On Fri, May 17, 2013, Daniel Roe wrote:
> If you want the average
> velocity of all atoms in a residue (assuming you are reading in a
> velocity trajectory instead of coordinates) you would need something
> like this for every residue:
>
> vector res1V center :1
> vector res2V center :2
> ...
>
>
> Once you have the vectors you can calculate the autocorrelation
> function for each vector with 'timecorr'.
We really need to change the names of some of these commands, which go way
back to prehistory. In particular, the 'timecorr' functions computes angular
averages needed for certain types of spectroscopy: you get the time-decay
of the cosine of the angle between a vector at time 0 and at time t (or the
square of the cosine for the P2 function). This is not the sort of velocity
autocorrelation function that is usually considered (e.g. to look at
diffusion), where you want more than just the angular part.
I don't think there is any current way to compute the "usual" velocity
autocorrelation functions within cpptraj. It would not be hard to add, but we
really should re-consider the user interface -- e.g. always compute a scalar
time series, then pass everything to "corr" itself. Right now the "timecorr"
function both computes the scalar and its time correlation function in a
single step.
[The next step is to have a nice interface to fit the output of the "corr"
function to a sum of exponential functions; this is something I'm hoping that
George and Dan can work on together.]
...dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri May 17 2013 - 11:30:05 PDT
