I am new to SMD (Steered Molecular Dynamics) and trying to calculate PMF. Using lammps colvar package. After simulations i can see trajectories coming out along with the accumulated work. But i have a doubt about the accumulated work. Is it the net work done of the entire system or the work done due to the movement of spring (manual says its work in moving the bias). Further how can we use jarzynski equality to calculate PMF from these work values?
I don't use LAMMPS, but I suspect that the work is the cumulative work done by the biasing potential as it moves along the reaction coordinate.
The Jarzynski equation equates the exponential of the free energy with the average exponential of the works:
exp(-beta*G(x)) =
Many years ago I wrote a program to compute a PMF from a file with a set of work profiles (actually the first code I ever wrote). The program takes a file of N+1 columns (where N is the number of SMD simulations you did). The first column is the location of the biasing potential on the PMF and each of the next columns is the value of the work for each of the N SMD simulations (so it's a simple table). It then spits out a file with the PMF.
You can find this code in the link I provided below. Compile it using gfortran or some other Fortran compiler (e.g., via "gfortran SMD_to_FE.f90"). The program is short and simple, so you should be able to grok how the FE is calculated and how to use the program just from the source code.
https://github.com/swails/JmsScripts/blob/master/SMD_to_FE.f90