Re: [AMBER] {SPAM?} Re: Problem with TI minimization

From: Matthew Guberman-Pfeffer via AMBER <>
Date: Fri, 2 Sep 2022 22:39:58 -0500

Hi Dr. Case,

Thanks for the comments. The issue with rising energies seems to be entirely because I accidentally used a submission script for pmemd.cuda; I know I’m not suppose to use GPUs for minimization! On a CPU, the energies are nicely becoming more negative with the input file at the end of this message.

However, a few questions remain.

1) I get the following error message with ntmin=1

ERROR: Minimizations with ifsc=1 require the steepest descent algorithm.
                 Set ntmin to 2 and restart.

I suppose I could minimize the lambda end points with ntmin=1 and the prmtop not set up for TI yet. Then, re-build the system with TLEaP and ParmEd to setup the prmtop for TI. Does this seem better than just doing steepest descent (ntmin=2) with the prmtop setup for TI?

Also, what should I look for in the output to make sure the TI calculations make sense? For example, I’m running minimizations with lambda = 0.0, 0.5, and 1.0. I see that DV/DL (at the current step) for these minimizations is respectively +481, 30, -428, which seems roughly to make sense—I realize I should really compare averages. What else should I check?

For example, is there a way to check that the charges on the TI but non-softcore atoms are being scaled correctly?

Input file:

   imin=1, ! Perform an energy minimization
   ntb=1, ! Boundaries (=1, constant volume)
   cut=10.0, ! Cutoff
   ntmin=2, ! Method of minimization
   ncyc=10000, ! Number of steepest descent cycles
   maxcyc=50000, ! Maximum number of minimization cycles
   igb=0, ! GB model (=0, explicit solvent)
   ntr=1, ! Turn on positional restraints
   restraint_wt=10.0, ! 10 kcal/mol.A**2 restraint force constant

   icfe = 1, !free energy calculations
   ifsc = 1, !1 SC potentials are used
   clambda = 0.50, !Zero corresponds to the unperturbed Hamiltonian V0. l=1 corresponds to the perturbed Hamiltonian V1
   scalpha = 0.5, ! default value is 0.5
   scbeta = 12.0, ! Default value is 12A2
   logdvdl = 0, !If set to .ne. 0,calculated during every step of the run will be printed out at the end of the simulation
   klambda = 1, !


> On Sep 2, 2022, at 3:24 PM, David Case via AMBER <> wrote:
> *Message sent from a system outside of UConn.*
>> ntr = 1, restraint_wt = 100.00,
>> restraintmask='!:WAT',
> This is very high restraint weight; are you sure that you don't want any
> of the waters to move?
> Sorry: I missed the "!" character. Still, if you want to minimize waters, be sure that SHAKE is turned on (ntc=ntf=2).
> [Not setting this can sometimes, but not aways, cause problems.]
> ....dac
> _______________________________________________
> AMBER mailing list
> _______________________________________________
> AMBER mailing list

AMBER mailing list
Received on Fri Sep 02 2022 - 21:00:06 PDT
Custom Search