Dear Dmitri,
Many thanks for this. It makes sense to me, except for the last part "larger than optimal for individual case". What do you mean by this (individual case).
Best regards
George
On Mar 13, 2012, at 8:19 PM, Dmitry Mukha wrote:
> Dear George,
>
> nonetheless, if you take snapshots with double frequency comparing to the
> optimal one, you will just have to calculate more. But you will not get any
> bias in your results. Also some parameters may appear to have periodic
> fluctuations so approach using autocorrelation will fail. The logic is to
> find 'convergence' when addition of new snapshot doesn't shift the result
> significantly. And if you want to use your own scripts you have to ensure
> that every calculation will be done with sufficient number of snapshots,
> larger that optimal for individual case.
>
> 2012/3/13 George Tzotzos <gtzotzos.me.com>
>
>> Thanks Aron,
>>
>> It makes sense. At the same time, the answer was given in the paper I
>> quoted (which I did not read carefully enough).
>>
>> In this the authors took 150 adjacent snapshots from 3ns production runs.
>> The authors quote that the sample points should be "nearly uncorrelated,
>> since the correlation times for relaxation of effective energy fluctuations
>> (1ps) are shorter than the time-interval (20ps) used for snapshot
>> extraction".
>>
>> Best regards
>>
>> George
>>
> --
> Sincerely,
> Dmitry Mukha
> Institute of Bioorganic Chemistry, NAS, Minsk, Belarus
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Mar 13 2012 - 12:30:04 PDT
