Re: OPES giving -nan entries with eABF

[Snow Summer]

If you are using Colvars, please post your issue on https://github.com/Colvars/colvars/issues. OPES is not the same as metadynamics as it has its own estimator which is different from metadynamics.

On Monday, June 9, 2025 at 11:15:44 AM UTC-5 Gourav Chakraborty wrote:

I am trying to use OPES in conjuction with eABF. Just trying to replicate a work where MetaDynamics is merged with eABF to accelerate sampling efficiency, but instead of Meta, I am using OPES. I am using COLVARS with NAMD3, as OPES is present there. Individually both eABF are OPES are working fine. But when I coupled the two, after a point I got -nan entries all over my kernels.dat file. How to avoid this? I tried everything, lowering barrier, using higher/lower hillsfrequency, using opes on only one of the two CVs' adding/removing harmonic walls, but nothing worked. It is working for some time, but eventually throwing -nan values. My question is, if metadynamics can work with eABF then so does OPES, right?