wasserstein-audit

Scan manuscripts and Python code for Wasserstein-1 versus Wasserstein-2 metric inconsistencies.

1|Updated Dec 13, 2025
One-click install
npx skills add https://github.com/ZK-Theory/TDL --skill wasserstein-audit-zk-theory
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: wasserstein-audit
Source: https://github.com/ZK-Theory/TDL/tree/main/.agents/skills/wasserstein-audit
Command: npx skills add https://github.com/ZK-Theory/TDL --skill wasserstein-audit-zk-theory

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Research projects that mandate Wasserstein-2 as the primary persistence-diagram metric often accumulate inconsistent W₁, unsubscripted W, or ambiguous W_p references across manuscript drafts and code. This Skill audits both sources and produces a structured reconciliation report so notation violations are found before they block publication milestones. ## Core Features & Use Cases - Manuscript Scanning: Flags W₁ usage, unsubscripted W, bottleneck-only metrics, and ambiguous Wasserstein references in Markdown drafts under papers/. - Code Scanning: Detects Python Wasserstein calls missing an explicit p=2 or order=2 argument, including gudhi, persim, and giotto-tda library calls. - Structured Report: Cross-references findings against the canonical notation standard and outputs a reconciliation table with suggested fixes, without modifying anything automatically. - Use Case: Before submitting a topology paper, run the audit across all manuscripts and domain packages to confirm every Wasserstein reference complies with the W₂ convention and receive a blocking/non-blocking status for Phase 0. ## Quick Start Ask the assistant to run a Wasserstein audit across all papers and code to list every W₁ or ambiguous Wasserstein usage with suggested fixes.

Frequently Asked Questions about wasserstein-audit

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I check my paper for Wasserstein notation inconsistencies?

Run the audit over your manuscript Markdown files to flag W₁ usage, unsubscripted W symbols, and ambiguous Wasserstein references. The report lists each finding with file, line, issue type, and a suggested fix aligned to your notation standard.

How to find Python Wasserstein calls missing the p parameter?

The audit scans Python files for wasserstein_distance calls without an explicit p= or order= argument, since defaults vary across library versions. It specifically flags gudhi calls lacking order=2 and any explicit p=1 or order=1 usage.

Does gudhi wasserstein_distance default to Wasserstein-2?

No, gudhi.wasserstein.wasserstein_distance requires an explicit order argument and its default is not guaranteed to be 2. In contrast, persim.wasserstein_distance and giotto-tda Wasserstein vectorisation default to p=2.

Does the Wasserstein audit automatically fix the issues it finds?

No, the audit only produces a structured reconciliation report with suggested fixes and never modifies files without confirmation. This keeps manuscript wording and code changes under explicit human review.

When is Wasserstein-1 usage acceptable in a manuscript?

W₁ is acceptable in theory proofs within Methods sections when clearly labelled and distinguished from the computational metric. Everywhere else, project convention mandates Wasserstein-2 as the primary persistence-diagram metric.