Jon Drobny

I believe I understand the cause of this issue. When using a planar surface binding potential, surface refraction in RustBCA is calculated when a particle passes through the energy barrier...

The parry2d switch is working well on the branch linked to the above pull request. Pending changes on that branch: - [ ] for small meshes, performance is slightly worse...

Potential cause: if the CPR rootfinder truncation threshold is too high, there may be an infinite loop caused by truncated terms leading to identical Chebyshev polynomials each iteration and the...

Interestingly, I cannot reproduce this on a different machine. More investigation is needed.

On the original machine with this issue, it appears to be limited to high energy, low nmax, and very small interval limits.

This issue has appeared again, running test_morse.py on multiple machines with a very high number of computational ions. I've been able to trace it to this issue to this nalgebra...

I've pushed an update to rcpr that appears to fix the issue. When convergence failure of the Schur decomposition is detected, the `find_roots` function now simply splits the current interval...

This fix has been stable on the dev branch in all my testing - I'm closing this for now, and can reopen if someone discovers an edge case that causes...

 Data digitized from above: H on Be ``` 20.004726, 0.0029361406 29.880848, 0.013341782 49.791046, 0.024575423 75.00529, 0.016207783 99.68844, 0.017530797 ``` T on Be ``` 20.013433, 0.0025699555 29.85846, 0.01879205...

I think there are ways to do it automatically now, but I'm too stubborn to trust a program to do it right, haha.