[Halld-offline] diagnosis and fix to photon gun failure

[Richard Jones]

Hello Jon, Naomi, [posted to mailing list so it persists after git issue is
resolved]

What you uncovered turned out to be a nasty tracking bug in the internals
of the G4 tracking library that was abandoning the tracking of neutral
particles under certain special circumstances. Although those circumstances
were relatively rare, only once per 10k events or so, if they happened to
occur on the last track in an event, then the tracking of all subsequent
events would fail because there are only neutrals coming from your particle
gun. I have submitted a new pull request that fixes this bug, and I ran a
single run over 1M events to verify it works. Here is the plot that I get.
Prior to this, I got the same results you did after event ~340,000. Now it
runs to 1M events no problem.

[image: image]
<https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F7832920%2F53695391-3de8ed80-3d89-11e9-9e69-2217f678fdd3.png&data=02%7C01%7CHallD-Offline%40jlab.org%7C74a354c1a9c94b805af208d69fdc0704%7Cb4d7ee1f4fb34f0690372b5b522042ab%7C1%7C1%7C636872164695841853&sdata=zcfWrwOu0th7%2Bhx3NO4cBXwodB09uX0wJBSAcsuqhH4%3D&reserved=0>

Please execute the pending pull request and verify that it works for you
before closing this issue. I will follow up with a detailed description of
the diagnosis and fix, in case you are interested.
-Richard
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/halld-offline/attachments/20190303/e75ee36e/attachment.html>