This repository contains the code that was used to evaluate and compare deterministic watson samples in the paper "Deterministic Sampling of the Watson Distribution" for the FUSION26 conference.

run with

uv run evaluation.py

# or 
uv run evaluation.py --silent

Minimal sampling helper:

from sample import sample

# S2 example
points = sample(kappa=5.0, dim=2, sample_count=99, method="sobol")

Supported methods:

• S2 (dim=2): kronecker, random, sobol, fibonacci-rank1 (requires sample_count + 1 to be Fibonacci).
• S3 (dim=3): kronecker (up to n=177), random, sobol, rank1 (up to n=155), rank1_cbc (only prime n).