QMCPACK
Need to consolidate TWF/WFC::evalGradSource
After the removal of finite difference https://github.com/QMCPACK/qmcpack/pull/5380, certain clean up can happen
- Removal of
ionsparticle set passing as an argument. This was needed because each walker needed a separateionsto operate moves. Since it is no longer needed, each component that involves ions can use its internal const reference to ions. - electron particleset should made possible if possible.
- There are two evalGradSource functions and preferably reduce to one.
- Remove all the clones of
ions.
@rcclay I had a hard time locating the per walker clone of ions and I'm wondering if the removed finite difference code actually worked properly in threaded runs. Do you remember where are the clones of ions?
Hi @ye-luo I thought I had told you I had a huge PR with all these modifocation with the linking of the TW functions to the batch version of the code. You asked me to not do a PR for now and now you open one... am I understanding differently? Would you mind waiting for our PR or should we just keep separate branches?
I don't have any PR prepared to address this issue which is mostly a TODO list I have in mind. A few of them was discussed when the finite difference derivatives in JeeI was first introduced. If your PR has already addressed part of it, that will be great.