[gmx-users] about thermal intergation calculations

[mircial at sjtu.edu.cn]

Hi All:

I am trying to do free energy calculations by thermal integration method and I have read the tutorials on the gromacs web site, they are really nice tutorials and help me a lot. But I still have two questions here:

1, After we define the necessary paramteres in the mdp file (such as the lambda values), when we run MD simulation using these mdp files, we get a file (default name dhdl.xvg) of the derivatives  of Hamilton relative to the lambda values (the dH/dLAMBDA) as a function of simulation time, and the free energy difference is the integration of this value, right?
My question is, how does this value calculated? Since in the simulations, we modified some interactions by using a specific lambda value, I can understand how to calculate the Hamilton or energy of the system under this specific lambda value, but how does the derivatives of Hamilton calculated?

2, General we use different lambda to modify the vdw or electrostatic interactions between ligand and it's environment, i understand how this work in gromacs. However, I am planning to restrain distances between specific atoms and use different lambda to change the equilibrium distances between these atoms. Is there any code implemented in gromacs to do this? Or I need to run simulations at each lambda value and calculated the potential energy as well as the derivatives and integrations by myself?

Thanks very much for your help!

RXG