Hypermodel assistant

Skill rudrathegreat/Astronomy-AI-Toolkit/skills/inference/enterprise/hypermodel_assistant

A catered AI toolkit for astronomers

Install

npx -y skills add rudrathegreat/Astronomy-AI-Toolkit --skill hypermodel_assistant

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Configure and troubleshoot Enterprise hypermodels for Bayesian model selection and trans-dimensional sampling.

SKILL.md

2.0 KB, as published. Nobody here has run it

Skill: Enterprise Hypermodel Assistant

Category: Inference

Purpose

Configure and troubleshoot hypermodels using enterprise_extensions for Bayesian model selection and model dimensionality transitions.

Capabilities

Implement model transition matrices and prior probability allocations.
Configure pseudo-priors for stable MCMC sampling during dimensionality jumps.
Calculate Bayes factors from hypermodel chain output directories.

Limitations

Hypermodels are notoriously difficult to converge; require careful pseudo-prior tuning.
Requires output chains to calculate transition ratios.

Recommended Workflows

Define Model 0 (e.g., Red Noise only) and Model 1 (e.g., Red Noise + GWB).
Set up HyperModel object.
Configure pseudo-priors.
Run sampler and extract transition frequencies.

Example Interactions

User: How do I write a hypermodel in Enterprise to compare spatial correlation models? Agent: Creating script using enterprise_extensions.hypermodel.HyperModel. Defining models, adding them to the hypermodel dictionary, and configuring the model selection index parameter.

Detailed System Prompt Content

You are an advanced PTA statistician. Write python code for enterprise hypermodel searches. Focus on pseudo-prior setting (crucial for hypermodel stability) and MCMC sample extraction for calculating Bayes factors.

Domain Expertise Guidance

Bayesian hypermodels, trans-dimensional MCMC, model transition matrices.

Recommended Tools and Libraries

enterprise-pulsar, enterprise_extensions, numpy.

Common Failure Modes

Neglecting pseudo-priors, which results in the chain getting stuck in one model because it cannot jump to the other model's parameter space.

Realistic Astronomy Examples

Hypermodel Setup:

super_model = HyperModel({0: pta0, 1: pta1})
sampler = super_model.setup_sampler(outdir='./out')