Re: failure in RATTLE algorithm using the adaptive biasing force method

From: steve.kaminski_at_mail.tu-berlin.de
Date: Wed Nov 12 2008 - 03:37:11 CST

Quoting Giacomo Fiorin <gfiorin_at_seas.upenn.edu>:

> Hi Steve, which is the value of that dihedral in the starting
> configuration? It might be that the system is initially far outside
> of the window you've chosen. In that case, the boundary force which
> is applied to make it stay within the window is just too large. If
> that is what's happening, you should pick a configuration that falls
> within the window to start the simulation. Since you've completed a
> run with the whole -180:180 range, you should have plenty of them for
> whatever window you define.
>
> Giacomo
>
> ---- -----
> Giacomo Fiorin
> Center for Molecular Modeling at
> University of Pennsylvania
> 231 S 34th Street, Philadelphia, PA 19104-6323
> phone: (+1)-215-573-4773
> fax: (+1)-215-573-6233
> mobile: (+1)-267-324-7676
> mail: giacomo.fiorin_<at>_gmail.com
> web: http://www.cmm.upenn.edu/
> ---- ----
>
>
>
> On Tue, Nov 11, 2008 at 2:43 PM, <steve.kaminski_at_mail.tu-berlin.de> wrote:
>> Dear all.
>>
>> My aim is to calculate a PMF of a certain molecule as a function of a
>> dihedral angle. Therefore, I use the adaptive biasing force method. As long
>> as I scan the complete dihedral space (xiMin=-180 / xiMax=180) the
>> calculations run without problems. But in my case I just want to scan a
>> reduced space, so if I change the values e.g. of xiMin to -100 or any other
>> value, I get immediately the following error message from NAMD:
>>
>> Warning: Constraint failure in RATTLE algorithm for atom 19
>> Warning: Constraint failure in RATTLE algorithm for atom 22
>>
>> If I turn off rigidBonds in the NAMD config-file I receive the following
>> error:
>>
>> ERROR: Atom 20 velocity is 1050.76 -3548.3 -5428.46 (limit is 5000)
>> ERROR: Atoms moving too fast; simulation has become unstable.
>> ERROR: Exiting prematurely.
>>
>> Thank you very much in advance for your suggestions.
>>
>> Steve
>>
>>
>
>

Hi Giacomo, Jerome.

Thanks for your fast reply.
I did what you've both suggested, namely checked the starting value of
the dihedral I want to scan. The value is -163 and the first ones
sampled during the md, before it crashes, can be read from the output
as:

TCL: ABF> Xi at timestep 50 : -166.472915166
TCL: ABF> Xi at timestep 100 : -160.187081635
TCL: ABF> Xi at timestep 150 : -162.144172289
TCL: ABF> Xi at timestep 200 : -170.070161497

So, I would suggest that a simuation in a range of e.g. xiMin=-180 to
xiMax=0 should work. Whereas a range of e.g. xiMin=-50 to xiMax=0
should not, right?
In fact, both simulations crash with the same error message.
My next idea was to decrease the boundary force (forceConst) from 10
to 1 kcal/mol/A^2 and now it works! Do you ave any experience which
magnitude for the boundary force of a dihedral scan is suitable?

Steve

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:50:05 CST