Wall combined with perturbed energy results in zero interaction for state B
For the perturbed energy, the state A is defined as (1-lambda) whereas B is set to
0 and correspondingly the wall interaction for type B is switched off.
I guess the correct expression for the B-state should be lamfac = lambda?
The attached patch against release-4-6 proposes this solution and seems to fix my troubles when performing thermodynamic integration with pbc=xy and nwalls=2.
#1 Updated by Alexander Schlaich about 3 years ago
Actually, I should add that the total interaction with the wall is accounted for in the derivative of the perturbed Hamiltonian, i.e. in the current implementation state A mimicks a system with the wall present and state B the system without any walls.
I'm not sure if this is the intended behaviour, anyway I did not find something like this in the documentation (maybe I overlooked it, though).
The trouble however starts if you want to determine the free energy of solvation of a particle enclosed between two walls using e.g. thermodynamic integration/BAR/...
Thanks for any feedback!