Understanding Output Files and Free Energy Derivatives in CP2K Alchemical Change Tutorial

[ice ZQing]

  Hi everyone,
I'm currently learning to perform molecular simulations with CP2K, and I’ve been following the official Alchemical Change (AC) tutorial, located at:  

  cp2k/tests/FE/regtest-1/  or cp2k/tests/FE/regtest-1 at master · cp2k/cp2k  

Using CP2K version 2022.2

  The tutorial simulates an alchemical transformation between two molecular fragments in water, where the main difference is the presence or absence of a hydrogen atom, likely in an alanine residue (ACE–ALA–NME) solvated by water. I try this simulation using the GENMIX method and runs a trajectory of 20,000 steps at λ = 0.5 with classical force fields (FIST).  

  I have a few questions regarding the output and how to process the results:

What does these Output files mean and which are important for free energy analysis? After the run, I get the following files:

Solv_alch_chng-1.ener
Solv_alch_chng-1.restart
Solv_alch_chng-mix-1.ener
Solv_alch_chng-pos-1.pdb
Solv_alch_chng-r-1.out
Solv_alch_chng-r-2.out

    I assume:

*.ener contains energy data,
*.pdb is trajectory,
*.restart is for restarting.

But I’m not sure what the -r-1.out, -r-2.out, and -mix-1.ener files are for, or what physical quantities they contain. Also, the .ener files contain only a few columns of data, with no clear labels, so I’m not sure how to interpret them.

Which of these output files are necessary for post-processing the free energy difference? And what is the best way to extract or average relevant observables like free energy gradients?

 I’m happy to share my input files if needed.  

[ice ZQing]

 Has anyone noticed this?

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/c2c8824d-ce60-4fa6-9a64-bab237f3af4cn%40googlegroups.com.

[Marcella Iannuzzi]

Hi ...

In this type of simulation, two calculations run in parallel to get the energy E1 and E2 (and forces as well). This is the reason why you get the additional out file, one for each independent calculation. The results are mixed as \lambda*E1+(1-\lambda)*E2. I suppose the mix energy file corresponds to that, it should be easy to check.

Regards

Marcella