Re: Low global exclusion count error

From: Brian Radak (bradak_at_anl.gov)
Date: Fri Dec 02 2016 - 09:25:30 CST

Setting margin to a nonzero value (perhaps 1.0 or 2.0) also helps on
occasion.

This may not be helpful, but I also tend to have this problem more when
running on a large number of nodes (since more communication is required
for non-bonded interactions). If your patience is wearing thin, you can
probably run on fewer processors and be more assured of stability.

HTH,

Brian

On 12/02/2016 04:43 AM, B.W.J. Irwin wrote:
>
> If the density of the system is a bit off, I have found one may need
> to restart a couple of times with NPT,
> if the box rescales too quickly some of the grids set up for the
> original volume exceed some threshold and the program crashes.
>
> Assuming the NPT runs for a while before crashing (if it crashes
> immediately it might be another problem)
> Remedy,
>
> Run NVT
> Run NPT from NVT coords saving restarts at an appropriate frequency
> Run short NPT from restart coords
> Run short NPT from restart coords
> Run short NPT from restart coords
> repeating until the error stops
> Run long NPT
>
>
> You should be able to verify if this was the problem by tracking the
> volume of the system as you make the transition from NVT to NPT
>
> Hope this helps,
>
> Ben
>
>
>
>
>
> On 2016-12-01 21:21, faride badalkhani wrote:
>> Dear NAMD users,
>>
>> I am performing MD simulations on a polymer in water. I performed
>> 20000 steps of minimization (time step : 1 fs) and then 2 ns of NVT
>> simulations without any problem. Then I wanted to change the ensemble
>> to NPT, but I got the following error:
>> Low global exclusion count errors!
>>
>> I wonder if the system is unstable because it has been running without
>> getting error for 2 ns before going to NPT simulation.
>>
>> P.S. The structure includes an alkyne group that I got its parameters
>> from CHARMM GENERAL force field, but there is no dihedral types in
>> CGen FF for it.
>>
>> Regards,
>> Farideh
>

-- 
Brian Radak
Postdoctoral Appointee
Leadership Computing Facility
Argonne National Laboratory
9700 South Cass Avenue, Bldg. 240
Argonne, IL 60439-4854
(630) 252-8643
brian.radak_at_anl.gov

This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:22:40 CST