Lubos,
I think you may be right about pmemd 3.1 in this regard. There are three
different runmd*.f90 implementations (non-cit, amber 6, and amber 7), and it
looks to me like they all contain the appropriate code. I presume my
earlier comments came from looking at 3.03, and when I added stuff for
sander 7 partial support, I changed the runmd code but never effectively
enabled it. Anyway, it is easy enough to try, but check your results
carefully against sander 7, being aware that any systems with intermolecular
crosslinks could give you slightly different virials (I think if you test
against the dhfr benchmark, you will be okay). The different virials issue
(basically, pmemd does virials like sander 6) goes away in pmemd 8.
Regards - Bob
----- Original Message -----
From: "Lubos Vrbka" <shnek.tiscali.cz>
To: <amber.scripps.edu>
Sent: Wednesday, March 31, 2004 2:44 PM
Subject: Re: AMBER: non-isotropic pressure scaling
> hi,
>
> > It would also be possible to patch pmemd 3.01 or better, 3.1 (which has
> > better performance and scaling, and is available on the amber web site)
to
> > support this; you would need to look at the code in mdin_ctrl_dat.f90
and
> > runmd_*.f90, and compare to equivalent code in sander 7 (and do
comparison
> > test runs to be sure you got it right).
> i had a look into the source and compared sander7 to pmemd 3.1. i got
> the impression that the only thing needed to get it running is to remove
> the initial test (if ntp=2 print that it is not supported) in
> mdin_ctrl_dat.f90, since ALL code that it is sander7 source (runmd.f,
> step 7, ~line 1365) is present in pmemd 3.1 (runmd.f90, step 7, ~line
> 713). there are all the tests whether ntp .eq. 1 or ntp .gt. 1...
>
> > The non-isotropic scaling
> > capability is not in pmemd 3.01 - 3.1 because these programs were
intended
> > to be mostly sander 6 compatible, and it did not support non-isotropic
> > scaling.
> so could it be, that the feature was just disabled in pmemd for
> compatibility reasons, but in fact is in the source "waiting to be
> uncommented"?
>
> i'm not fortran programmer, so i may be wrong, but all the code
> indicates to me that i might be right :)
>
> thanks for your help. regards,
>
> --
> Lubos
> _._"
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Wed Mar 31 2004 - 21:53:00 PST
