Re: Error in implicit solvent MD

From: Aron Broom (broomsday_at_gmail.com)
Date: Wed Mar 27 2013 - 15:39:39 CDT

you could also try what Giocomo mentioned about the option to directly
compute things, it's in a previous thread from today or something like that.

But, as to what I had originally suggested, you could set:

switching off
cutoff 995
pairlistDist 999
staticatomassignment yes

this will ensure full calculation of all the interactions, and will not
bother updating the pair list because there is no longer any value in that.

If you use a smaller value like 16 it could all be fine. Basically what
will happen is that you're energy will no longer be conserved, because some
forces just won't be calculated any longer at certain timesteps, but then
will be calculated later, so the sums might be changing.

If you were trying to run an NVE simulation (i.e. no temperature control)
this would be bad. If you are doing NVT, say with a langevin thermostat,
the energy of the system will continually be moderated by the thermostat,
so it might not matter much (but that's a bit speculative). I think you
can probably find more discussion on this kind of topic, because in
particular, a lot of people using NAMD set the overall timestep to
something like 2fs, but only calculate the nonbonded interactions every 4fs
(so called "multistepping"), and I believe this also has the same issue of
not conserving energy, which may or may not matter (I think in general
you'll find more people of the mind that it doesn't matter).

Also, are you doing this on a GPU? Because if so, there is almost no
improvement in performance for having a cutoff, so best just not to use
one, but on a CPU I imagine this could be quite different.

~Aron

On Wed, Mar 27, 2013 at 4:08 PM, Fugui <theoreticalchem_at_163.com> wrote:

> Dear all,
> I am running an implicit solvent MD, and i tried to set the cutoff value
> to 9999, however, i got an error like that:
>
> Charm++ fatal error:
> FATAL ERROR: pairlistDist must be >= cutoff distance
> But there is no problem if the cutoff value is lower than 8000, because i
> tried value from 10 to 8000 and there was no error. However, if the value
> is 9000, i got the error.
>
> Finally, i set the cutoff value to be 16 which is from NAMD website :
> http://www.ks.uiuc.edu/Research/namd/2.8/ug/node31.html
>
> I do not know whether the cutoff value will influence the result
> significantly, can anyone give me some information on that?
> BW
> Fugui
>
>
>

-- 
Aron Broom M.Sc
PhD Student
Department of Chemistry
University of Waterloo

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:23:06 CST