LISTSERV 16.5 - HPS-SOFTWARE Archives

I only know of a flaw in the CLHEP::RanecuEngine, which is used by GEANT4. The flaw is that for very different seeds the engine will give the exact same sequence. The periodicity of the engine (i.e. starting with one seed, how long before the sequence repeats) is huge, as it is for all the CLHEP random engines. So on a single run, with a single seed, you should NOT see any repeat or pattern in this engine. I alerted Lynn Garren of this flaw back in 2010 and they were going to look into it, but never heard anything since. Best to avoid that particular engine.

There are many other things that can cause a pattern like the one you are seeing. One possibility is that there is some subtle effect in roundoff. Are the cosmics coming from a single point good distance from the ECAL, and could it thus be numerical limits on the accuracy of the angles? Something like that?

On May 31, 2014, at 9:45 AM, Holly Vance <[log in to unmask]> wrote:

Hi,

When I am running my own events in SLIC from a macro, for small numbers of events (3k) the distribution looks ok. However, when I run something a bit larger like 100k events, I keep getting strange checkered patterns. I've been setting a /random/seed number for different runs, but this seems to have no effect (using this vs not using this and setting it at different values).

I am attaching a photo of what the cosmic run looks like. While the underlying events look good, there is a very distinct pattern that I need to get rid of. I also see this same pattern if I shoot electrons at the ECal. Any suggestions would be very useful.

I am wondering if GEANT4 is using the TRandom random number generator? I know there are errors with this. I am not sure what is being used with the SLIC interface.

--

Respectfully,
Holly

Use REPLY-ALL to reply to list
To unsubscribe from the HPS-SOFTWARE list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=HPS-SOFTWARE&A=1
<cosmics.png><electrons.png>