ASD Research Downloads

The Chinese and English Chapter 1/2 PDFs and AI collaboration resources are collected at /en/ai/build-your-own. Chapter PDFs support full reading and offline archiving; the AI pack transfers public website knowledge. They are separate downloads. Browsing needs neither a download nor programming.

Current articles record page-specific adoption of final combined manuscript ASD-BOOK-ddbade997b45. The four public Chapter 1/2 PDFs remain independent existing artifacts with unchanged checksums; website migration does not make them chapters or final English translations of this combined edition. The AI pack carries published website adaptations, conditions and source identities, not the complete final manuscript file.

ASD Researcher helps you locate passages, trace equations and premises, examine arguments, and carry an open problem into a research task. Reading the website requires no download.

The collaboration knowledge pack is a portable copy of public texts, equations, sources, rules and tasks for use in another AI environment. It is not a website update: unpacking this ZIP into the GitHub website project does not update the site.

The online workspace reports its actual capabilities. Retrieval and context export remain available without a connected model. Downloading the pack does not connect a model or transfer subscriptions or API credits.

First decide whether you want to read one article, resume one research session elsewhere, or build a broader reference collection. These downloads are not interchangeable.

• Collaboration knowledge pack — Download ASD_AI_Research_Pack.zip, unzip it, and select the relevant texts, rules and checks for reading or an AI. Select the ZIP in the public download center; chapter PDFs are not included in it.
• Single-page content — The MDX download on an article preserves its source. Ordinary Markdown readers may not expand MDX components. knowledge/markdown/ in the pack provide readable exports.
• Markdown / JSON research record — Export the current question, retrieved passages and notes from Researcher. This is not the whole knowledge pack; it saves this record without publishing or submitting it. Save drafts you need before refreshing.
• Website engineering update — Files under upload/ in an engineering delivery go to the corresponding GitHub paths according to that delivery’s instructions. They update website code and content; they are not the AI collaboration knowledge pack.

Preview opens the PDF without incrementing a count. Download retrieves the file directly and attempts to record one initiated request; no login or programming is needed. Files remain downloadable when statistics are unavailable.

Download counts record initiated download requests and do not represent unique users or completed transfers. Direct file access and downloads with JavaScript disabled are not counted; repeat clicks and automated requests cannot be fully excluded.

Check the language, pages, bytes and SHA-256 in File version and checksum. SHA-256 identifies whether a downloaded file matches the website artifact. A PDF file version is derived from its checksum, not a publication date or theory version. A revised PDF needs a new checksum; counts for the same resource ID accumulate across revisions.

These are manuscript chapters the author has authorized for public access; a public download does not establish formal publication. The files are provided for reading and research discussion. Follow the existing site copyright policy for citation and republication; no new open licence is granted.

Study this chapter with Researcher opens the relevant public page context. It does not upload or read a local PDF, nor does making a PDF public replace a professional module’s theory baseline. If supplying a PDF to another AI, check its file support, capacity and what it actually read.

You do not need Python or Node.js. Start with one concrete question, such as “Why do the SI definitions alone not derive LVE?”, and select material relevant to it.

1. Download and unzip the collaboration pack, read README.md, and note the snapshot_id at the top of MANIFEST.json for your question.
2. Select rules/SYSTEM_RULES.md and the relevant .md articles under knowledge/markdown/. Chinese articles use theory/, overview/ and similar folders; their English counterparts are under en/. Readable exports of the same home-page source are home.zh.md and home.en.md. For the metrology example, choose en/theory/metrology.md and en/overview/origins.md.
3. Add the selected articles through your AI tool’s file feature and supply the rules as your own collaboration instructions. Check that tool’s supported formats, capacity and actual reading result; do not assume the entire ZIP must be uploaded.
4. Copy the first question below, replacing the snapshot placeholder with the value in your manifest. Ask the AI to identify the files and passages it actually read and explicitly name anything missing.
5. Compare the answer with the original page or Markdown, then use the checks below to inspect citations and boundaries. Save the question, sources, new assumptions and next task if you continue.

Copy this first question

Use knowledge snapshot [insert snapshot_id from MANIFEST.json] as the baseline. First list the files and relevant sections you actually read; a README or URL is not evidence that you read the article. Using the Metrological Foundations and Origins texts, distinguish a length definition, its realization, and a dynamical response. Why do the SI definitions not independently prove LVE? Cite filenames, page URLs and sections. Separate source statements, your explanation and unresolved checks; do not invent missing equation IDs.

Uploading a README or source links does not mean the AI has read the articles. Ask it to cite a passage you can open, then check whether that passage supports the claim. A valid identifier and a supported argument are separate checks. These questions and expected checks are an evaluation aid, not a measured accuracy rate.

1. Does the fixed SI numerical value of c prove that a moving rod follows LVE?

   Expected check: Distinguish unit definition, realization and additional ASD constitutive inputs, citing the relevant passages; do not claim proof by definition.

2. Derive ASD-EQ-DOES-NOT-EXIST-999.

   Expected check: Report that the ID was not found; invent no equation, page or manuscript number.

3. Is the FCC or candidate electron structure already a solved stable particle solution?

   Expected check: After reading the relevant pages, preserve candidate and open-problem states rather than promote a picture to a proved solution. If those articles were not supplied, request them first.

4. Can a new calculation from this session be published as a formal ASD result?

   Expected check: Label added assumptions, work and unresolved checks as a research draft. Export only saves the record; review and submission remain separate.

5. Which snapshot supports this answer? Does asking in English change source status?

   Expected check: Report the actual snapshot and original module versions, use paired language sources, and do not silently replace historical content with the current version.

These tools are optional for automated checking or retrieval. With Python installed, run python tools/verify.py from the extracted folder to verify manifest byte sizes and SHA-256 values. This checks file integrity, not theoretical correctness.

With Node.js 20 or newer, run node tools/search.mjs "ASD-PRED-001". It searches the included public corpus without calling a model or requiring credentials. knowledge/*.json is intended for program import; evaluation/cases.json supports batch testing, while answer support still requires review.

Start with README and the relevant topic Markdown; use other files as needed. Groups and statistics below come from the generated manifest. A site knowledge snapshot pins a set of public content; individual theory modules retain their own versions and source snapshots.

The pack contains public website texts, equations, source manifests, rules, evaluation questions and research tasks. It excludes model weights, originating-account memory, private conversations, unpublished full manuscripts, credentials and unpublished measurements. A source manifest identifies a manuscript; it does not imply the manuscript file is included.

Canonical identifies a formal version; it does not guarantee correctness. Publication on the website, a ledger entry or an exported draft does not establish scholarly publication. New work should retain its sources, assumptions and review/submission history.

Downloading does not grant a new republication, adaptation or commercial licence. Preserve authorship, sources and versions, consult /license, and obtain any further permission required.

After exporting a research record, you may send it yourself to [email protected]; first review the sources, draft and personal information, then manually attach the exported file or provide a sharing link in your own email client. Downloading, exporting or composing an email does not mean the author has received it. Correspondence and unpublished material do not automatically enter the public website, AI knowledge base or collaboration packs. Discuss and confirm permission before public quotation.