In my endeavors I think I’ve come across a bug that may be specific to the Windows only installations but I’m not totally confident in reproducing it yet without some help.
Basically, my egs_chamber inputs crash on Windows when only IPSS is enabled. I don’t find this issue running the exact same input file on a Linux parallel computing cluster. The Windows input files will run successfully if CS is enabled through the subgeometries and subgeom regions blocks but will crash if those are commented out or absent.
In a somewhat related issue, I was wondering if it’s possible to run a simulation with both correlated and uncorrelated geometries successfully. I want to make use of the statistical accuracy sought feature for my most challenging geometry (a small volume sphere of phantom) but this geometry is not part of the correlated set. So in essence I want the following scoring objects:
1. PHSP Scoring for IPSS
2. Small Volume
3. Full Chamber with correlated subgeoms -> No Stem -> No Cel -> No Wall -> Repl
When I try to run my inputs this way, the Small Volume geometry always gets zeroed out even though only the PHSP scoring object says this MUST be 0. If I use IPSS only (on the Linux cluster) or disable IPSS altogether, then it appears to work as intended.
So in short, I think that 1. My Windows installation of EGS at least will crash if only IPSS is enabled and 2. It would be neat (as someone previously mentioned on this board) if statistical accuracy sought could follow a user-specified scoring geometry and not just the first one