mailRe: Segmentation Fault on Relax - 1.3.7


Others Months | Index by Date | Thread Index
>>   [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Header


Content

Posted by Edward d'Auvergne on January 27, 2011 - 17:48:
Hi,

Thanks for testing that.  The optimisation failures are not failures
of relax, but that it is a new platform where optimisation is a tiny
bit different.  I'm collecting all the information about these, and
adding exceptions for each machine.  I might have to turn this off
though.  You'll see that the optimised pararameters will be
nevertheless very close to those in the test name
'test_opt_constr_bfgs_mt_S2_0_970_te_2048_Rex_0_149'.  If you could
email more of the information from the system test, this would be
quite useful to make this test pass.  There should be a long print out
like:

--------------------------------------
System: Darwin
Release: 9.8.0
Version: Darwin Kernel Version 9.8.0: Wed Jul 15 16:55:01 PDT 2009;
root:xnu-1228.15.4~1/RELEASE_I386
Win32 version:
Distribution:
Architecture: 64bit
Machine: i386
Processor: i386
Python version: 2.6.2
numpy version: 1.3.0

s2: 0.9699999999999785
te: 2047.9999999962433
rex: 0.14900000000039709
chi2: 5.2479491342506911e-24
iter: 162
f_count: 758
g_count: 169
h_count: 0
warning: None
--------------------------------------

before the traceback message you posted.  This is the information
needed.  But everything should be fine, just that you won't be able to
run relaxation curve-fitting for the R1 and R2 until the module is
compiled.  The python version I compiled on (2.6.5) is similar enough
to your 2.6.2 - this is not the issue.  Maybe the numpy version is the
problem - I compiled on 1.4.1 and you are using 1.2.1.  They could
have introduced an incompatibility.  If you can install the numpy
headers package and scons, you'll be able to compile without problem
by typing 'scons'.

Cheers,

Edward




On 27 January 2011 12:14, Tiago Pais <tpais@xxxxxxxxxxx> wrote:
Thank you Edward,
Renaming the relax_fit.so did the job! Nice catch!

However, when running the test suite it is falling in the System/Functional
tests:

Traceback (most recent call last):
 File "/usr/local/relax-1.3.7/test_suite/system_tests/model_free.py", line
677, in test_opt_constr_bfgs_mt_S2_0_970_te_2048_Rex_0_149
   self.value_test(spin, select, s2, te, rex, chi2, iter, f_count, g_count,
h_count, warning)
 File "/usr/local/relax-1.3.7/test_suite/system_tests/model_free.py", line
2051, in value_test
   self.assert_(spin.f_count in f_count, msg=mesg)
AssertionError: Optimisation failure.


And this is the output information from relax --info (v1.3.8)

Hardware information:
   Machine:                 i686
   Processor:

System information:
   System:                  Linux
   Release:                 2.6.28-17-generic
   Version:                 #58-Ubuntu SMP Tue Dec 1 18:57:07 UTC 2009
   GNU/Linux version:       Ubuntu 9.04 jaunty
   Distribution:            Ubuntu 9.04 jaunty
   Full platform string:
Linux-2.6.28-17-generic-i686-with-Ubuntu-9.04-jaunty

Software information:
   Architecture:            32bit ELF
   Python version:          2.6.2
   Python branch:           branches/release26-maint
   Python build:            release26-maint, Apr 19 2009 01:56:41
   Python compiler:         GCC 4.3.3
   Python implementation:   CPython
   Python revision:
   Numpy version:           1.2.1
   Libc version:            glibc 2.4

Python packages (most are optional):

Package              Installed       Version         Path
minfx                True            Unknown
/usr/local/relax-1.3.8/minfx
bmrblib              False
numpy                True            1.2.1
/usr/lib/python2.6/dist-packages/numpy
ScientificPython     True            2.4.11
/usr/lib/python2.6/dist-packages/Scientific
wxPython             False
mpi4py               False
epydoc               False
optparse             True            1.5.3
/usr/lib/python2.6/optparse.pyc
Numeric              True            24.2
/usr/lib/python2.6/dist-packages/Numeric/Numeric.pyc
readline             True
/usr/lib/python2.6/lib-dynload/readline.so
profile              True
/usr/lib/python2.6/profile.pyc
bz2                  True
/usr/lib/python2.6/lib-dynload/bz2.so
gzip                 True
/usr/lib/python2.6/gzip.pyc
os.devnull           True
/usr/lib/python2.6/os.pyc

Compiled relax C modules:
   Relaxation curve fitting: False

I have not tried the "scons" possibility that you suggested.

Regards,
TiagoP



-----Original Message-----
From: edward.dauvergne@xxxxxxxxx [mailto:edward.dauvergne@xxxxxxxxx] On
Behalf Of Edward d'Auvergne
Sent: quarta-feira, 26 de Janeiro de 2011 19:56
To: Tiago Pais
Cc: relax-users@xxxxxxx
Subject: Re: Segmentation Fault on Relax - 1.3.7

Hi,

This is strange!  Do the source code versions work ok?  It could be
the a strange combination of python and numpy, together with relax's C
modules.  You could possibly remove (or rename) the
maths_fns/relax_fit.so file and try again.  It would be useful to have
information about the system to work out what is happening.  You can
get all the relevant info by typing:

$ relax --info

This is assuming relax will run at all.  Another option would be if
you have 'scons', 'numpy', and 'numpy-devel' (or something like that)
installed.  You could then type:

$ scons clean_all

to remove the compiled C module, and

$ scons

to build it again.  I hope this helps!

Cheers,

Edward



On 26 January 2011 19:35, Tiago Pais <tpais@xxxxxxxxxxx> wrote:
Dear all,
Surely this should be a very silly mistake of mine but I can not manage to
put to work any of the relax versions above 1.3.6.
I just get the message "Segmentation fault" when I type "relax"

Do I need to uninstall the previous 1.3.4 version?
For now I would like to keep the old one since I have things analyzed on
that one.

Any help would be much appreciated.
Cheers
Tiago P



__________ Information from ESET NOD32 Antivirus, version of virus
signature
database 5821 (20110126) __________

The message was checked by ESET NOD32 Antivirus.

http://www.eset.com



_______________________________________________
relax (http://nmr-relax.com)

This is the relax-users mailing list
relax-users@xxxxxxx

To unsubscribe from this list, get a password
reminder, or change your subscription options,
visit the list information page at
https://mail.gna.org/listinfo/relax-users


__________ Information from ESET NOD32 Antivirus, version of virus signature
database 5822 (20110126) __________

The message was checked by ESET NOD32 Antivirus.

http://www.eset.com




__________ Information from ESET NOD32 Antivirus, version of virus signature
database 5823 (20110127) __________

The message was checked by ESET NOD32 Antivirus.

http://www.eset.com






Related Messages


Powered by MHonArc, Updated Fri Jan 28 11:40:17 2011