histogram2d_op

→vector<TensorImplPtr>

int histogram2d_op(const int & a, const int & b, int bins_a, int bins_b, double lo_a, double hi_a, double lo_b, double hi_b, bool density)

source edit

Python wrappers

pylucid.ops.composite.histogram2dfunction

Compute a 2-D histogram from two equal-length 1-D tensors.

Treats paired observations (a_k, b_k) as samples on an $\mathrm{bins\_a} \times \mathrm{bins\_b}$ grid of uniform rectangular cells and counts the number of pairs falling in each cell. When density is true each cell value is divided by (step_a * step_b * N).

See Also

histogram_op : 1-D variant. histogramdd_op : General N-dimensional variant.

Parameters

aTensorImplPtr

1-D tensor of x-coordinates.

bTensorImplPtr

1-D tensor of y-coordinates. Must match a in length, dtype, and device.

bins_aint64

Number of bins along the x-axis.

bins_bint64

Number of bins along the y-axis.

lo_a, hi_adouble

Range of the x-axis.

lo_b, hi_bdouble

Range of the y-axis.

densitybool

If true, normalise to a probability density.

Returns

vector<TensorImplPtr>

[counts, edges] where counts has shape (bins_a, bins_b) and edges has shape (bins_a + 1 + bins_b + 1,) — the x-edges followed by the y-edges concatenated. Both dtype F64.

Notes

Non-differentiable. Executes on the CPU stream regardless of input device.