LOGCOMBINER on the terminal

661 views
Skip to first unread message

Susana Freitas

unread,
May 7, 2016, 1:45:07 PM5/7/16
to beast-users
Dear all
I have been trying to combine the tree files of a long run and I have struggling a bit, When running the GUI, I as using the options (which I already found in this forum):
> convert numbers from scientific to decimal annotation
> resample states at lower frequency 10000
> burnin of 30%

but LogCombiner would still crash.

So I decided to test if on an external server would be better. I tried this command
java -Xms3024m -Xmx4g -cp beast.jar beast.app.tools.LogCombiner -b 30 -log species_1462381066716.trees -log species_1462381068273.trees -log species_1462392241557.trees -o spstree_LC.tree

but I still get this error:
There is insufficient memory for the Java Runtime Environment to continue.
# Native memory allocation (malloc) failed to allocate 39328 bytes for Chunk::new
Can't open file to dump replay data. Error: Cannot allocate memory

Is there any way I can run LC with more than one core? Can I add the option of resampling in the command line as well? I did not see that option in the logcombiner help, and I think that might help since the error was when LC was writing to file.
I would appreciate any insight, I really don't know what else can I do.

Thanks in advance!

Susana


Eduardo Ruiz-Sanchez

unread,
May 7, 2016, 2:00:33 PM5/7/16
to beast...@googlegroups.com
It is possible that your log files are so big...

--
You received this message because you are subscribed to the Google Groups "beast-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to beast-users...@googlegroups.com.
To post to this group, send email to beast...@googlegroups.com.
Visit this group at https://groups.google.com/group/beast-users.
For more options, visit https://groups.google.com/d/optout.



--
__________________
Eduardo Ruiz Sánchez


Mira mis fotos (pictures) en Panoramio: http://www.panoramio.com/user/470729

Andrew Rambaut

unread,
May 7, 2016, 2:42:36 PM5/7/16
to beast...@googlegroups.com
Try using the LogCombiner that is in the BEAST 1.8.3 package. I think the BEAST 2 one is a reimplementation so may have different requirements.

Best,
Andrew

Miller, Mark

unread,
Jul 2, 2016, 11:24:22 PM7/2/16
to beast...@googlegroups.com

Hi,

I am trying to expose Logcombiner in CIPRES, and I have run into an issue.

I can run some command lines; for example, these work:

 

logcombiner logfile1.txt logfile2.txt logfile3.txt logfile4.txt logfile5.txt saveme.txt

logcombiner -scale 3 logfile1.txt logfile2.txt logfile3.txt logfile4.txt logfile5.txt saveme.txt

 

However, several others fail;

logcombiner -scale 3 -burnin 1 logfile1.txt logfile2.txt logfile3.txt logfile4.txt logfile5.txt saveme.txt

logcombiner -decimal -renumber logfile1.txt logfile2.txt logfile3.txt logfile4.txt logfile5.txt saveme.txt

logcombiner_1.8.3_comet -renumber logfile1.txt logfile2.txt logfile3.txt logfile4.txt logfile5.txt saveme.txt

 

The failure message says

xception in thread "main" java.lang.ArithmeticException: / by zero

              at dr.app.tools.LogCombiner.<init>(Unknown Source)

              at dr.app.tools.LogCombiner.main(Unknown Source)

srun: error: comet-10-33: task 0: Exited with exit code 1

 

Any advice is most welcome.

Thanks

Mark

Miller, Mark

unread,
Jul 5, 2016, 12:15:37 PM7/5/16
to beast...@googlegroups.com

Hi again,

 

I studied this further and the error seems to be associated with the -renumber flag.

I get this failure not just on Linux, but on Windows distributions as well, when the flag is activated

Through a checkbox in the GUI.

I won’t expose this option in our interface, but wanted to flag the issue.

 

Best,

Andrew Rambaut

unread,
Jul 5, 2016, 1:34:41 PM7/5/16
to beast...@googlegroups.com
Thanks,

I will add it to the bug tracker.

Best,
Andrew

Andrew Rambaut

unread,
Jul 5, 2016, 1:36:40 PM7/5/16
to beast...@googlegroups.com
Thanks,

I will add it to the bug tracker.

Best,
Andrew
On 5 Jul 2016, at 17:15, Miller, Mark <mmi...@sdsc.edu> wrote:

Reply all
Reply to author
Forward
0 new messages