Comparison of free energy changes upon particle removal in the (a) all-particle and (b) implicit solvent, density dependent cases. The density dependent potential introduces a secondary free energy change due to the change in energy models associated with a change in global average solute density of the system.
Comparison of the RDF generated using the converged potential for (open circles) and the target RDF (solid line).
Pairwise potentials that reproduce the target RDF across a range of global solute densities, . Bare LJ potential shown for comparison as dark solid line. (a) Solute densities below the total particle density [(—) ; (⋯) ; (---) ; (-⋅-⋅) ]. (b) Solute densities above the total particle density [(—) ; (⋯) ].
Self-interaction energy, , for , .
Comparison of measured (open circles) and target (solid line) RDFs at for reduced temperatures: (a) , (b) , (c) , and (d) . (a) and (b) are below the fitting temperature; (c) and (d) above.
Comparison of measured (open circles) and target (solid line) RDFs at for system volumes: (a) , (b) , (c) , and (d) . (a) and (b) are more dense than the fitting density; (c) and (d) less dense.
Excess chemical potential as a function of system temperature (circles, all-particle simulation results; squares, ; crosses, ).
Comparison of simulated values with target value as a function of (circles, original simulation using global density values; squares, after optimization; solid line, all-particle simulation; dashed lines, desired simulation accuracy).
Comparison of simulated RDF with worst fit (open circles) and the all-particle RDF (solid line).
Comparison of simulated values with the exact value for the all-particle simulation as a function of for (circles, simulation data; solid line, all-particle simulation; dashed lines indicate an error of ).
Solute fraction as a function of distance from center of the box [(—) all-atom; (---) local density approach; (⋯) global density approach)].
Test system properties.
Transferability of the original effective potential, as measured by the error in the per-particle energy with changing temperature.
Transferability of the original effective potential, as measured by the error in the per-particle energy with changing density.
Errors produced by use of a local density dependent potential. For the case , errors are below the specified tolerance of 1%. However, a smaller cutoff radius introduces significant errors in the per-particle energy.
Article metrics loading...
Full text loading...