pub fn draw_batch(
input: PolyaGammaBatchInput<'_>,
) -> Result<Array1<f64>, String>Expand description
Top-level entry point: dispatches to GPU when available, otherwise CPU.
Both paths use the same per-row XORWOW seeding so the GPU result is a
bit-equivalent of the CPU result up to IEEE rounding of exp/log/
tan/tanh/sqrt (which the device evaluators round to within 1 ULP
of the CPU libm).