<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<META NAME="Generator" CONTENT="MS Exchange Server version 6.5.7652.24">
<TITLE>double precision minimization accuracy limits.</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/plain format -->
<BR>
<P><FONT SIZE=2>Hi,<BR>
<BR>
Im attempting to minimize a protein-solvent system (>10,000 waters) using the double precision program<BR>
to minimize structures dumped from a single precision dynamics run.<BR>
<BR>
Ive switched off all constraints, including switching to flexible SPC waters, added switching functions to the<BR>
LJ and coulombic potentials to make a smooth cutoff.<BR>
<BR>
However, all minimizations (conjugate gradients, l-bfgs) all seem termate to terminate at what I consider<BR>
to be large derivatives for a double precision program.<BR>
<BR>
e.g.<BR>
<BR>
Polak-Ribiere Conjugate Gradients converged to machine precision in 123 steps,<BR>
but did not reach the requested Fmax < 0.01.<BR>
Potential Energy = -1.64197713818716e+06<BR>
Maximum force = 2.32094495259875e+02 on atom 31761<BR>
Norm of force = 6.19775726238427e+00<BR>
<BR>
<BR>
Low-Memory BFGS Minimizer did not converge to Fmax < 0.01 in 1501 steps.<BR>
Potential Energy = -1.65198292382305e+06<BR>
Maximum force = 2.70928575898524e+02 on atom 25485<BR>
Norm of force = 6.58974288803592e+00<BR>
<BR>
Is there some step size variable I should be making smaller to allow the minimizers to make even smaller<BR>
changes to the coordinates (Im assuming from the message its at 10-12 changes or less).<BR>
<BR>
Am I expecting too much of the potentials, forces etc. to deliver lower forces?<BR>
In particular Id like to have the Maximum force under 10<BR>
<BR>
By the way I made some modifications to the tpbconv program to allow me to pass full double precision coordinates<BR>
to the program as otherwise writing out to a .gro file jumps the max force and norm significantly and I dont<BR>
see a way to continue from steepest descent with conjugate gradients or l-bfgs without performing a new mdrun run.<BR>
- and lbfgs seemed to take forever to work on its own and get nowhere - cg was way below it in force/energy after<BR>
far fewer iterations.<BR>
<BR>
Thanks<BR>
<BR>
David<BR>
<BR>
<BR>
<BR>
<BR>
</FONT>
</P>
</BODY>
</HTML>