I recently tried GROMACS tpi module to calculate Henry's law constant for a couple of molecules in Poly(ethylene). As I found a great inconsistency between my results and those reported before, I started to investigate the reason(s). This led to a weird contrary between output files, namely time dependent and distribution results.

Results for Argon-PE system at 1 atm and 480 K has been attached. 1.xvg is time dependent results and dist1.xvg is the probability distribution for energy.

1) The term U in the time dependent file (1.xvg) is supposed to be the DeltaE of insertion. But it seems that it is representing the chemical potential (Mu=-DeltaE), which is negative of DeltaE. I am still struggling on this. which one is correct?

2)If we assume that U is representing the DeltaE: we can calculate DeltaE (or U) as a function of time using 1.xvg. averaging the value results in 5.3577 (a positive number). But if you look at dist1.xvg for the same system, the weighted probability distribution results in a negative number for U (or DeltaE) which is -11.64. I think question (1) explains the difference in sign of these numbers. The value obtained from 1.xvg should be multiplied with a (-). But how come the absolute value is different?

3) using the direct energy probability distribution results in a very big positive number which is not compatible with calculated Delta E.