[Traminer-users] pb of memory to compute distance matrix

[Alexis Gabadinho]

Hi Pierre,

I think this is a lack of memory, a 20000x20000 distance matrix requires 
a huge amount of memory.  You can try the "full.matrix=FALSE" option so 
that only half of the -symmetric- distance matrix is returned by the 
seqdist function. Remember also that having other big objects in your 
current R environment, like other distance matrices - uses a part of the 
available memory.

Regarding weights, these are defined once when creating a sequence 
object with seqdef, and used automatically by seqdplot and other plot 
and sequence analysis functions. Please see our freely available article 
in the journal of statistical software and help(seqdef) for more details.

All the best,
Alexis


Le 11. 02. 13 13:14, Pierre Blavier a écrit :
>
> Hi TramineR-users,
>
> 2 questions :
>
>
> 1) pb of memory computing distance matrix
>
> Everything works perfectly (couts<- seqsubm(seq.seq, 
> method="CONSTANT", cval=2) / range(couts) / round(couts, 2) )
>
> Until :
>
> seq.om <http://seq.om><-seqdist(seq.seq, method="OM",indel=1.1, sm=couts)
>
> I obtain :
>
> Erreur : impossible d'allouer un vecteur de taille 1.5 Go
>
> R(351,0xac7312c0) malloc: *** mmap(size=1617408000) failed (error code=12)
>
> *** error: can't allocate region
>
> *** set a breakpoint in malloc_error_break to debug
>
> R(351,0xac7312c0) malloc: *** mmap(size=1617408000) failed (error code=12)
>
> *** error: can't allocate region
>
> *** set a breakpoint in malloc_error_break to debug
>
> My dataset is rather large (43 periods, n=20 000, 6 states) but it is 
> rather suprising, how can I solve the problem ? ? (I have a brand-new 
> macintosh with 8Go of live memory, so I think this is not the pb).
>
> 2) How can we use the weights in TramineR, for example in order to 
> make it take them into account in commands like
>
> > seqdplot(seq.seq, withlegend=T, title="1988", weights="pond")
>
> It does not seem to work...
>
> Thanks a lot and best regards,
>
> Pierre
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> _______________________________________________
> Traminer-users mailing list
> Traminer-users at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20130211/26109b4e/attachment.html>