About this English edition
This practice course teaches you to read, write, check, and share mathematics so that its claims, sources, computations, corrections, and limits of certainty can be traced and examined.
- Title: Traceable Mathematical Work.
- Language and version: English,
en, 2026.08.31. - Scope: the complete original nine-unit O017 / D120 course, with its delivery and assessment package, self-study guidance, and English backend.
- English HTML reader: open the reader.
- English PDF and downloadable packages: open the English release.
- English editable source: open the English source tree.
- English-additive version DOI: 10.5281/zenodo.22209140.
- Indonesian edition preserved unchanged: HTML reader, version DOI 10.5281/zenodo.22073823.
- Work’s preservation lineage: https://doi.org/10.5281/zenodo.22051866. The Indonesian version DOI above identifies the Indonesian source edition, not this English translation.
- PDF accessibility boundary: the PDF is a print-oriented, untagged edition. Use the semantic HTML reader for reflow, keyboard navigation, heading structure, alternative descriptions, and native MathML.
- Production: OpenAI Codex gpt-5.6-sol, Ultra, acting on the human creator’s instructions. KokunoYumeto is the public human creator and rights-holder identity for protectable original material. No human peer review is claimed.
- Rights: original expression CC BY-SA 4.0; methodological donor content CC BY 4.0; software, runtime, and fonts retain their component licenses. LICENSES.md, ENGLISH_EDITION_RIGHTS.md, and THIRD_PARTY_NOTICES.md give the controlling map.
The course assumes that you can already write elementary proofs, use discrete mathematics, and run a reproducible mathematical computation. It does not teach Python or proof techniques from the beginning. Its subject is how to use those skills in mathematical research whose evidence and reasoning can be audited.
Begin with Unit 1: Reading arguments as networks of claims.
0.1 Nine units
- Reading arguments as networks of claims
- Tracing the literature and assessing source authority
- Reconstructing abbreviated proofs
- Citation, provenance, and source notebooks
- Auditable mathematical exposition
- Reproducible computation as empirical evidence
- Errata, corrections, and limits of certainty
- Seminar notes, referee reports, and responses
- A bounded community contribution and final portfolio
After the nine units, use the delivery and authentic-assessment package. It distinguishes simulated practice from evidence of actual activity, specifies independent assessment where the learning claim requires it, and never treats a local draft as a completed community contribution. These are assessment definitions, not a prerequisite for accessing or publishing this course.
0.2 Core technical glossary
This glossary distinguishes related terms that must not be used interchangeably.
- A proof is a mathematical argument establishing a claim from stated assumptions; proving is the process of constructing or checking such an argument. Empirical evidence, including successful computations, is not automatically a proof of a universal mathematical statement.
- A graph has vertices and edges. A connected graph has a path between every pair of vertices. A spanning tree is a connected, acyclic spanning subgraph. A minimum spanning forest is not a spanning tree of a disconnected graph.
- A dependency graph displays logical relationships between claims. A software dependency is a package or component needed by an implementation. Neither use should obscure which kind of dependence is being asserted.
- Function names the mathematical object. Mapping is used when its action or the mapping perspective is specifically intended.
- A sequence is distinct from a series. Recursion is a process; recursive describes its form; a recurrence relation specifies terms by their relationship to other terms.
- Continuous describes a function; continuity is the property. An integral is the mathematical object and integration the operation or method.
- Numerical error concerns an approximation or computation. An erratum is a publication correction; it is not the same as numerical error.
- Reproducibility concerns repeating a stated process using specified artifacts and conditions. A minimal reproducer is a small, sufficiently complete case that exhibits a reported behavior. Reproducibility does not by itself establish implementation correctness or a theorem.
- Computation is the broader computational activity; a calculation is a particular calculation. Literature search, testing, test results, data, and algorithm each name different parts of the work.
- Provenance is the auditable history of origins, versions, transformations, and evidential relationships. A manifest records artifact identities, sizes, hashes, rights, and states. A version DOI identifies a particular release; a concept DOI can identify a work across releases.
The Indonesian source edition checked its terminology directly against the Indonesian TeX source of arXiv:0807.4609v1 and official 2026 mathematics teaching materials by Rinaldi Munir at Institut Teknologi Bandung. The arXiv witness supports distinctions involving computation, calculation, literature study, testing, data, and algorithms; it is not a witness for theorem, lemma, proof, peer-review, or erratum terminology. The Indonesian edition’s separate QA records preserve those boundaries and the exact source identities; the witness documents are not redistributed here.
This English edition translates the original O017 course, not The Turing Way or Research Software Engineering with Python a second time. Their already-English, commit-pinned methodological sources, citations, and credits are retained. English prose is checked against the meaning of the mathematical arguments and the unchanged formulas, identifiers, exercises, and code. Local filenames and anchors retain their source-edition forms so that language versions remain interoperable. The D/A/K/J/T annotation codes are retained as stable labels even where the English words differ from their Indonesian initials.
Every unit includes learning outcomes, worked mathematical cases, guided practice, six exercises, six answer guides, and a completion task. The visible stable identifiers also appear in the backend so that another language edition does not need to reconstruct identities from HTML.
0.3 Course boundaries
Programming and computation are prerequisites. This reader does not reteach Python syntax, package APIs, numerical methods, environments, software testing, or deployment; those topics belong to B80. O017 focuses on the status of claims, proofs, sources, provenance, exposition, errata, experiment versus proof, scholarly criticism, and bounded contributions.
The English edition is an additional language-access surface for the same course, with the same O017-U01 through O017-U09 unit identities. It neither replaces the Indonesian edition nor reopens its completed production scope. Edition production and the separate curriculum-admission decision are not evidence for each other.
0.4 Rights and bundled software
Content, code, factual data, methodological donors, fonts, and the HTML runtime have separate licenses. Read LICENSES.md, the English-edition attribution and change notice, and THIRD_PARTY_NOTICES.md. The editable-source package includes the component rights map and the build and validation instructions.