VMD-L Mailing List

From: Hall, Lisa Michelle (lhall_at_sandia.gov)
Date: Fri May 07 2010 - 10:46:27 CDT

Hi Axel,

Thanks for looking at this.

On 5/6/10 6:46 PM, "Axel Kohlmeyer" <akohlmey_at_gmail.com> wrote:
>> if i get you right, what you would
>> need is some kind of exclusion that states if two atoms have the
>> same residue/fragment/chain/... info, they should be excluded from
>> the g(r) calculation.

That's right; in my own python script, it simply tests each pair to see if they are on the same molecule, and doesn't add the same molecule pairs to the histogram. I do agree that it should be normalized correctly already if you give it the full atom selections and then exclude some from the histogram while adding them up. Personally, reading in LAMMPS dump files (thanks to you again!) my molecules become 'residues'.

>>how soon would you (or others) need this to be fully implemented
>>in measure gofr? realistically?

I'm in no hurry. This script is fast enough for me right now, (I haven't timed it, but it is considerably (at least 10x) more speedup versus my python script than the 16x speedup one would naively expect from VMD running 16 threads, for an overall speedup of more than 100x) it's just slow and clumsy for me to change it and apply to other related systems, since I don't know tcl.

Best,
Lisa

PS. I meant to mention for anyone trying to do this, if you are reading in a LAMMPS trajectory, at least for me it perfectly turns molecules into residues only if the first configuration is sorted so that the atoms of each molecule are read in consecutively without being broken up (else it will occasionally break up a molecule into 2, but I need to sort anyway).