How to test convergence of ABF simlation

[sathishdasari]

Hi, I am running an Adaptive Biasing Force (ABF) simulation of alanine dipeptide in water with two walkers. I can plot the final free energy profile after simulation is over but I do not know how to test the convergence of the ABF method with simulation time. Is there any command or code available like in plumed software where you can plot the free energy difference between two minima.

Thanks

[mquevill]

The "output_frequency" parameter sets how often a histogram is output from SSAGES. While we do not necessarily have a feature to determine convergence automatically, it should be relatively easy to script something to determine convergence based on the outputs from ABF over time.

[sathishdasari]

Hello Sir, In continuation of the previous question, If I want to plot a free energy profile at different time intervals how can I do that? Suppose I ran a 250 ns total simulation time and I want to plot free energy profile at time intervals of each 50 ns like 50 ns, 100 ns, 150 ns, 200 ns.

[mquevill]

A simple way of performing this would be to run successive 50 ns simulations. Make sure that the restart: True flag is turned on (in JSON) to continue the ABF from where it left off. As long as you use different filenames (by setting "output_file", "Nworld_output_file", and"Fworld_output_file"), you should be able to get the successive analysis that you are wanting.

If you wanted to be able to do this programmatically, you could modify the ABF::WriteData() function. This is not a functionality that is currently in ABF, but it could be a future enhancement of the method.