AMBER Archive (2005)

Subject: Re: AMBER: SANDER BOMB: volume of ucell too big

From: Robert Duke (rduke_at_email.unc.edu)
Date: Thu May 12 2005 - 15:44:11 CDT


I was not following this to see if there is something there that pmemd can't
handle, but if we are talking pme simulations here, pmemd DOES handle things
like changing density in the unit cell (or at least should - all storage
allocation is dynamic, and I check for the various "ewald bomb" conditions
and reallocate storage as needed).
Regards - Bob Duke

----- Original Message -----
From: "David A. Case" <case_at_scripps.edu>
To: <amber_at_scripps.edu>
Sent: Thursday, May 12, 2005 4:27 PM
Subject: Re: AMBER: SANDER BOMB: volume of ucell too big

> On Thu, May 12, 2005, Stern, Julie wrote:
>
>> It is a periodic simulation of alkylthiol chains. The sulfurs
>> are restrained, but apparently did not stay in place. There is
>> no solvent at all. I did look at the simulation (22 frames at
>> .1 ps/frame was all I got). The volume expanded roughly 6 angstroms
>> in both the x and y directions but hardly anything in the z-direction.
>
> I'm assuming that you are trying to simulate a neat liquid of alkylthiol
> chains (how long is the chain)? The results reported above suggest that
> your
> initial density is way too high (hence the volume is expanding rapidly,
> even over 2 ps.)
>
> Amber is not well suited to simulations where the volume changes
> significantly
> as a function of time: certain data structures are allocated based on the
> initial structure, and are not updated correctly when the volume changes
> by a
> large amount. Mike can probably suggest which variables to change to
> allow
> bigger changes, but you could also try the following:
>
> Set taup=20.0 or so, and run a few picoseconds; then restart.
> Do this repeatedly until the volume equilibrates
>
> This will slowly increase the unit cell volume, and the restarts will
> force
> all of the data structures to be rebuilt. Be *sure* to visualize your
> trajectories to be sure that the structures you get are meaningful.
> Unless
> you have a good reason to restrain atom positions, it is best to avoid
> that.
>
> ....good luck....dac
>
> (note: if you are not trying to simulate a neat liquid, then you may be
> heading in a wrong direction? Do you expect your system to have a
> well-defined density at the given temperature and pressure?)
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
>

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu