Hi Fatima -
There undoubtedly IS a performance advantage to using pmemd; it just won't
be as large as you may be used to for the rest of pmemd functionality.
There are a bunch of optimizations that you can take advantage of, but you
will hit the scaling bottleneck quicker due to the less efficient coordinate
and force propagation scheme. So while you could use sander.mpi, I would
recommend trying pmemd if the jarzynski relation will cover your needs
(glancing at the manual, there is more new stuff in the category of "steered
md" out around section 4.6 I think, perhaps requiring remd - not sure as I
don't use it - but this stuff is not supported in pmemd).
Regards - Bob
----- Original Message -----
From: <fatima.chami.durham.ac.uk>
To: <amber.scripps.edu>
Cc: <rduke.email.unc.edu>
Sent: Saturday, May 17, 2008 6:26 AM
Subject: Re: AMBER: STEERED MD
> Quoting Robert Duke <rduke.email.unc.edu>:
>
>>
>> Okay, "minor" correction here. I DO support Jarzynski relation stuff in
>> pmemd, ie. jar = 1. I think this falls under the perview of nmropt = 1,
>> and
>>
>> probably is not particularly fast because there we use simpler
>> communications protocols when a bunch of the nmr code is invoked (has to
>> do
>> with not touching some ancient code, basically). I was thinking of some
>> stuff that came out of Rebecca Wade's group; that is actually supported
>> only as a patch in earlier releases, though, so I don't know to what
>> extent
>> that would be considered "supported". Sorry for my confusion (I have
>> heard
>> steered MD used in different contexts).
>> Regards - Bob Duke
>
> Thanks for the feedback.
> So the Jarzynski relation is implemented in PMEMD but not improved to run
> at
> high parallel performance...so I better just use SANDER.MPI.
>
> Another thing is that Umbrella sampling method necessites the use of the
> weighted histogram to get free energy...I downloaded the WHAM utility
> program
> but I have not used it yet ...
> For the Steered MD...just using the Jarzynski relation will yield the free
> energy..
> am I getting things right here ...
>
> regards
> Fatima
>
>>
>> ----- Original Message -----
>> From: "Robert Duke" <rduke.email.unc.edu>
>> To: <amber.scripps.edu>
>> Sent: Friday, May 16, 2008 1:15 PM
>> Subject: Re: AMBER: STEERED MD
>>
>>
>> > Sorry, no. Anything really fancy has three hurdles to pass: 1) are
>> > there
>>
>> > enough potential users, 2) is there actual potential to speed it up
>> > (some things introduce requirements that basically run counter to good
>> > performance), and 3) is it something interesting enough to the folks
>> > that
>> > pay me that I can justify doing it. At least in the last round, this
>> > one
>> > was not mentioned.
>> > Regards - Bob Duke
>> > ----- Original Message -----
>> > From: <fatima.chami.durham.ac.uk>
>> > To: <amber.scripps.edu>
>> > Sent: Friday, May 16, 2008 12:25 PM
>> > Subject: AMBER: STEERED MD
>> >
>> >
>> >> Dear Amber users,
>> >>
>> >> does the PMEMD supports the steered MD calculation?
>> >>
>> >> best wishes
>> >> Fatima
>> >> -----------------------------------------------------------------------
>> >> The AMBER Mail Reflector
>> >> To post, send mail to amber.scripps.edu
>> >> To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
>> >> to majordomo.scripps.edu
>> >>
>> >
>> > -----------------------------------------------------------------------
>> > The AMBER Mail Reflector
>> > To post, send mail to amber.scripps.edu
>> > To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
>> > to majordomo.scripps.edu
>> >
>>
>> -----------------------------------------------------------------------
>> The AMBER Mail Reflector
>> To post, send mail to amber.scripps.edu
>> To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
>> to majordomo.scripps.edu
>>
>>
>
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
> to majordomo.scripps.edu
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
to majordomo.scripps.edu
Received on Sun May 18 2008 - 06:08:05 PDT
