Err: Out of memory re-packing data for sectorial searches.

[ross.mounce]

Dear all,


Occasionally I hit this error message:

"
COMMAND: XMULT
Repl. Algor. Tree Score Best Score Time
Rearrangs.
Out of memory re-packing data for sectorial
searches.
Increase slack for sectors to 6 or more (with sect:slack 6), and try
again
Error reading batchfile.tnt
"

The commands inside my batch file are of this structure:
"
proc file1.nex; xmult=rss fuse 5 drift 10 ratchet 20; macro= ; stats ;
macro- ;
proc file2.nex; xmult=rss fuse 5 drift 10 ratchet 20; macro= ; stats ;
macro- ;
...
proc file500.nex; xmult=rss fuse 5 drift 10 ratchet 20; macro= ;
stats ; macro- ;
"

Could anyone perhaps explain to me a little more about why this error
happens and how I can avoid it (within the context of a batch run,
with macros OFF)?

I've read http://tnt.insectmuseum.org/index.php/Commands/sectsch, but
i'm struggling to get the syntax right to set sect:slack 6 (or more)
within a batch mode environment...

Thanks,

Ross

[Pablo Golo]

Hi Ross,

you have to issue a command just like the error message tells you (obviously before doing the searches).

That is:

proc file1.nex; sect: slack 10 ; xmult=rss fuse 5 drift 10 ratchet 20; macro= ; stats ;
macro- ;
proc file2.nex; sect: slack 10 ; xmult=rss fuse 5 drift 10 ratchet 20; macro= ; stats ;
macro- ;
...
proc file500.nex; sect: slack 10 ; xmult=rss fuse 5 drift 10 ratchet 20; macro= ;
stats ; macro- ;

The "slack" has to do with internal memory management when creating buffers for doing the sectorial searches. These buffers are allocated when the data are packed initially (i.e. almost no memory re-allocation is done when creating the sectors), and by default TNT does not allocate any extra space, i.e., it assumes the sectors will have no more packed characters than the full data set.  The number of packed characters can however change; this is uncommon, but possible.  This could happen, for example, if in the full data set you have the packs of 2-state characters full (i.e. a number of 2-state informative characters which is an exact multiple of 16), and a number of 4-state characters, some of which become 2-state in the sector (but which leave the last package of 8 4-state characters non-empty).  Under those conditions, you may need one more package of 2-state characters, and the same number of 4-state characters.  TNT tries to squeeze those all the time when packing the data, but these differences occassionaly lead to more packed characters for the sector than there were for the full data set.  That is what the "slack" controls: the percentage of extra cells initially allocated for subsequent use by the sectorial searches.

Best regards,

--Pablo

2010/10/21 ross.mounce <ross....@gmail.com>

--
Has recibido este mensaje porque estás suscrito al grupo "TNT-Tree Analysis using New Technology" de Grupos de Google.
Para publicar una entrada en este grupo, envía un correo electrónico a tnt-tree-analysis-u...@googlegroups.com.
Para anular tu suscripción a este grupo, envía un correo electrónico a tnt-tree-analysis-using-n...@googlegroups.com
Para tener acceso a más opciones, visita el grupo en http://groups.google.com/group/tnt-tree-analysis-using-new-technology?hl=es.

[ross.mounce]

Great. :)
Thanks Pablo.

I've never heard the phrase 'packed characters' before.

Ronquist (1998) appears to provide a good explanation in case anyone
else was interested...

Ref: Ronquist, F. (1998) Fast Fitch-Parsimony Algorithms for Large
Data Sets. Cladistics. 14(4), 387-400.

> 2010/10/21 ross.mounce <ross.mou...@gmail.com>

>
>
>
> > Dear all,
>
> > Occasionally I hit this error message:
>
> > "
> > COMMAND: XMULT
> > Repl. Algor.     Tree        Score       Best Score   Time
> > Rearrangs.
> > Out of memory re-packing data for sectorial
> > searches.
> > Increase slack for sectors to 6 or more (with sect:slack 6), and try
> > again
> > Error reading batchfile.tnt
> > "
>
> > The commands inside my batch file are of this structure:
> > "
> > proc file1.nex; xmult=rss fuse 5 drift 10 ratchet 20; macro= ; stats ;
> > macro- ;
> > proc file2.nex; xmult=rss fuse 5 drift 10 ratchet 20; macro= ; stats ;
> > macro- ;
> > ...
> > proc file500.nex; xmult=rss fuse 5 drift 10 ratchet 20; macro= ;
> > stats ; macro- ;
> > "
>
> > Could anyone perhaps explain to me a little more about why this error
> > happens and how I can avoid it (within the context of a batch run,
> > with macros OFF)?
>

> > I've readhttp://tnt.insectmuseum.org/index.php/Commands/sectsch, but

> > i'm struggling to get the syntax right to set sect:slack 6 (or more)
> > within a batch mode environment...
>
> > Thanks,
>
> > Ross
>
> > --
> > Has recibido este mensaje porque estás suscrito al grupo "TNT-Tree Analysis
> > using New Technology" de Grupos de Google.
> > Para publicar una entrada en este grupo, envía un correo electrónico a
> > tnt-tree-analysis-u...@googlegroups.com.
> > Para anular tu suscripción a este grupo, envía un correo electrónico a

> > tnt-tree-analysis-using-n...@googlegroups.com<tnt-tre e-analysis-using-new-technology%2Bunsu...@googlegroups.com>

> > Para tener acceso a más opciones, visita el grupo en

> >http://groups.google.com/group/tnt-tree-analysis-using-new-technology...
> > .

[Ana Carolina Pavan]

Hi. I am having the same problem..

I can't perform any bootstrap analysis using new tech searches in TNT because it returns the following error:

"out of memory re-packing data for sectorial searches. Increase slack for sectors to 9 or more (with sect:slack 9) and try again."

I am using an windows version with menu interface, so I am not familiar with command lines.

My attempt to solve the problem was setting the memory of the program to 600Mb and the number of trees to 10.000 (settings>memory) but it didn't work.

Do you know how do I solve this without using command lines?

Thanks!