Re: [AMBER] QM/MM in amber, PME when qm_theory='extern'

From: Andreas Goetz <agoetz.sdsc.edu>
Date: Tue, 14 Oct 2014 22:00:19 -0700

QM/MM calculations with periodic boundary conditions are possible. With external QM programs, PME will be used for the long range electrostatic interactions among the classical MM point charges.

Long range electrostatics between the QM charge density and the MM point charges and the periodic images of the QM charge density are neglected (beyond the cut off which is set by qmcut in the qmmm namelist). To my knowledge a PME compatible theory for general QM/MM models has not been worked out - NDDO and DFTB models in Amber use a discrete representation of the QM charge density (Mulliken point charges) beyond the cutoff. For NDDO models this is an approximation, for DFTB it actually turns out to be exact. For other QM models, as supported by external programs, Mulliken point charges are not numerically stable (nor sensible to use) thus the best one can do is to use a cutoff. Other charge models for which derivatives can be formulated might be useful (e.g. CHELPG), but I know of no example in the literature that would have tested this for MD (I think Herbert has worked out the theory and tested this for QM/MM geometry optimizations recently). For most cases, I would expect a negligible effect on the simul
ations, if your QM/MM cutoff is sufficiently large.

All the best,
Andy

--
Dr. Andreas W. Goetz
Assistant Project Scientist
San Diego Supercomputer Center
Tel  : +1-858-822-4771
Email: agoetz.sdsc.edu
Web  : www.awgoetz.de
On Oct 6, 2014, at 2:27 PM, Brian Radak wrote:
> I believe the 'extern' module literally just passes coordinate and atom 
> type information to the external program. Thus long range electrostatic 
> methods are limited to the specific package.
> 
> I don't think many quantum chemical programs have 'proper' Ewald or PME 
> implemented period (and in this regard SQM is rather ahead of the 
> curve), much less in conjunction with AMBER. In my opinion, simple 
> electrostatic embedding is the next best thing, but may be considered 
> lacking with PBCs. You could do a droplet type calculation, but AMBER is 
> not really designed to support these any more. 'Proper' reaction field 
> methods (like stochastic boundary) are probably even less common than 
> Ewald at this point.
> 
> Sorry if that isn't very helpful. I spent my PhD surrounded by Ewald 
> developers...
> Brian
> 
> On 10/03/2014 09:57 AM, Aixiao Li wrote:
>> Hey,
>>   I am currently doing QM/MM calculations with Amber14, with ORCA as external QM code. And I realized that Amber interface doesn't support the PME calculations by now with external QM codes. So what method instead could be implied for the long rang electrostatic effect? Is the reaction field application possible?
>> 
>> thanks you very much
>> 
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
> 
> -- 
> 
> 	Current Address
> Brian Radak 	BioMaPS Institute for Quantitative Biology
> PhD candidate - York Research Group 	Rutgers, The State University of 
> New Jersey
> University of Minnesota - Twin Cities 	Center for Integrative Proteomics 
> Room 308
> Graduate Program in Chemical Physics 	174 Frelinghuysen Road,
> Department of Chemistry 	Piscataway, NJ 08854-8087
> 
> 	radakb.biomaps.rutgers.edu
> 
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber

Received on Tue Oct 14 2014 - 22:30:02 PDT