AMBER Archive (2009)

Subject: Re: [AMBER] Extreme optimization problems

From: case (
Date: Thu Aug 06 2009 - 10:51:23 CDT

On Thu, Aug 06, 2009, wrote:
> we have a really strange and mad goal - to optimize protein (4881 atoms)
> so the GMAX is less or equal to 1e-7. Is there somewhere a good and
> experienced soul, that can help?

We do this routinely, using the xmin minimizer in NAB or sander. With
Newton-Raphson minimization, you can easily get below 10*-11. Such a
simulation makes good sense in implicit solvent (like GB); it might not work
so well with explicit solvent and periodic boundaries.

Look at amber10/test/nab/ Increase maxiter and
decrease the grms_tol value, and things will probably work.

...good luck....dac

AMBER mailing list