From: DENILSON FERREIRA DE OLIVEIRA (denilson_at_dqi.ufla.br)
Date: Sun Apr 29 2012 - 18:11:56 CDT

Dear Chris,

Thank you very much for your help.

Best wishes.

Denilson

Denilson F. Oliveira
Laboratório de Produtos Naturais
Departamento de Química
Universidade Federal de Lavras

----- Mensagem original -----
De: Christopher G Mayne <cmayne2_at_illinois.edu>
Para: DENILSON FERREIRA DE OLIVEIRA <denilson_at_dqi.ufla.br>
Cc: <vmd-l_at_ks.uiuc.edu> <vmd-l_at_ks.uiuc.edu>
Enviadas: Sun, 29 Apr 2012 18:13:30 -0300 (BRT)
Assunto: Re: vmd-l: Force Field Toolkit: scaling, weight etc

> 1) In “Opt.Charges” -> “Advanced Settings” -> “Optimize settings”, there are two “modes”: “downhill” or “simulated annealing”. Any suggestion/advice ?

Downhill mode simply runs the objective function downhill, meaning that it cannot get out of local minima. Simulated annealing, in contrast, provides better sampling and is capable of finding other minima in the global energy landscape (if they exist and if you sample enough). My suggestion is to start with Simulated Annealing until you feel that you have found suitable charges near the global minimum, and then run downhill until you reach a reasonable convergence on a set of charges.

> 2) For the optimization of bond and valence angle force constants a scaling factor of 0.89 must be applied to the MP2/6-31G* force constant matrix. Is this scaling included in the calculation carried out in the “Cal. Bonded” window of FFTK ?

We use the hessian calculation directly output from QMTool. The hessian transformation itself is written, at least partially, in C, and I am unfamiliar with the details. Jan (the author of QMTool and ParaTool) used this in ParaTool, and designed it to work for MP2/6-31G* hessian calculation, for what it's worth. I'm sorry that I can't answer any more directly than that.

> 3) In the “Water Int.” window FFTK generates three compound-water Gaussian input files for the oxygen of each carbonyl group. Am I supposed to use the three corresponding log files (output of Gaussian calculation) in the “Water Interaction Energy Data” of the “Opt. Charges” window ?

The authors of the CGenFF paper suggest optimizing charges against both the 120 degree (QM min) and 180 (MM min) positions of the water interaction with carbonyls. It is a compromise such that charges are reasonably accurate for either geometry.

> 4) In the “Opt Charges” -> “Water interaction Energy Data”, one of the parameters is “weight”. Is there any rule for choosing the appropriate value for this parameter ?

I do not have any hard and fast rules. In general, getting the interaction energy correct is much more important than the interaction distance. The latest 1.9.2-alpha versions (available through BioCoRE) include COLP (see the charge optimization "Results" section), which can be used to visualize the individual contributions of energy, distance, and dipole to the objective function. These data should provide good feedback for tuning both the distance and dipole weights.

Chris

On Apr 29, 2012, at 6:37 AM, DENILSON FERREIRA DE OLIVEIRA wrote:

> Dear Chris,
>
> As I told you before, I am trying to develop parameter and topology files for a dianion using Force Field Toolkit (FFTK). Consequently, I would appreciate if you could help me with:
>
> 1) In “Opt.Charges” -> “Advanced Settings” -> “Optimize settings”, there are two “modes”: “downhill” or “simulated annealing”. Any suggestion/advice ?
>
> 2) For the optimization of bond and valence angle force constants a scaling factor of 0.89 must be applied to the MP2/6-31G* force constant matrix. Is this scaling included in the calculation carried out in the “Cal. Bonded” window of FFTK ?
>
> 3) In the “Water Int.” window FFTK generates three compound-water Gaussian input files for the oxygen of each carbonyl group. Am I supposed to use the three corresponding log files (output of Gaussian calculation) in the “Water Interaction Energy Data” of the “Opt. Charges” window ?
>
> 4) In the “Opt Charges” -> “Water interaction Energy Data”, one of the parameters is “weight”. Is there any rule for choosing the appropriate value for this parameter ?
>
> Thank you very much for your attention.
>
> Best regards.
>
> Denilson.
>
>
> Denilson F. Oliveira
> Laboratório de Produtos Naturais
> Departamento de Química
> Universidade Federal de Lavras