I need to confirm the convergence of REMD simulation within 200 ns per replica so how can I calculate RMSD values with their corresponding free energy in equal-length non-overlapping time windows like the following figure?
Convergence can be accessed by various things. Most easily, taking a cumulative average of a single observable, such as RMSD. But, you can also compare histograms (or free energy landscapes) for different fractions of your trajectories. The best prove however, is having two independent REMD simulations and comparing their outcome.