Either use pymatgen version if the fix is in their version, or make fix suggested by Daniil in our copy, or suggest to pymatgen to fix:
"
One thing I noticed while doing this though, is you can speed up the code 10x at least by moving the "dV = np.gradient(V)" line outside the for loop in string_relax() - the potential never changes in this approximation so re-evaluating the gradient every step is not necessary, and slows down the code a ton.
"
Suggestion implemented. It is much faster. Submitted pull request to implement in pymatgen's version as well. Next MAST version could reference outside pymatgen version.
Either use pymatgen version if the fix is in their version, or make fix suggested by Daniil in our copy, or suggest to pymatgen to fix:
" One thing I noticed while doing this though, is you can speed up the code 10x at least by moving the "dV = np.gradient(V)" line outside the for loop in string_relax() - the potential never changes in this approximation so re-evaluating the gradient every step is not necessary, and slows down the code a ton. "