Re: AMBER: Restrained minimization problem

From: Bill Ross <ross.cgl.ucsf.edu>
Date: Mon, 29 Nov 2004 15:25:58 -0800 (PST)

>> > How can I resolve this problem to move on?
>
>> You might skip using restraints since minimization is very local.
>
> So, if I can simply skip restraints in the minimization, does this hold
> true for the actual MD part as well (because if I need to use restraints
> there, I'm assuming I'd run into the same problem again).

Maybe you won't :-) There are alternatives to the restraint method of
equilibration, but it seems to be the the most popular one, so I'll
assume that you need it for MD.

If you run your minimization or dynamics from the command line, you
may see a system error that will tell you what the real problem is,
which may well have nothing to do with your restraints. When faced
with a problem, it is always good to try a slight permutation without
the assumed source to check assumptions :-)

A really big system might be a solute of 50-100K atoms. Even smallish
systems can require more memory than the default allowed to a unix
user, in which case a system error should be printed. The 'limit'
unix command allows some adjustment of this.

> When running the minimization without
> restraints on the solute, the program terminates with the following message:
> ... [ewald problem]
> ... Increase lastist in the &cntrl namelist
> ... It seems to me that I either the system is too large,
> and that there is a parameter I need to increase in the program itself.
> What would that be and where do I find it?

Try lastist per the msg. I see Tom Cheatham has added some good advice.
Usually the error messages are worth a careful read, even though parts
may be cryptically programmer-oriented.

> Further, if I
> can possibly skip these things, why are they recommended in the amber
> tutorials? Is it just to ensure a "smoother" run or is there something
> more behind this?

It is more foolproof for a variety of user errors (bad solvent/ion
placement), is my guess. Although one could make a philosophical
argument that the solute shouldn't move at all until solute and
ions move, the effects of minimization are usually so slight that
I would never bother to restrain solute for, say, minimization of
leap-placed solvent/ions if I were running simulations. For initial
MD it is critical if you want to start data collection from your
initial config without equilibration-induced artifacts.

Bill
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Mon Nov 29 2004 - 23:53:01 PST
Custom Search