AMBER Archive (2002)

Subject: Re: About the Memory allocation in MM_PBSA and SANDER

From: Alan Leo (ralyboy_at_sohu.com)
Date: Fri Oct 04 2002 - 07:46:15 CDT

Hi, Holger:

I have resolved the problem. Actually the Maxpr should a little larger
than Natom*(Natom-1)/2.

Thanks for your help.

Best,
Alan

----- Original Message -----
From: "Holger Gohlke" <gohlke_at_scripps.edu>
To: "Alan Leo" <ralyboy_at_sohu.com>
Sent: Thursday, October 03, 2002 8:41 PM
Subject: Re: About the Memory allocation in MM_PBSA and SANDER

> Dear Alan,
>
> I think you mixed something up below. If GB=0, igb should be 0. In that
> case, the maximum possible number of pairs can be natom * (natom-1) / 2
> and, hence, this is the value for MAXPR in sizes.h in case of an
> "infinite" cutoff. Actually, I do not see at the moment why your value
> of 86*10^6 should not be large enough? If GB=1, igb > 0. In that case,
> MAXPR is not relevant. Are you sure you re-installed sander after
> increasing MAXPR?
>
> Best regards
>
> Holger
>
> >>>>>
> Alan Leo wrote:
> I meet with the MM calculation problems when I performed MM_PBSA.
>
> If I use the following option to avoid MM calculation in mm_pbsa.in file
> ***********************
> #
> MM 1
> GB 0
> PB 1
> MS 1
> #
> ***********************
> The generated sander input file is GB MM calculation(igb=4).
>
> If I use the following option involving MM calculation in mm_pbsa.in
> file
> ***********************
> #
> MM 1
> GB 1
> PB 1
> MS 1
> #
> ***********************
> The generated sander input file is MM calculation(igb=0), which doesnot
> use
> GB.
>
> The problem is that the former sander MM input file calculation can be
> through, and
> the later one cannot through because of the number of nonbonded
> pairs(MAXPR overflow),
> even though I adjust it to a little bit larger than the largest distance
> between two
> atoms(cut=99.0). I recompile the sander by rewrite the sizes.h file in
> sander package
> (adjust it to MAXPR=86000000, 4000000 in original sizes.h file). The
> later sander
> input file generated by MM_pbsa (I had modified the cut=49.99) still
> failed.
>
> Q1: My calculation system is about 7500 atoms protein excluded water
> and Na+,
> the number of non_bonded interaction should be equal to 7500*7499,
> shouldn't it?
> And the memory requirement is about 7500*7499*8( or 32 for double float)
> bytes,
> I am not sure if it is right or not? How to adjust MAXPR in sander
> sizes.h file
> according to protein size?
>
> Q2: What is difference between the GB and PB MM energy of Protein in
> Gas?
> If the difference is not big, of course I will choose GB MM calculation.
> <<<<<
>
> --
> +++++++++++++++++++++++++++++++++++++++++++++
> Dr. Holger Gohlke
> Dept. of Molecular Biology, TPC15
> The Scripps Research Institute
> 10550 N. Torrey Pines Rd.
> La Jolla CA 92037 USA
> phone: +1-858-784-9788
> fax: +1-858-784-8896
> email: gohlke_at_scripps.edu
> +++++++++++++++++++++++++++++++++++++++++++++
>
>