![]() Essentially, the dE column is equivalent to elec(l+dl) - elec(l) added to vdW(l+dl) - vdW(l). I don't know if the sum of Elec and vdW is the full change in energy.Ĭorrect. Yes, that should be the case for both statements. In my notation then you would apparently have Elec and vdW evaluated at lambda0 (l) and lambda1 (l+dl). " I don't have an additional set of lambda values other than these. You can see that new windows are declared by "#NEW FEP WINDOW: LAMBDA SET TO. Here's a web example of a NAMD output file. NAMD joins the different lambda windows in its standard routine. If you have one output file at lambda=0, one at lambda=0.5, and one at lambda=1 then your parser needs to read all the data files and join them together in one big DataFrame. I am looking at the data (the following needs PR alchemistry/alchemtest#29 to work): The columns are the evaluations of the Hamiltonian (or the potential energy U) at other lambdas (sometimes called "foreign lambdas"): take positions sampled at fep-lambda lambda0, x_lambda0 and evaluate the energy function U(x, lambda=lambda1) with the lambda set to foreign lambda1: U(x=x_lambda0(t), lambda=lambda1) - this gives you row lambda0 with column lambda1 (at time t). The timesteps (or trajectory "frames") are each tagged with the lambda at which they were sample, i.e., data from the lambda=0.5 file will all have fep-lambda 0.5. The "fep-lambda" index states at which lambda this particular frame was sampled. The DataFrame should contain all data needed to run BAR/MBAR or one of the FEP estimators (which we don't have implemented yet). My understanding of the above was that the different columns represent the lambda value of sampling, but that would mean the fep-lambda column is doing something else since it has values of both 0 and 1 in the example in the docs. For datasets that sample only a single λ parameter, then the DataFrame will feature only a single index in addition to time, with the values of λ for which reduced potentials were recorded given as column labels.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |