[Gate-users] "Segmentation violation" aborts my run

Chih-Chieh Liu jjliu6 at gmail.com
Thu Nov 19 12:43:35 CET 2009

Dear Ida and Marcin,
    Thanks for your experience sharing. I had similar situation with yours
and there are two possible problems:
1) Overlapping geometric components
   Please check it out after initialization and use this command,
/geometry/test/recursive_test. And then you will see the name of overlapping
components and the exact position.
2) Source outside of phantom
   Just go to the macro of phantom.mac and source.mac and check it out.

By the way, my system configuration is:
  CPU: Intel C2D E8400
  RAM: 4GB
GATE configuration is:
  GATE v5.0.0_p01
  ROOT v5.22.00
  lmf v3.0


CC Liu

2009/9/24 Ida Häggström <ida.haggstrom at radfys.umu.se>

> Dear Marcin,
> thank you so much for your very thorough reply! I will surely try and find
> out about this process 7508. As for point #2, It seems very plausible that
> that could cause problems for the system. However, I have already tried
> disabling all visualization but I always get the same errors. =( I will try
> and find some unix expert that can come have a look at my errors and maybe
> tell me the origin. For point #3, I have only simulated back to back 511keV
> photons as a first simple case. I figured I'd get that to work, then go via
> the positrons... And unfortunately I only have the one computer I'm working
> on. I will look into the geometry test errors as well (my other post
> "geomTest Error: overlapping daughters and overshooting daughter volume"),
> and see if that is connected somehow. Either way, if I manage so solve the
> problem I will post the solution! Thank you very much again, I really
> appreciate your help!
> sincerely,
> Ida
> 2009/9/24 Marcin Balcerzyk <m.balcerzyk at pluri.ucm.es>
>>  Dear Ida.
>> I am no expert in debugging Gate, but I have three suggestions:
>> 1.       Double *check what is the process 7508 doing* (number in your
>> description of the problem from 23 Sept.). If this is Gate itself, there is
>> no clue, but if any other program – it may be indicative.
>> 2.       *Disconnect visualization of gamma tracks.* I have seen it once
>> in early days of my Gate simulations, that graphic subsystem simply chokes
>> at high rate or amount of the events or when there is simply too many. If
>> you use 10Bq it does not seem as high activity at all, but… As the problem
>> appears only at longer times of simulation, it is rather not the Gate macro
>> itself. Note first sensible (to me) message of stacking the trace in your
>> debugging message: #5  0x011aaa56 in TUnixSystem::*StackTrace. *All* *else
>> seems like very deep Unix system error, distantly related to Gate.
>> 3.       For the same reason of speed, as in point 2: if you have a
>> choice, *try to run the same macro on a significantly slower system*. I
>> experienced segmentation violations when positrons did not annihilate on the
>> way from the source and did not find any material to react with and escaped.
>> On a slow system a few escapes during few tens of minutes of simulation were
>> not a problem, Gate could handle this, but on a same Gate version with the
>> same macro on a system with 4x the speed, the escaping positrons induced
>> rapid segmentation violation.
>> If anything of that works, let me know. I am just curious.
>> Best regards
>> Marcin Balcerzyk
>> Instituto Pluridisciplinar
>> Universidad Compultense de Madrid
>> Paseo Juan XXIII n. 1
>> 28040 Madrid, Spain
>> tel. +34 91 3943261
>> fax. +34 91 3943266
>> movil +34 697 322 126
>> Skype: balcerzm
> _______________________________________________
> Gate-users mailing list
> Gate-users at lists.healthgrid.org
> http://lists.healthgrid.org/mailman/listinfo/gate-users

Chih-Chieh Liu (劉志傑), PhD student

Laboratory for Preclinical Imaging and Imaging Technology of the Werner
Roentgenweg 13, 72076 Tuebingen, GERMANY
phone: +49 (7071) 29 83639
Fax:     +49 (7071) 29 5541
e-mail: Chih-Chieh.Liu at med.uni-tuebingen.de
web: www.preclinicalimaging.org

More information about the Gate-users mailing list