allotax 0.3.0

Fast allotaxonometer computation powered by Rust

allotax

Python bindings for allotax-core — fast allotaxonometer computation powered by Rust via PyO3.

Install

pip install allotax

Usage

import allotax

sys1 = {"types": ["word_a", "word_b", "word_c"], "counts": [100.0, 50.0, 25.0]}
sys2 = {"types": ["word_b", "word_c", "word_d"], "counts": [80.0, 40.0, 20.0]}

# Single alpha — lean display result
result = allotax.compute_allotax(sys1, sys2, alpha=1.0)
# result keys: normalization, delta_sum, diamond_counts, max_delta_loss, wordshift, balance, alpha

# Multiple alphas at once (shared combElems step, Rayon-parallelized)
result = allotax.compute_allotax_multi_alpha(sys1, sys2, alphas=[0.5, 1.0, float("inf")])
# result keys: balance, alpha_results

# Full intermediate data (for custom downstream processing)
result = allotax.compute_allotax_full(sys1, sys2, alpha=1.0)
result = allotax.compute_allotax_multi_alpha_full(sys1, sys2, alphas=[0.5, 1.0])

API

Function	Returns
compute_allotax(sys1, sys2, alpha, wordshift_limit=200)	Lean display dict for a single α
compute_allotax_full(sys1, sys2, alpha)	Full intermediate result for a single α
compute_allotax_multi_alpha(sys1, sys2, alphas, wordshift_limit=200)	Lean display dict for multiple α values
compute_allotax_multi_alpha_full(sys1, sys2, alphas)	Full intermediate result for multiple α values
rank_turbulence_divergence(ranks1, ranks2, counts1, counts2, alpha)	RTD only (pre-combined data)

Input systems are plain dicts with "types" (list of str) and "counts" (list of float).

References

Gallagher et al. (2021). Generalized word shift graphs. EPJ Data Science, 10(1), 4.