Different result with fixed random seed

[anonymous]

Dear SHERPA authors,

In GENSER installation and testing infrastructure we have introduced quite simple SHERPA tests.

For example we use the following configuration file:

​https://svnweb.cern.ch/trac/lcgsoft/browser/trunk/lcgcmake/generators/sherpa/tests/LHC_ZJets_riv2.dat

Running with sherpa 2.1.0 this configuration file in two separate directories(two independent runs with same initial random seed) I am getting different results:

< |  Total XS is 1977.99 pb +- ( 19.2023 pb = 0.97 % )  |
---
 > |  Total XS is 1976.93 pb +- ( 19.4595 pb = 0.98 % )  |
379c381
<     From "Jet_Evolution:CSS": 1369 (12370) -> 11 %
---
 >     From "Jet_Evolution:CSS": 1305 (12338) -> 10.5 %
382c384
<     From 
"Hadron_Decay_Handler::RejectExclusiveChannelsFromFragmentation": 334 
(0) -> 334.
---
 >     From 
"Hadron_Decay_Handler::RejectExclusiveChannelsFromFragmentation": 354 
(0) -> 354.
385c387
<     From "Decay_Channel::GenerateKinematics": 38 (544443) -> 0 %
---
 >     From "Decay_Channel::GenerateKinematics": 44 (546769) -> 0 %

I understand that running these configuration files the sherpa performs ME generation, integration and then generation of 1000 events.

But when I run SHERPA once again with pregenerated ME and integrated cross-section I get same result every run (and it is good). Therefore it means that integration is not fixed by RANDOM seed.

I would like to know if it is expected behavior or not. Could you please clarify?

Thank you in advance,

Dmitri

[Stefan Hoeche]

Hi Dmitri,

I cannot reproduce this behaviour. What are the exact commands you used? What was the Sherpa configuration command line?

Thanks Stefan

[Stefan Hoeche]

Please reopen ticket if necessary