The CASINO forum

How do I go about the preparation or generation of a gwfn.data file.

Dear Osarodion
It is mentioned in quite detailed form in page 101 of the manual. However, I am going to describe what I have done to generate gwfn.data files.
First of all we need Gaussian program which provides us two important files, the checkpoint file name.chk and the output file NAME.LOG (or name.out), depending on the system we use. We should change the name.chk file to a formatted checkpoint file using the FormChk utility (Gaussian window-->Utilities--> FormChk; in my G03W. A window will pop up then so as to allow us to select the name.chk file.). It will convert the file to name.fch.
In my case I have just renamed it as name.Fchk, and it worked.
When I use Gaussview to feed Gaussian, it always gave NAME.LOG as output. And gaussiantoqmc code in CASINO do not accept it as the output file. I have to save the name.gjf file first, then open it using Gaussian to run the calculation and get the output as name.out.
After we have both the name.Fchk and name.out files in the same directory, we can just run gaussiantoqmc and follow the prompts.
The name of both the .Fchk and .out files must be the same.
It will give name.qmc file, which is then renamed to gwfn.data and should be saved to our working directory, where we have put the corresponding input file , since CASINO reads only gwfn.data (but not something like gwfn1.data or gwfnh2o.data, as far as I know).
This worked for me. If you are a G03W and CASINO in Ubuntu user, hope it will also work for you. If it didn't, experts are here.
-Jhalak

You can also use the GAMESS and CRYSTAL programs to generate gwfn.data files in addition to the GAUSSIAN program. Full instructions are provided in section 8 of the manual starting on p.95 (the full manual is included in the list of links at the top left of this page).

Mike

Dear Mike,

Can we prepare gwfn.data file for periodic systems (crystals) using Gamess (US) ?

Jhalak

Hi Jhalak,

As far as I know, GAMESS cannot do periodic systems, but CRYSTAL and GAUSSIAN can (of these three codes I have personally only used CRYSTAL).

Of course, as always, Google is your friend and one can find a table such as the following:

http://en.wikipedia.org/wiki/List_of_qu ... s_software

which has a column stating which codes support which periodicities and this appears to confirm what I say.

Which is better for periodic systems - CRYSTAL or GAUSSIAN? Well, CRYSTAL was designed for periodic systems
from the outset (the outset being in about 1978), whereas in GUASSIAN it was bolted on much later. I remember one of the
CRYSTAL developers laughing hysterically when I asked him if the GAUSSIAN implementation was any good - they are in general very rude about the competition. Of course, that was ten years ago - I have no idea whether the GAUSSIAN developers have since improved it.

Hope this helps,
Mike

And an additional point which occurs to me. The guy who wrote the CASINO-GAUSSIAN interface (Andrew Porter) was a student with us about 12 years ago at which point the GAUSSIAN periodic implementation was either non-existent or not working very well and so he will have ignored it. So the converter program which switches GAUSSIAN output to CASINO input will not support periodic systems.

Thus the current state of play is that if you want to do periodic quantum Monte Carlo calculations with Gaussian basis sets then you can only use CRYSTAL.

This is a general problem we have with CASINO being dependent on trial wave functions produced by other programs. Any in-house expertise we have for particular programs tends to disappear when the person in question moves onto other things. Thus our GAUSSIAN expert (Andrew Porter) left in 2002, our GAMESS expert Alexander Badinski about 5 years ago, and they have never been replaced. Luckily I am the in-house man for CRYSTAL so I ought to be able to help with you that, and I'm unlikely to disappear any time soon. Well, I hope not.

Of course, if you're looking for a project, why not volunteer to make the GAUSSIAN converter work for periodic systems.

Mike

Dear Mike,

To my personal experience (which is very little), G03W deals periodic systems very poorly. I am unaware about the capabilities of G09 (I do not use it). If it works fairly well, I can spare some months as a part-time volunteer (I have to teach on another part ).

Jhalak