Re: [AMBER] NVT equilibration crash using CUDA

From: Jason Swails <>
Date: Mon, 9 Jun 2014 11:39:34 -0400

On Mon, Jun 9, 2014 at 10:59 AM, James Starlight <>

> Jason,
> thanks for suggestions. Could you also tell me 1) may I use initial heating
> of my system in NPT (not NVT) which could prevent such density fluctuations
> during this stage using appropriate pressure coupling constant?

​Probably.​ Depends on the other settings. I typically heat over ~100 ps
with restraints on the solute using NVT and then switch to NPT to stabilize
the density. But I've also used NPT to heat before without adverse
consequences. You should inspect your trajectory if you are unsure whether
there are problems or not.

However, heating with NPT will not lessen how much the unit cell will
shrink. The only thing that it will help with is preventing the formation
of bubbles if you see bubbles begin to form during your heating stage.

2) What
> problems might be occurred during simulation in octahedral boxes?

​Truncated octahedrons don't support anisotropic scaling (at least in
Amber). However, it typically doesn't matter what shape periodic cell you
use. [1]


[1] Except when it does matter, like for crystal simulations (that need to
use the same shape box as the asymmetric unit cell) or instances where
anisotropic pressure control is required, like for surfaces and layers
(that typically require an orthorhombic box).

Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Mon Jun 09 2014 - 09:00:02 PDT
Custom Search