#2 Periodic Wavefunctions

[frozenflip]

<snippets from emails>

Two major features that keep QMcBeaver from being a produciton
level code are psd. potentials and periodic wavefunctions. Both
are easy on paper but may be a major coding project. I really
don't know enough about them to know how big of task they are.
These are big enough things that you shouldn't go into them
without some well thought out plans.

<snip>

Psd. pot. are a weird can of worms. Before starting to implement
any of this, you really need to read the lit and determine if you
trust them enough to include. The will speed up the calcs, but they
leave questions about the ultimate accuracy of your calcs.

<snip>

CASINO does the psd. pot. and is periodic (can request a copy of
the code but need to sign a bunch of stuff -> might not give it to
QMcBeaver writers, however). They have to add all kinds of stuff
to check how screwed up the calc is. Like Chip says psd. pot.s will
introduce errors. The size of these errors are generally on the
same order as the errors you are trying to correct for (the whole
reason to use QMC). Granted you are getting the point of view of
a couple of anal guys. The other side says that there is no point to
using QMC if you can't do bigger interesting systems and get
converged estimators. Reality is probably in the middle
somewhere. "Fahy" (I think -> check spelling) is a good guy to
start reading on these topics.