I have some problems with the element beryllium. I am working with the Jeff 3.2 crosssections and a conda forge environment. I try to build a model with a beryllium reflector. When I try to execute I get the following error message:
’
Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
Backtrace for this error:
Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
When I use the ENDF crosssections or a different reflector material I have no problems at all. May someone had this error before or an idea what could be the problem.
I append a simple example of a uranium sphere with a beryllium reflector.
I was able to reproduce the problem you’re running into and it helped to uncover a bug with how we are keeping track of inelastic scattering. The bug only occurs when (n,2n) reactions in an evaluation are split into levels (MT=875+). That is, an evaluator is allowed to specify separate cross sections for an (n,2n) reaction with the residual nucleus in ground state (MT=875), 1st excited state (MT=876), etc. rather than just a total (n,2n) cross section (MT=16). We actually hadn’t come across such an evaluation until you ran a simulation with Be9 from JEFF 3.2 where this does occur. The good news is that I have a fix for the problem that is now under review. The bad news is that if you want to use it, you’ll have to wait until it is merged into the main branch of OpenMC, reinstall from source, and regenerate/download an updated data library. I just generated a JEFF 3.2 library that works with the developmental branch of OpenMC and can share it, so that will save you from having to do it yourself, but you would still need to reinstall OpenMC.
A less proper way to get around this would be to “hack” the Be9 data file to make it appear as though only a total (n,2n) reaction is present. If the above steps sound too onerous, let me know and I can try to send you a hacked file that should work with your current install.
Hi Paul,
first of all thank you very much for your answer!
Is there the chance that you could share me the JEFF 3.2 library that works with the developmental branch and send me the “hacked” Be9 file?
Then I would try first the “onerous” way and if this dosen’t work, I still have the chance to do it with the “hacked” Be9 file.
You can either wait for that pull request to be merged and then clone the main repository, or you can run off my branch (where I’ve fixed the bug) directly by running something like:
Dear Paul,
I am sorry, but I think there is still (or another problem). I installed openmc and the crosssections I got from you. But now I got sometimes an error which looks like the following (this is from the debugger mode):
I was also able to reproduce this and found an indexing error that is now fixed in this branch (sorry, this is what happens when you live on the bleeding edge). If you want to get that branch right away you can run:
Otherwise, if you wait a day or two, the fix will probably be merged in the main ‘develop’ branch. Thanks again for reporting and being my unwitting beta tester
