Thanks for the suggestions. I will try different schemes to see if the results are consistent. If using D_pbc vs 1/L to compute D and eta, is there any special reason for the NVE ensemble? Or can NVT be adopted?

Have a nice day.

Jeff