Re: AMBER: Restrained minimization problem

From: Thomas E. Cheatham, III <cheatham.chpc.utah.edu>
Date: Mon, 29 Nov 2004 16:02:44 -0700 (Mountain Standard Time)

> I am starting an explicit solvent simulation in Amber7. As suggested in
> the amber manuals, I at first run a minimization with the solute
> restrained. The solute is quite large with 4548 atoms.

This is not considerably large.

> Exceeding lastist in get_istack
> lastist = 2000000
> top_stk= 1786044
> isize = 232008
> request= 2018052
> Increase lastist in the &cntrl namelist

If you read and follow the message, the problems should go away.
Specifically, your restraint information was read in and then the job was
trying to start to allocate memory from the available integer stack. A
request was made for 2018052 ints and this blew the compiled limit of 2M
integers as 1.78M were already allocated. You can get around this error
by increasing lastist in the &cntrl namelist, i.e. in your script, add in
lastist > 2M

 The input file is as follows:
 &cntrl
  imin = 1,
  maxcyc = 1500,
  ncyc = 500,
  ntb = 1,
  igb = 0,
  cut = 12,
  ntr = 1,

  lastist = 4000000,

 &end
Group input for restrained atoms
500.0
RES 1 249
END
END


Note that 500.0 kcal/mol-AA is OK for minimization but way too large for
(subsequent) MD as the frequency of vibration would be too fast. For MD,
lower the restraint value to something like 1.0.

Restraints, depending on the problem, may or may not be important. Their
purpose in the tutorial is under the assumption that you want to initially
stay close to your starting geometry. As water, ions etc are placed
around the solute, the placement is not optimal. Initial equilibration
removes hot spots and lets the ions/water "relax". In general, you do not
want this initial relaxation to perturb your starting structure (assuming
you are mimicking an experimental geometry). If you were looking at an
unfolded protein, then restraining it may not be important...

Good luck.


-----------------------------------------------------------------------
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