Approximate multigroup velocity #3766
Conversation
paulromano
left a comment
paulromano
left a comment
There was a problem hiding this comment.
@nelsonag would love to get your thoughts on this one!
|
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. |
|
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? |
|
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:
***@***.***>
|
This reverts commit f7b0fd4.
|
Now this PR is ready again. |
|
Looking good! One more thing, sorry I should have realized sooner. We should update the docs to explain what happens when inverse-velocity is not provided in an Mgxs.h5 |
|
@nelsonag, I've added a section in the theory manual. |
Minor editorial correction to cross_sections.rst
GuySten
added
the
Merging Soon
| .. math:: | ||
|
|
||
| v_g = \frac{1}{c \log\left(\frac{E_{\text{max}}^g}{E_{\text{min}}^g}\right)} | ||
| \left[ 2(\arctan(k_\text{max}) - \arctan(k_\text{min})) + (k_\text{max}-k_\text{min})) \right] |
There was a problem hiding this comment.
I couldn't get this solution independently. As best as I can tell, the solution should use arctanh(1/k) instead of arctan(k). Here's the solution from ChatGPT 5.4 Thinking. Also, it's confusing to me to use k as a normalization constant above and then use k in a change or variables too.
There was a problem hiding this comment.
You are right, the solution should use arctanh instead of arctan.
You can find the solution here:
https://www.wolframalpha.com/input?i=integrate+%28x%2Ba%29%2F%28x*sqrt%28x*%28x%2B2*a%29%29%29+
I will update the PR.
There was a problem hiding this comment.
I've updated the formula and documentation.
I also changed the normalization constant to alpha.
|
@paulromano, do you agree with my changes? |
|
@paulromano @GuySten, Looking forward to seeing this get merged for my use case. Just wanted to make sure it's still on your radar |
|
I think this PR is ready to be merged. I am waiting for @paulromano to say if he agrees. |
paulromano
left a comment
paulromano
left a comment
There was a problem hiding this comment.
Ok, I just had a chance to go through again. I made one update to avoid complex math by using the following relationship that holds for k > 1:
Re[arctanh(k)] = arctanh(1/k)
The equation in the documentation also needed a few fixes but is good to go now. Thanks @GuySten!
4 participants
Description
Currently, void regions in multigroup mode makes openmc crash (#3723).
This PR implements the following changes:
Checklist