Re: [AMBER] questions about RESP

From: FyD <>
Date: Mon, 20 Jul 2009 21:46:04 +0100

Dear Jeffrey,

>       Thanks very much for your information. I read that paper and
> make it clear of the different level of optimization. Another
> question is that why the structure should be optimized before
> calculating ESP?

To define (a) "correct/well-defined/canonical_?"
structure/conformation(s) to be used in MEP computation.

> How about a structure near the mimimum point, but not at the real
> mimimum when calculating the ESP?

I see... your structure is too big & you cannot reach convergence... Right ?

When we optimize a structure we always use "Tight" convergence,
instead of using default convergence criteria to better define the
corresponding minimum.

> For large molecules, it is difficult to optimize the structure at
> 6-31G* level. So how about using a conformation obtained at a low
> level optimization?

Did you ever optimize H3O+ (for instance) using HF/3-21G instead of
HF/6-31G* ?

When you have a big structure, decreasing the theory level in QM might
appear an attractive approach. Others in the list might have different
experiences - my feeling is that removing polarization functions for
instance will lead to inaccurate structures. HF/6-31G* & HF/6-31G**
provide reasonable results for organic and bio-organic structures, and
are widely used.

Another problem that shows up when you have a big structure is how
rigorously defining the corresponding conformation(s) ? Thus, quite
often the conformation(s) selected in this case is(are) are not
rigorously defined but rather generated after crude construction &
geometry optimization. This might be a problem as conformation
strongly affect charge values based on MEP computation.

Thus, I would suggest you to split your big molecule into two or more
fragments (with well defined conformations), and use R.E.D. to compute
RESP charges & build the corresponding FF libraries.

regards, Francois

>>       If the optimization and ESP calculation are performed at 
>> different level, e.g.  HF/3-21G for optimization and HF/6-31G* for 
>> ESP calculation, how will it effect the fitted atomic charges?
>> Are there any papers or materials discussing this topic?
> You might find some information about that in
> Hariharan, P. C.; Pople, J. A. Chem Phys Lett 1972, 16, 217.

AMBER mailing list
Received on Mon Jul 20 2009 - 18:08:54 PDT
Custom Search