WangLandau

class WangLandau : public feasst::Bias

Wang-Landau flat histogram bias.[1]

CriteriaWriter outputs the following:

• num_flatness: The number of times the visited states histogram was found to be flat.

• rows for each Macrostate with the following:

  • ”visited” the number of times a state has been visited since the last time this visited states histogram was reset after it was deemed to be sufficiently flat.

References:

[1]

Fugao Wang and D. P. Landau. Efficient, Multiple-Range Random Walk Algorithm to Calculate the Density of States. Physical Review Letters, 86(10):2050–2053, March 2001. doi:10.1103/PhysRevLett.86.2050.

Arguments

• min_flatness : Number of flatness checks required for completion.

• add_to_ln_probability : The initial amount to add to the natural log of the macrostate probability upon visiting that state (default: 1.0).

• reduce_ln_probability : Reduce the amount to add to the natural log of the macrostate probability by multiplcation of this factor upon reaching a sufficiently flat histogram (default: 0.5).

• flatness_threshold : The visited states histogram is determined to be flat when the percentage between minimum visisted states and average reaches this threshold (default: 0.8).

• min_visit_per_macro: The minimum number of visits for each macrostate required during flatness check (default: 10^3).