You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, there is no script available for processing GNDS files in the repository. Given the increasing use of GNDS in nuclear data, it would be beneficial to have a dedicated script to convert GNDS files into HDF5 libraries for OpenMC.
Proposal
I propose adding a script that:
Converts GNDS files into HDF5 format.
Integrates smoothly with existing OpenMC workflows.
Provides documentation and examples for users to understand its usage.
Benefits
Expands the versatility of OpenMC by supporting additional data formats.
Encourages more users to utilize GNDS data for their simulations.
I would be keen to see this feature. I think it might also help avoid some of the nuclear data issues we see appeaing in ACE files occasionally. Would it be more suitable in Paul's endf-python package and then we wrap it in ipenmc incident neutron class
If there is other software (e.g. Fudge) that can convert GNDS to ENDF, there's not a huge motivation for us to reinvent the wheel. If and when evaluators start putting data in evaluations that can only be represented in GNDS, then it may become a real issue but I don't think we're at that point yet.
Also, provided that NJOY adds support for GNDS, that's also something that we can leverage since we mostly take data from ACE files rather than ENDF files.
Technically GNDS supports HDF5 as a serialization format so the issue is more converting GNDS into to transport ready HDF5 schema for OpenMC.
Re: NJOY. From my discussions with the devs I don't believe GNDS support is not a priority for them, and isn't really planned. But once they support GNDS as an output is when I would switch to GNDS support, as it is more powerful than ACE.
Activity
jon-proximafusion commentedon Oct 10, 2024
I would be keen to see this feature. I think it might also help avoid some of the nuclear data issues we see appeaing in ACE files occasionally. Would it be more suitable in Paul's endf-python package and then we wrap it in ipenmc incident neutron class
paulromano commentedon Oct 10, 2024
If there is other software (e.g. Fudge) that can convert GNDS to ENDF, there's not a huge motivation for us to reinvent the wheel. If and when evaluators start putting data in evaluations that can only be represented in GNDS, then it may become a real issue but I don't think we're at that point yet.
Also, provided that NJOY adds support for GNDS, that's also something that we can leverage since we mostly take data from ACE files rather than ENDF files.
MicahGale commentedon Jan 28, 2025
I'm late to the game. My thoughts:
Technically GNDS supports HDF5 as a serialization format so the issue is more converting GNDS into to transport ready HDF5 schema for OpenMC.