VegasHistogram¶

class madspace.VegasHistogram(self: madspace._madspace_py.VegasHistogram, dimension: SupportsInt, bin_count: SupportsInt)¶

Bases: FunctionGenerator

Per-bin weight accumulator that adapts a VegasMapping grid.

During the warm-up run it sums the sample weights falling in each bin of each dimension. The accumulated values and counts are then used to refine the VEGAS grid (Sec. 3.2.2 of [1]).

batch is the leading batch dimension.

Arguments

latent – float, shape (batch, dimension) – the sampled coordinates.
weights – float, shape (batch,) – the per-sample weights.

Returns

values – float, shape (dimension, bin_count) – summed weight per bin.
counts – int, shape (dimension, bin_count) – sample count per bin.

References

[1] T. Heimel, O. Mattelaer, R. Winterhalder, “MadSpace”, https://arxiv.org/abs/2602.06895 (Sec. 3.2.2)

Parameters:

dimension – Number of dimensions.
bin_count – Number of grid bins per dimension.