# Notices

## Textbook and Blueprint website

- Sinho Chewi, *Log-Concave Sampling*
  - Source: https://chewisinho.github.io/main.pdf
  - Role: primary mathematical textbook reconstructed by ASTIS.
  - Copyright boundary: the public draft exposes no explicit license for
    wholesale republication. ASTIS uses original faithful paraphrase, precise
    source correspondence, and separately identified supplemental
    derivations/formalizations. Sinho Chewi does not participate in, endorse,
    or maintain ASTIS.
- Sho Sonoda, Lean-Ridgelet
  - Repository: https://github.com/shosonoda/lean-ridgelet
  - Blueprint:
    https://shosonoda.github.io/lean-ridgelet/blueprint/html-multi/overview/#Lean-Ridgelet-Blueprint--L2-theory___-arXiv___2106___04770v2-implementation-map
  - License: Apache-2.0.
  - Role: inspiration for the Blueprint-style organization and implementation
    map. ASTIS copied no Lean-Ridgelet source code, template, or style.
    Sho Sonoda does not participate in, endorse, or maintain ASTIS.

## Automation workflow

This project includes an automation workflow inspired by:

- `DakeBU/Quantum-Computing-Block-Encoding`
  - Reference branch: https://github.com/DakeBU/Quantum-Computing-Block-Encoding/tree/wip/ghl2025-faithful-20260518-0201
  - Role: plain-file agent orchestration, six-hour proof loops, conversion
    windows, proof-obligation ledgers, and reviewer gates.
- `wanshuiyin/Auto-claude-code-research-in-sleep`
  - Repository: https://github.com/wanshuiyin/Auto-claude-code-research-in-sleep
  - Role: autonomous research-loop framing and adversarial multi-agent review.
- Jiayi Weng, Learning Beyond Gradients
  - Article: https://trinkle23897.github.io/learning-beyond-gradients/
  - Artifact repository: https://github.com/Trinkle23897/learning-beyond-gradients
  - Role: append-only trial memory, failure memory, and search-system
    maintenance patterns.
- `math-ai-org/mathcode`
  - Repository: https://github.com/math-ai-org/mathcode
  - Role: Lean proof diagnostics, hidden-assumption scans, theorem-reuse
    memory, tree-of-subgoals decomposition, proof-statistics review, and
    skills/tools/plugins as proof-workflow concepts.
  - Note: the local checkout inspected during development did not include a
    top-level license file, so ASTIS records citation/source links and does
    not reuse MathCode source code.
- `YuanheZ/lean-stat-learning-theory`
  - Repository: https://github.com/YuanheZ/lean-stat-learning-theory
  - Article: https://arxiv.org/abs/2602.02285
  - Role: nearby Mathlib-based formalization reference for probability,
    concentration, log-Sobolev/Poincare, entropy duality, and one-step
    discretization results.