[gmx-users] Velocity autocorrelation

[David van der Spoel]

On 2012-04-01 20:05, Ignacio Fernández Galván wrote:
> --- On Sun, 1/4/12, David van der Spoel<spoel at xray.bmc.uu.se>  wrote:
>
>> It depends what you want to compare to.
>
> Fair enough. For the moment I don't want to compare to anything. I just want to extract a number of uncorrelated configurations from the simulation, and to know what's the minimum interval at which I should sample configurations.
>
> Something like what's explained in pp. 164-ff of this book:
> http://books.google.es/books?id=GT9oPnAEkEQC
>
>> The relaxation time of a
>> velocity averaged per atom can be analyze by fitting the vac
>> to y = exp(-t/tau). This gives 5 fs in your graph.
>
> 5 fs? Don't you mean 5 ps? In that case, how do arrive to such value? It can't be a simple exponential fit, because the function is oscillating. Do you just guess it by the way the amplitude shrinks?
>
> This is my main problem, I don't see how one can fit an exponential to an oscillating function.
>
Of course you can, but what does it mean?

Indeed if I take the absolute value of the function I get 150 fs.
No time to read your book, but I think the data is fine.

> Thank you,
> Ignacio


-- 
David van der Spoel, Ph.D., Professor of Biology
Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205.
spoel at xray.bmc.uu.se    http://folding.bmc.uu.se