year  =c(2009,2009),
>>                               val     =c(.3,180000),
>>                               #min     =c(NA,180000),
>>                               quantity=c("f","ssb")))
>>
>> stk.=fwd(stk,ctrl=target ,
>>               sr=eql)
>>
>> fbar(stk.)[,ac(2009)]
>> An object of class "FLQuant"
>> , , unit = unique, season = all, area = unique
>>
>>       year
>> age   2009
>>    all 0.21768
>>
>> units:  f
>>> ssb( stk.)[,ac(2009:2010)]
>> An object of class "FLQuant"
>> , , unit = unique, season = all, area = unique
>>
>>       year
>> age   2009   2010
>>    all 227124 180000
>>
>> units:  NA
>>
>>
>> Note that the SSS constraint refers to the SSB at the end of the year if harvest.spwn.
>>
>> Laurie
>> On 27/10/14 10:36, Clara Ulrich wrote:
>>
>> Hej, Thanks for help. I send my object - a stf from North Sea Haddock with 10 iterations.
>>
>>
>>
>> Clara
>>
>>
>>
>>
>>
>>
>>
>> -----Original Message-----
>>
>> From: flr-list-bounces at lists.r-forge.r-project.org<mailto:flr-list-bounces at lists.r-forge.r-project.org> [mailto:flr-list-bounces at lists.r-forge.r-project.org] On Behalf Of laurie
>>
>> Sent: 27 October 2014 09:47
>>
>> To: flr-list at flr-project.org<mailto:flr-list at flr-project.org>
>>
>> Subject: Re: [FLR-list] ssb constraint with FLash
>>
>>
>>
>> Is it a bug or poor documentation?
>>
>>
>>
>> We used to have  a page on TYFLR with all the different HCRs implemented using FLash. This included the STECF ones with numerous sequential constraints, on F, TAC, SSB etc. I have also implemented some simpler interfaces for fwd. I am currently redoing this see http://rscloud.iccat.int/Tutorials/MSE/hcr.pdf
>>
>>
>>
>> As well as being documentation this will also provide a set of tests to make sure no bugs creep in during the winter.
>>
>>
>>
>> As Iago says could you provide an example, and if it is not already included in hcr.pdf I will add it. After checking there isnt a bug of course.
>>
>>
>>
>> Laurie
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> On 27/10/14 09:28, Iago MOSQUEIRA (JRC) wrote:
>>
>> Hi,
>>
>>
>>
>> I am trying to replicate the issue, using our dear ple4, but I failed
>>
>> to do so. If I set a min value for ssb in 2010, which is higher than
>>
>> what
>>
>> f=0.8 in 2009 would give me, I get as expected a lower F and the right
>>
>> ssb, try code below.
>>
>>
>>
>> Can you send me your FLStock object?
>>
>>
>>
>> Cheers,
>>
>>
>>
>>
>>
>> Iago
>>
>>
>>
>>
>>
>> library(FLash)
>>
>> library(FLAssess)
>>
>>
>>
>> data(ple4)
>>
>>
>>
>> ple4 <- stf(ple4,3)
>>
>>
>>
>> maxF <-fwdControl(data.frame(year=2009,val=0.8,quantity="f")) #  an
>>
>> example with high F that decreases biomass a lot
>>
>>
>>
>> ple4 <-
>>
>> fwd(ple4,ctrl=maxF,sr=list(model="mean",params=FLPar(1)),sr.residuals=
>>
>> FLQuant(3500000,dimnames=list(year=2009:2011)))
>>
>>
>>
>> fbar(ple4)[,"2009"]
>>
>>
>>
>> maxF <-fwdControl(data.frame(year=2009,val=c(0.8,NA),min=c(NA,350000),
>>
>> quantity=c("f","ssb")))
>>
>>
>>
>> ple4 <-
>>
>> fwd(ple4,ctrl=maxF,sr=list(model="mean",params=FLPar(1)),sr.residuals=
>>
>> FLQuant(3500000,dimnames=list(year=2009:2011)))
>>
>>
>>
>> fbar(ple4)[,"2009"]
>>
>> ssb(ple4)[,"2009"]
>>
>> ssb(ple4)[,"2010"]
>>
>>
>>
>>
>>
>> On 25/10/14 21:35, Clara Ulrich wrote:
>>
>> Hi all,
>>
>>
>>
>> We are just back from ICES MIXFISH, where we once again made much use
>>
>> of FLash power... (still on R 2.15.3 windows 32 bits, we had some
>>
>> troubles at the advice meeting in May with using the new FLFleet
>>
>> package for R3.1, we haven't tried again - Flash 2.5.0)
>>
>>
>>
>> This time we wanted to some SSB-based HCR - Typically, we have a target F with some SSB minimum constraints - for example the biomass must stay above Bpa, or if the biomass is below Bpa, the target is reduced such as SSB should be at least increasing...
