Approximate multigroup velocity

[GuySten]

Description

Currently, void regions in multigroup mode makes openmc crash (#3723).

This PR implements the following changes:

1. When inverse-velocity data is missing default to an approximate formula (group collapse of inverse-velocity data with 1/E spectrum).
2. Ensure that lowest neutron energy boundary is at least 1e-5 (ensure 1/E spectra is defined)

Checklist

• I have performed a self-review of my own code
• I have run clang-format (version 15) on any C++ source files (if applicable)
• I have followed the style guidelines for Python source files (if applicable)
• I have made corresponding changes to the documentation (if applicable)
• I have added tests that prove my fix is effective or that my feature works (if applicable)

[paulromano]

@nelsonag would love to get your thoughts on this one!

[nelsonag]

Why not tally a 1/v based on the group midpoint energy (I assume logarithmic midpoint is best) if no specific data is present? This could be done void or not.
I agree having data is better than not, but can assume this would work great for a UFG collapse sort of approach.

[GuySten]

Do you mean instead of defaulting to speed 0 when 1/v data is missing using velocity that match the logarithmic energy midpoint of the group?

[nelsonag]

Yep!

…

On Fri, Feb 6, 2026, 3:10 PM GuySten ***@***.***> wrote: *GuySten* left a comment (openmc-dev/openmc#3766) <#3766 (comment)> Do you mean instead of defaulting to speed 0 when 1/v data is missing using velocity that match the logarithmic energy midpoint of the group? — Reply to this email directly, view it on GitHub <#3766 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAH5GM276B2D4K5FJGJCUMT4KT7NLAVCNFSM6AAAAACT4QSIHSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTQNRSGU4DENBQGA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[GuySten]

@nelsonag, can you look at this PR again?
I've implemented your suggestions.

[meltawila]

I’m very interested in this to support use cases in Cardinal. Is there any testing/support I can provide to help see this PR merged?

[GuySten]

Currently this PR waits for a review of either @nelsonag or @paulromano.
Maybe one of them can answer your question?

[nelsonag]

Two highly-related comments: why not generalize for any MGXS without 1/v data present?

[GuySten]

@nelsonag, I've implemented your suggestions and now the code is alot simpler.
I decided to put searching for mgxs data named void for another PR (maybe we can do without)

Now the approximate inverse velocity data is stored only once.