I was doing adsorption energy calculation of molecules on iron ore surface. During optimization of molecule + surface system, I noticed that energy change was very slow after ~300000 steps and gnorm values were repetitive. So i changed option stepmx from 1.0 to 0.1 and then whole system converged within 200 steps.
My question is am I chosing right option or a different option will be helpful.
I am using ReaxFF and other option values were set to default.
If the criteria for optimisation are met then you should be at a stationary point. Optimisers can get stuck if there are discontinuities in the energy (which can happen with ReaxFF) and so lowering the step size is fine. Only for a harmonic system with an exact Hessian would you be likely to move 1.0 in a single step.