From: Aron Broom (broomsday_at_gmail.com)
Date: Thu Jun 20 2013 - 16:22:34 CDT
Yeah I think what you have works. What you're doing in essence is
converting all the points from energy to probability and then summing for
each value of J, then converting back to energy.
You may also have to make a trivial adjustment to get back to having an
actual 0 energy value, if you care about looks.
On Thu, Jun 20, 2013 at 4:28 PM, Kasra Fattah <kasra.fattah_at_gmail.com>wrote:
> Hi all,
> I'm calculation the pmf of a system using ABF method. One I did it for a
> 1D collective variable that is the distanceZ of a molecule from a specific
> point along y-direction. Next I'm doing a 2d pmf calculation using the
> collective variable along both x and y direction and I get a surface of
> pmf. Now to test if the results from 2d are consistent with my 1d results:
> Is the following procedure right?
> I average the results while keeping one of the reaction coordinates
> constant (eg. at a constant y) with respect to the Boltzmann weight that
> is, if I have pmf(i,j) then doing:
> - <pmf ( 1:imax , J)> with the weight of
> - exp(-beta * pmf( 1:imax , J) )/ integral ( exp(-beta * pmf( 1:imax ,
> J) ))
> and doing this for all the points along y (incrementing j) will give a pmf
> for along the y reaction coordinate so I can compare it with the 1D case of
> finding the pmf from ABF.
> I hope I was clear enough please let me know if it's not clear. I'd
> appreciate any comment and help on this.
-- Aron Broom M.Sc PhD Student Department of Chemistry University of Waterloo
This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:23:21 CST