12:48:13 PM PDT - Thu, Oct 11th 2012 |
|
Okay, you may want to look for yourself at the code in esp_fit.F Looking at it, there is a definition for q0. Digging a little further and seeing where it is defined, turns out there is an not-advertized input variable called qh, which represents q0 in the RESP fitting. You can try setting:
qh <real q0>
in the esp block. I don't know why it is not included in the manual, may be it doesn't work properly. But, based on the equations you listed, it is the q0 you're looking for.
Bert
Quote:P99 Oct 11th 9:37 amQuote:Bert Oct 9th 1:16 pmI have no clue what RESPA stands for and can't find anything online about it either.
Sorry for the typo, of course not RESPA, but RESP.
Quote:Bert Oct 9th 1:16 pmI guess you want to tell ESP to give an atom a minimum or maximum charge. This is not available in NWChem's ESP module. You probably could write the code yourself, modifying ESP.
I would like to do as described in the article, which introduced the RESP J Phys Chem 1993, v.97. p.10269
It describes three models with penalty functions:
(model 1) x2 = a (q 0 – q) 2
(model 2) like (1) but q 0=0
(model 3) x2 = a ((q 2 +b 2) 1/2 – b)
My first guess is that there is no control in NWChem for q 0. So I can use (model 2) and (model 3), but not (model 1). Is this true?
I see that the directive "constrain" does not provide the exact charges.
So my second assumption is that "constrain" not the charges fixing, but really is the (model 1). May be this true?
|