[CP2K:1870] Re: methane in a water box with FIST

[Fawzi Mohamed]

On 12-mar-09, at 18:28, Axel wrote:

> On Mar 12, 1:13 pm, Fawzi Mohamed <fa... at gmx.ch> wrote:
>> On 12-mar-09, at 18:03, Axel wrote:
>>> On Mar 12, 11:39 am, Teodoro Laino <teodor... at gmail.com> wrote:
>>>> This is not possible (NPT ensemble supports only as possible
>>>> thermostats
>>>> NOSE and CVSR).
>>>> I agree with you that LANGEVIN is just another kind of thermostat  
>>>> and
>>>> should stay between the
>>>> other thermostats. But it was developed as a different ensemble..
>>>> maybe
>>>> one day it will disappear
>>>> and reappear between the thermostats.. For the time being  I would
>>>> recommend you for this case
>>>> the CSVR . It's quite effective (have a look at the original  
>>>> paper).
>>
>>> marius,
>>
>>> two comments.
>>
>>> if i understood the CSVR algorithm correctly, it is actually
>>> not very different from using langevin dynamics with a small  
>>> friction
>>> coefficient (which is what people using NAMD typically do, for
>>> example).
>>
>> the main difference is that it influences only the total kinetic
>> energy, (it does a langevin dynamics only for that).
>> It has many nice properties, but does not act differently on  
>> different
>> parts of the systems and conserves the rotational moment.
>
> how about using it with REGION set to MASSIVE?
> you'd have one CSVR per DOF...

yes indeed that is close to real langevin dynamics

> cheers,
>    axel.
>
>> Langevin dynamics is applied on all the particles, and so it might be
>> more efficient at equilibrating imbalanced energy distribution within
>> the system (you don't have to count only on energy transport in the
>> system to equilibrate them).
>> On the other hand it influences the system in a more global way and
>> does not preserve the rotational moment.
>>
>> ciao
>> Fawzi
>