[ESPResSo-devel] Fwd: Re: Implementing Lowe-Andersen thermostat

[Kai Szuttor]

-------- Weitergeleitete Nachricht --------
Betreff: Re: [ESPResSo-devel] Implementing Lowe-Andersen thermostat
Datum: Sat, 28 Feb 2015 14:22:41 +0100
Von: Kai Szuttor <address@hidden>
An: Axel Arnold <address@hidden>

Hi Axel,

yes i actually took that into account, but still my energy is diverging.
Is there any other C-core function which could cause another scaling
of the velocities i am changing in my function?

Kai

Am 28.02.15 um 10:28 schrieb Axel Arnold:
> Hi Kai,
>
> Did you consider that In the C-core velocities are scaled by the
> time step?
>
> Axel
>
> Dr. Axel Arnold Email: address@hidden
>
>> Am 27.02.2015 um 23:39 schrieb Kai Szuttor
>> <address@hidden>:
>>
> Hi everyone,
>
> currently i´m trying to implement the Lowe-Andersen thermostat (C.
> P. Lowe 1999 Europhys. Lett. 47 145 doi:10.1209/epl/i1999-00365-x)
> in the Espresso source code and having some issues i wasn´t able to
> resolve. So far i wrote a function that performs the thermalization
> of the particles relative velocities. This function is called by
> the function "integrate_vv" in integrate.cpp just after the GHMC
> thermostat is doing its ghmc_momentum_update()-call. I thought that
> would a good idea since the Lowe-Andersen thermostat is also only
> changing particles velocities and not the forces acting on them.
> The problem i'm having now is that the energy grows to infinity and
> i was not able to find out why. To check for the algorithm itself i
> implemented the thermostat on the tcl-level and it seems to work,
> so the algorithm itself should do the right thing. Is there
> anything i should know about espressos internal structure regarding
> the velocity handling? Maybe someone can give me a hint.
>
> Thanks in advance.
>
> Kai Szuttor Insitute for Computational Physics University of
> Stuttgart