I am currently working on implementing MBAR with just the single state point (near the critical point).
This plot shows the squared difference between the weights of the liquid and vapor phases for the initial simulation:
From this figure, a cut-off in N around 81 gives equal area under the peaks.
This is reasonable considering the 1-D histogram plot:
Previously, my Ncut was fixed at 75. I think I just visually selected this value. It would probably be best to find the point that introduces the least uncertainty.
However, we need to recall that the dividing surface should also depend on U:
With the Ncut of 81, this plot shows the squared difference of weights at T1 = 500 K (10 K less):
Recall that Potoff reported an equilibrium line around -4100 K. So this appears to be working quite well.
This plot shows what the expected value of N in the liquid and vapor phases are as a function of mu (T=500 K).
And this shows the predicted rho:
This is at T=480 K:
Note that previously my optimal mu for 480 K was -4023 K. These densities appear to be pretty good too.
I am currently working on implementing MBAR with just the single state point (near the critical point).
This plot shows the squared difference between the weights of the liquid and vapor phases for the initial simulation:
From this figure, a cut-off in N around 81 gives equal area under the peaks.
This is reasonable considering the 1-D histogram plot:
Previously, my Ncut was fixed at 75. I think I just visually selected this value. It would probably be best to find the point that introduces the least uncertainty.
However, we need to recall that the dividing surface should also depend on U:
With the Ncut of 81, this plot shows the squared difference of weights at T1 = 500 K (10 K less):
Recall that Potoff reported an equilibrium line around -4100 K. So this appears to be working quite well.
This plot shows what the expected value of N in the liquid and vapor phases are as a function of mu (T=500 K).
And this shows the predicted rho:
This is at T=480 K:
Note that previously my optimal mu for 480 K was -4023 K. These densities appear to be pretty good too.