>>
>>
>>
>> But I kept having troubles making it to work - I had tried this some years ago and it worked OK as far as I remember, but last week I did not manage to get around the problem.
>>
>>
>>
>> For example (with target F for 2014, and SSB constraint applying on
>>
>> SSB at 1st January 2015 at the end of the fwd year)
>>
>>
>>
>> maxF <-fwdControl(data.frame(year=2014,val=0.8,quantity="f")) #  an
>>
>> example with high F that decreases biomass a lot Proj <-
>>
>> fwd(MyObj,ctrl=maxF,sr=list(model="mean",params=FLPar(1)),sr.residual
>>
>> s=FLQuant(3500000,dimnames=list(year=2014:2015)))
>>
>>
>>
>> This works :
>>
>> fbar(Proj)[,"2014"]
>>
>>       An object of class "FLQuant"
>>
>>       iters:  10
>>
>>       , , unit = unique, season = all, area = unique
>>
>>            year
>>
>>       age   2014
>>
>>         all 0.8(0)
>>
>>       units:  f
>>
>>
>>
>> ssb(Proj)[,ac(2014:2015)]
>>
>>       An object of class "FLQuant"
>>
>>       iters:  10
>>
>>       , , unit = unique, season = all, area = unique
>>
>>            year
>>
>>       age   2014      2015
>>
>>       all 172983(0)  77323(0)
>>
>>       units:  NA
>>
>>
>>
>> but this does not work (here an example where the SSB constraint
>>
>> should be activated)
>>
>>
>>
>> maxF
>>
>> <-fwdControl(data.frame(year=2014,val=c(0.8,NA),min=c(NA,150000),
>>
>> quantity=c("f","ssb"))) Proj <-
>>
>> fwd(MyObj,ctrl=maxF,sr=list(model="mean",params=FLPar(1)),sr.residual
>>
>> s=FLQuant(3500000,dimnames=list(year=2014:2015)))
>>
>>
>>
>> fbar(Proj)[,"2014"]
>>
>>       An object of class "FLQuant"
>>
>>       iters:  10
>>
>>       , , unit = unique, season = all, area = unique
>>
>>            year
>>
>>       age   2014
>>
>>         all 0.016914(0)
>>
>>       units:  f
>>
>>
>>
>> ssb(Proj)[,ac(2014:2015)]
>>
>>       An object of class "FLQuant"
>>
>>       iters:  10
>>
>>       , , unit = unique, season = all, area = unique
>>
>>            year
>>
>>       age   2014      2015
>>
>>       all 172983(0) 172983(0)
>>
>>       units:  NA
>>
>>
>>
>>
>>
>> and this doesn't work either when the SSB is not constraining (e.g.
>>
>> SSB limit so low that it is always achieved). Only if I specify ssb
>>
>> as a value inn fwdControl does it compute correctly, but that is not
>>
>> the idea, it should be a minimum
>>
>>
>>
>>
>>
>> I have tried several ways around this but cannot find out... Any idea of what is getting wrong?
>>
>>
>>
>> Thanks in advance,
>>
>> Cheers
>>
>> Clara
>>
>> _______________________________________________
>>
>> flr-list mailing list
>>
>> flr-list at flr-project.org<mailto:flr-list at flr-project.org>
>>
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/flr-list
>>
>>
>>
>>
>>
>>
>>
>> _______________________________________________
>>
>> flr-list mailing list
>>
>> flr-list at flr-project.org<mailto:flr-list at flr-project.org>
>>
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/flr-list
>>
>>
>>
>>
>> _______________________________________________
>>
>> flr-list mailing list
>>
>> flr-list at flr-project.org<mailto:flr-list at flr-project.org>
>>
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/flr-list
>>
>>
>>
>>
>> _______________________________________________
>> flr-list mailing list
>> flr-list at flr-project.org
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/flr-list
>>
>
>

More information about the flr-list mailing list