Re: ABF simulation with extendedLagrangian on keeps crashing

From: JťrŰme Hťnin (jhenin_at_ifr88.cnrs-mrs.fr)
Date: Tue May 03 2011 - 10:23:17 CDT

On 3 May 2011 15:11, Ajasja Ljubetińć <ajasja.ljubetic_at_gmail.com> wrote:
> Hi,
> I tried that, but it still crashes. I'll try to reduce the integration step
> even further (to 0.5 or 0.2 fs) and play around with temperature control
> (perhaps disable it).

I'd only expect disabling it to make things worse. If anything we need
a thermostat on the extended DOF.

> A while ago I boasted, that I will try to derive the analytical derivative
> for the angle colvar, but even after pulling in a couple of
> theoretical physicists we failed miserably. It would take me (us) too much
> time to grasp the basic concepts and the maths behind ABF.

Good news on that front: I have the expressions for the angle.
Implementing it is on my short- term TODO list.

Jerome

> On Fri, Apr 22, 2011 at 22:38, J√©r√īme H√©nin <jhenin_at_ifr88.cnrs-mrs.fr>
> wrote:
>>
>> Hi Ajasja,
>>
>> Just a guess, but the force constants for the walls are a tad high.
>> You may want to try something on the order of 1, especially if the
>> crashes happen near the boundaries.
>>
>> Jerome
>>
>> On 22 April 2011 16:27, Ajasja Ljubetińć <ajasja.ljubetic_at_gmail.com> wrote:
>> >>
>> >> I'll try running two 1D simulations, just to see if this works better.
>> >
>> > It's the angle colvar (theta) that is blowing up.
>> > Is there anything particularly illegal in this colvar?
>> > colvar {
>> >   name theta
>> >   width 5
>> >   lowerboundary 0
>> >   upperboundary 180
>> >   lowerWallConstant 10
>> >   upperWallConstant 10
>> >
>> >   extendedLagrangian on
>> >
>> >   angle {
>> >
>> >     group1 { dummyAtom (3.283, -0.213, 99.366) } ; #100 Ang above CB
>> >     group2 { atomnumbers 204 } ;#CB
>> >     group3 { atomnumbers 173 } ;#O1
>> >   }
>> > }
>> >
>> > Thank you & best regards,
>> >
>> > Ajasja
>
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 05:23:53 CST