Re: [AMBER] NVE Energy/Temperature Drift Issue

From: Andreas Goetz <>
Date: Tue, 2 Jul 2013 13:59:24 -0700

On Jul 2, 2013, at 5:13 AM, David A Case wrote:

> On Mon, Jul 01, 2013, James W. Snyder, Jr. wrote:
>> Also, I am using an explicit solvent model/solvation sphere.
>> From what I understood, this warranted using ntb=0 and igb=0.
> I'm betting that you would be much better off treating this system with
> periodic boundary conditions. It will be *far* faster, and probably more
> physically realistic (since you avoid the vacuum/water interface). Even
> if you have to add a lot more water molecules to make a big enough box,
> you would have a net win.
> Although Amber (with enough care) probably can do an NVE simulation as
> you are trying to do, it is designed for very different types of simulation.
> Basically, the two "designed" options are explicit solvent with ntb=1, or
> GB/PB with ntb=0 and igb>0.

I would follow Dave's advice unless you are planning to run QM/MM (which you might as I infer from the commented ifqnt line in your input) using an external QM program. In this case the speed of the MM code in AMBER will not matter and you have to choose between two evils: a) having a vacuum/water interface or b) using a cutoff for QM-MM electrostatics.

All the best,

Dr. Andreas W. Goetz
Assistant Project Scientist
San Diego Supercomputer Center
Tel  : +1-858-822-4771
Web  :
AMBER mailing list
Received on Tue Jul 02 2013 - 14:00:02 PDT
Custom Search