Re: [AMBER] &ewald skinnb=2.0d0 /

From: Ross Walker <>
Date: Mon, 01 Jul 2013 21:40:07 -0700

Hi Tu,

Skinnb is the value by which the cutoff is padded when calculating the non
bond list. In principal it should have no effect on the actual dynamics.
It is simply a tuning parameter controlling how often the non-bond list is
rebuilt heuristically. The bigger you set it the more expensive each list
build is but the less often it has to rebuild it.

The error is coming because the GPU code does not rebuild the non-bond
cells as the box size changes - this is fine for most simulations that are
equilibrated but causes problems if the box size shrinks by more than the
skinnb size. This is what is happening in your simulation. Essentially
your box is shrinking by a lot, probably because your initial density is
too low. The simplest approach is to not mess with skinnb but to just
restart the calculation which will build a smaller box and will continue
to run until your box shrinks too much again. At worst you should only
have to do this a couple of times at most and then your density will have
equilibrated and the simulation should proceed without stopping again.

All the best

On 7/1/13 7:47 PM, "\ӵ" <> wrote:

>Dear all,
>Im using amber12 pmemd.cuda_SPFP( Version 12.3) to run MD simulation to
>study the protein folding problem. But the running process is always
>terminated soon. There was nothing error in the md2.out. The log.txt
>exports: Nonbond cells need to be recalculated, restart simulation from
>previous checkpoint with a higher value for skinnb.when I restarted it ,
>this happened again after some time.I find the skinnb
>default is 2.0 angstrom,but when I added *&ewald skinnb=3.0d0 / *to the
> file,this error didn't happen, How does the skinnb value affect the
>motion ?
>Any help will be appreciate. Thanks a lot.
>Tu Zhiping
>AMBER mailing list

AMBER mailing list
Received on Mon Jul 01 2013 - 22:00:03 PDT
Custom Search