QMCPACK / qmcpack

Main repository for QMCPACK, an open-source production level many-body ab initio Quantum Monte Carlo code for computing the electronic structure of atoms, molecules, and solids with full performance portable GPU support
http://www.qmcpack.org
Other
298 stars 139 forks source link

Jastrow pade2 optimization is misbehaving #412

Open ye-luo opened 7 years ago

ye-luo commented 7 years ago

Reported by @lshulen . During an optimization. The first iteration VMC energy seems correct but the later ones have lower values than they should be capable of variationally.

lshulen commented 7 years ago

Looking at the functor in the code (Pade2ndOrderFunctor), I'm not sure it makes sense. The form it says at the top of the functor is ar +cr^2 / (1+ b*r). I find this weird that the functor would grow as r at large distance. Also the implementation is inconsistent. It looks like it evaluates ar+br^2 / (1+br).

I think another functor in the file makes more sense to target. It is PadeTwo2ndOrderFunctor. It evaluates ar+br^2 / (1+cr + dr^2). There are some inconsistencies to clean up for it as well, but it makes more sense to me.

ye-luo commented 7 years ago

Let us keep the one makes sense and clean up the all the rest.