bayesloop 2.0.0

The first major release in nearly a decade — a full modernization of bayesloop.

Highlights

• New snake_case API across the entire library (load_data, set_observation_model, get_parameter_distribution, …)
• Drop-in backwards compatibility: import bayesloop.v1compat runs existing 1.x scripts unchanged, emitting deprecation warnings that point to the new names — it also loads study files saved with 1.x
• Faster fits through adaptive likelihood caching and optimized built-in observation models
• Parallel hyper-studies: HyperStudy and ChangepointStudy fan out across cores via joblib (now a core dependency, replacing the optional pathos)
• Modern packaging: src layout, pyproject.toml, Python 3.10–3.13
• Refreshed documentation notebooks, and a new homepage at bayesloop.com with an in-browser playground

Breaking changes

• All camelCase names (methods, attributes, keyword arguments, module names) are snake_case now — or use bayesloop.v1compat
• The probability Parser and Study.eval() were removed
• Default parameter names of some transition models changed: 'tChange' → 't_change', 'tBreak' → 't_break', 'log10pMin' → 'log10p_min'
• Python ≥ 3.10 required

Full Changelog: 1.5.7...2.0.0