[Gate-users] Monte Carlo seed

[Albert Hirtl]

Dear Colleagues,

Some time ago I posted a problem concerning artefacts in the SPECT 
simulation output. It was solved by changing from "Ranlux64" to 
"MersenneTwister". In principal everything is fine now. However, 
sometimes some splits of the simulation on a Condor-grid seem to get 
lost. Meaning, they run much longer than all other splits. Until now we 
solve that by halting the respective simulation, change the seed 
manually and restart it again. Then, usually, the simulation finishes in 
a reasonable time. Did anyone observe that behaviour as well? Any ideas 
what might be the reason, or how that can be solved? It is only a minor 
problem, but if you have larger simulations with many splits it can 
become somewhat cumbersome to keep track on lost jobs.

The interesting part of our mac-file is:
---
/gate/random/setEngineName MersenneTwister
/gate/random/setEngineSeed auto
/gate/random/verbose 1
---
The simulation is simply a water filled tank with 50 MBq Tc-99m solution 
with 10 min of DAQ.

Any help would be greatly appreciated!

I wish you nice holidays,
Albert

-- 
Medizinische Universität Wien - Allgemeines Krankenhaus Wien
Universitätsklinik für Nuklearmedizin
Währinger Gürtel 18-20
1090 Wien
Austria

Tel.: +43 (0) 1 / 40400 5545

Web:
http://www.meduniwien.ac.at/user/albert.hirtl
http://www.meduniwien.ac.at/moccamed/

-------------- next part --------------
A non-text attachment was scrubbed...
Name: albert_hirtl.vcf
Type: text/x-vcard
Size: 364 bytes
Desc: not available
URL: <http://lists.opengatecollaboration.org/mailman/private/gate-users/attachments/20111223/e45de12f/attachment.vcf>