relation between time step used and nfreq value in ave/correlate

hi all,

I have a doubt regarding the nfreq used in ave/correlate command. I am calculating thermal conductivity of Ar Cu nanofluid system using Green - Kubo method. If I am using very small timestep like 0.01 fs, whether I should change Nfeq in ave/correlate command to a very higher value? correlation decay approaches zero anyway when I tried with same nfreq (2000) for both, means 0.01fs and 10fs. Plese anyone could help…

yes if you use a smaller tilmestep, you should in general
compute your correlations at longer tilmestep (same delta time)
intervals. What is best for your model is something you’ll
have to expt with.

Steve