Re: colvars not scaling well in NAMD 2.10

From: Giacomo Fiorin (giacomo.fiorin_at_gmail.com)
Date: Thu Feb 23 2017 - 21:58:43 CST

Hi Chitrak, I presume that you know that in NAMD 2.12, Colvars now uses the
NAMD's internal functions that compute centers of mass in a parallel
fashion (as also done in tclForces). They will still carry a bit of
overhead, but it's not bad unless you request something like the entire
system.

If you need REST2, you may want to ask Sunhwan and Wei to invest some
effort into porting the patch to the current NAMD. There are several
innovations in newer versions of NAMD that do not make it worth sticking to
an old one.

Giacomo

On Thu, Feb 23, 2017 at 6:36 PM, Jérôme Hénin <jerome.henin_at_ibpc.fr> wrote:

> Hi,
>
> Your problem is entirely the number of atoms. if you pick just one atom
> per lipid (e.g. a glycerol carbon), you should get the same center position
> at a much reduced cost.
>
> Best,
> Jerome
>
> On 24 February 2017 at 00:25, Chitrak Gupta <chgupta_at_mix.wvu.edu> wrote:
>
>> Hello,
>>
>> I am doing solute tempering simulations with a POPC/peptide system using
>> NAMD 2.10. To the best of my knowledge, this is the only version of NAMD
>> that can work with the REST2 implementation (Jo and Jiang, JCTC, 197:304,
>> [2015])
>>
>> I also have 3 colvars to prevent my peptide from interacting with the
>> lipids in the periodic image along Z dimension. These are distance colvars,
>> defining the distance of the N-terminal residue, center residue, and
>> C-terminal residue with the lipid heavy atoms. Number of atoms in each
>> colvar is ~ 10500.
>>
>> From what I understood from the following link, it is the number of
>> colvars and not number of atoms in a colvar that could affect the scaling,
>> correct?
>>
>> http://www.ks.uiuc.edu/Research/namd/mailing_list/namd-l.201
>> 3-2014/1135.html
>>
>> Has anyone else faced this issue? And any suggestions on how best to
>> address this? Would using the absolute Z value of the residues as the
>> colvar give the same effect with a better scaling?
>>
>> Thanks and regards,
>> Chitrak.
>>
>
>

-- 
Giacomo Fiorin
Associate Professor of Research, Temple University, Philadelphia, PA
Contractor, National Institutes of Health, Bethesda, MD
http://goo.gl/Q3TBQU
https://github.com/giacomofiorin

This archive was generated by hypermail 2.1.6 : Sun Dec 31 2017 - 23:21:06 CST