Can this be used for astrophysical data analysis?

Yes, it is designed to integrate with tools like Comrade.jl for applications such as Event Horizon Telescope black hole imaging.

How does this skill handle random number generation?

It utilizes SplittableRandoms.jl to create independent, deterministic random streams that can be split across workers without losing reproducibility or statistical integrity.

What is Strong Parallelism Invariance (SPI)?

SPI is a property where a parallel algorithm produces bit-for-bit identical results regardless of the number of parallel workers or threads utilized, provided the initial seed is the same.

Which MCMC explorers are supported?

The skill supports various explorers including SliceSampler for robust general-purpose sampling and AutoMALA for high-dimensional, auto-tuned Hamiltonian Monte Carlo.

Deterministic Parallel Tempering (Pigeons SPI)

Name: Deterministic Parallel Tempering (Pigeons SPI)
Author: plurigrid

byplurigrid

•

Data Science & ML

Implements Strong Parallelism Invariance (SPI) for reproducible parallel tempering MCMC sampling regardless of worker count.

The pigeons-spi skill enables Claude to implement and manage sophisticated Markov Chain Monte Carlo (MCMC) workflows using the Pigeons.jl framework. Its core strength lies in Strong Parallelism Invariance (SPI), ensuring that a single seed produces identical results whether running on one CPU core or a thousand. This skill is essential for scientific reproducibility, high-dimensional Bayesian inference, and complex modeling tasks where deterministic behavior across varying hardware environments is critical. It integrates advanced explorers like SliceSampler and AutoMALA while providing unique triadic diagnostics and color-coded visualizations via the broader topological chemputer ecosystem.

Key Features

01Strong Parallelism Invariance (SPI) for 100% reproducible sampling

02GF(3) triadic pattern analysis for sample/constraint conservation

038 GitHub stars

04Deterministic splittable random streams via SplittableRandoms.jl

05Advanced Parallel Tempering MCMC with auto-tuned explorers

06LogDensityProblems interface integration for custom target models

Use Cases

01Scientific research requiring exact reproducibility across different compute clusters

02Bayesian inference where deterministic color-coded diagnostics aid in chain analysis

03High-dimensional astrophysical modeling such as black hole imaging with Comrade.jl

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add plurigrid/asi pigeons-spi

For use in Claude.ai and ChatGPT

Download Skill