Path-integrals status

[Harald Forbert]

Hi Axel,

On Oct 16, 5:30 pm, Axel <akoh... at gmail.com> wrote:
> can somebody please comment on the status of path-integral support in
> cp2k?

basic support is in and it should work, though you probably have to
know what you are doing (but then don't you always :) ).
Mainly the initial setup of a path-integral calculation could be made
easier.

Although supposedly any type of force evaluation should work,
I have so far only tested pure QM (QS) cases. And I need to
check if it works with the latest changes to how restarts are done.

Only basic estimators for kinetic/potential energy etc are in, so
if you need other things you have to post-process the
trajectory(ies) at the moment. (And I have only limited plans atm
in what to add. Maybe Fawzi is also working on something there?)

But is anybody besides me actually using path-integrals in CP2K?

> it has been almost exactly a year since this has been asked the last
> time and since the respective code was changed in the cvs (except for trivial
> stuff).
>
> is somebody actually working on it and not committing to the cvs or
> is the development halted/delayed/postponed/abandoned/...?

I am still working on some aspects of it, though most of that is not
ready for
public consumption (and my code additions are at the moment pretty
monolithic, so it likely breaks compilers other than the ones I use :)
I heard
pint_types.F is problematic for compilers as is already).
Besides I don't expect many people to actually want to use these
additions,
as they are highly specific.

I have a small general patch to (hopefully) improve speed in parallel
that
I always planned to submit, but so far haven't gotten around to it. I
also need
to test the code against the latest changes (there have been some
conflicts in the last few days commits which I think I resolved, but
hadn't
time to actually check). And it touches parts of the code that are not
path-integrals only, so I like to be careful there...

Anyways, if you tell me what you are interested in, I might be able to
tell you if that already works, should already work, is easy to add or
not....

Harald