Hi @JaGeo ,

I had a look at the phonopy source code changes between v2.28 and v.2.27, I could not really pinpoint what is causing the issue with different values of the average Gruneisen parameter here.

I guess it has something to do with following PR. But am not sure about it

phonopy/phonopy@c013a19

How large are the changes? Could you give more details? Thanks!

The average Gruneisen value now comes out to be 1.271535057337552 using v2.28 (using v2.27 we used to get 1.1882292157682082)

Yeah, I think we need a bit more analysis here. This is quite a large change

The latest version of phonopy causes phonon workflow test to also fail (cause of failure seems to originate from force constants computation default being changed)

I would need more details to act. E.g., How large is the change? Why does this only affect the Grüneisen parameter workflow? What is different?

We can do this privately if you prefer.

I would need more details to act. E.g., How large is the change? Why does this only affect the Grüneisen parameter workflow? What is different?

We can do this privately if you prefer.

As I have mentioned above, now with latest version phonon workflow test is also affected. I will try to share the details in next days

@naik-aakash i see! Thanks!

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 76.63%. Comparing base (4244da9) to head (abbc041).
Report is 2 commits behind head on main.

@@            Coverage Diff             @@
##             main    #1006      +/-   ##
==========================================
+ Coverage   72.82%   76.63%   +3.81%     
==========================================
  Files         187      187              
  Lines       13637    13627      -10     
  Branches     1370     1372       +2     
==========================================
+ Hits         9931    10443     +512     
+ Misses       3161     2626     -535     
- Partials      545      558      +13     

@naik-aakash one last question: do you need all test files still or can one set be removed?

@naik-aakash one last question: do you need all test files still or can one set be removed?

Hi @JaGeo, Thanks for spotting this. I Have now removed the test files that are no longer used

@naik-aakash thanks. This looks good. I will merge.

For everyone else seeing this and wondering why we are changing tests:

By accident, one of the phonon tests used wrong test files. This only appeared as a problem when phonopy added more strict handling of the forces.

The Grüneisen parameter computation was also affected by a phonopy change. However, as this mostly affected an unrealiable value at gamma, we simply remove it from the mesh computation. We should have likely done this already before.