Re: Minimisation not converged

From: Kevin C Chan (cchan2242-c_at_my.cityu.edu.hk)
Date: Wed Apr 01 2015 - 22:26:19 CDT

Thanks Jerome,

Problem was solved by running more steps of sd. I have also loaded the
trajectory and it was flipping of a flexible loop.

However in fact with sd minimisation, the energy keeps decreasing instead
of converges. I understand sd is a relatively "slow" minimisation but how
can I tell when to stop? Now I have run more than 100,000 steps and I
choose to step as the decrease of energy between steps was <0.01 kcal/mol.

Regards,
Kevin

Sent from my iPhone

On 1 Apr, 2015, at 15:14, Jérôme Hénin <jerome.henin_at_ibpc.fr
<javascript:_e(%7B%7D,'cvml','jerome.henin_at_ibpc.fr');>> wrote:

Hi Kevin,

Two things you can try:
1) open that trajectory in VMD and see exactly what is oscillating, just in
case it indicates a deeper problem;
2) run a few steps of steepest descent.

Jerome

On 1 April 2015 at 08:20, Kevin C Chan <cchan2242-c_at_my.cityu.edu.hk
<javascript:_e(%7B%7D,'cvml','cchan2242-c_at_my.cityu.edu.hk');>> wrote:

> Dear Users,
>
> I am currently doing minimisation for a protein in water simply in a
> stepwise manner. For the first minimisation in which I fixed the protein
> and only water and ion could move, it worked fine. However for the second
> one in which I fixed water and backbone of the protein instead and let the
> side chains to relax, when I look at the energies using NamdPlot (
> https://www.dropbox.com/s/3kq9ql1z1l7wd9y/mini-2.png?dl=0), it converges
> to two distinct energy values and jumps between them every few steps.
>
> I have tried adjusting parameters like minBabyStep or minLineGoal. It does
> not alter the fact of bouncing energy values but changes the higher energy
> value, i.e. it now bounces from -1.0087e6 to -1.00865e6 instead of
> -1.0084e6.
>
> I am wondering whether I am having problem with the protein structure or
> minimisation parameters.
>
>
>
>
> Kevin
> PhD Candidate
> Department of Physics and Material Science
> City University of Hong Kong
> ukevi_at_gmx.hk <javascript:_e(%7B%7D,'cvml','ukevi_at_gmx.hk');>
>
>

This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:21:02 CST