Finding open-access mathematics sources
Where to look for a paper to replicate — and how to check it before you commit
Short, practical guidance on where the source for your LaTeX Project should come from, and how to confirm it is usable before you build a project around it. This pairs with Picking a Paper to Replicate.
What “open-access” means for this course
For our purposes, a source is open-access if:
- you can read the full text — not just the abstract — without paying and without a library login;
- it is a real, findable work, with authors, a title, and a stable link or DOI you can cite;
- you are allowed to read and cite it (which you almost always are for openly posted work).
You are reproducing a small unit of mathematics in your own words and notation and citing the source. You are not republishing the original, so you do not need a special reuse license to write a replication — but you do need to be able to read the whole thing and cite it cleanly. See the note below on the difference between reading a source and copying it.
Recommended starting points
- arXiv — https://arxiv.org. The largest open repository of mathematics preprints. Every paper has a free full-text PDF and a stable identifier (e.g.,
arXiv:2401.01234). Browse by subject (math.NT, math.CO, math.PR, …) or search by topic. A great first stop for short, self-contained results. - Open-access mathematics journals — journals that publish full articles free to read. The Directory of Open Access Journals (https://doaj.org) lets you filter for mathematics journals whose articles are openly available. Many society and university journals are open-access too.
- Public-domain and classic texts — older mathematics (out of copyright) is often freely and legally available through, for example, Project Gutenberg, the Internet Archive, or university digital libraries. A clean excerpt from a classic text can make an excellent, very self-contained unit.
If a source does not fit one of these — for instance, it sits behind a paywall or you can only see an abstract — find a different source. There is no shortage of open mathematics on almost any topic this course cares about.
Reading a source vs. permission to copy
Two different things, easy to confuse:
- Can you read it? — Is the full text freely available to you? This is what “open-access” is about, and it is what you need to replicate a result.
- Can you copy it verbatim? — Reusing the original’s exact text, figures, or layout is a licensing question. You sidestep it entirely by writing the unit in your own words and notation and citing the original, which is exactly what a replication is.
So: read freely, write it yourself, and cite the source. Do not paste the original’s prose or lift its figures; reproduce the mathematics in your own document.
Capturing a citation
Once you pick a source, grab a clean citation:
- On arXiv, the abstract page has an “Export BibTeX citation” link that gives you a ready-to-use
.bibentry. Confirm the fields (authors, title, year) match the paper. - For a journal article, the article or DOI page usually offers a “cite” or “export citation” option in BibTeX format.
- For a book or classic text, build a simple entry by hand: author, title, year, publisher or source.
Save the entry in your project’s references.bib, cite it with [@yourkey], and let the document generate the reference list — the workflow you learned in Week 4.
AI can invent papers — verify before you trust
An AI assistant can suggest topics and summarize papers, and that is a fine way to start looking. But assistants routinely invent sources: a real-sounding author, title, journal, year, and even a working-looking DOI — for a paper that does not exist. They will also confidently call a paywalled paper “freely available.”
So treat any source an assistant hands you as a lead, not a fact. Before you commit:
- find the source yourself at its canonical page (arXiv, the journal, the archive);
- open the full text and confirm you can read it without paying;
- confirm the details — authors, title, year — match what you were told.
If you cannot find it, it does not exist. Pick something you can open with your own eyes.
A source-check checklist
Before you build your project on a source, confirm all five:
- I found it at its real, canonical page (not a re-shared mirror or an AI summary).
- I can open the full text without paying.
- The mathematics is something I can follow and explain.
- There is one self-contained unit — a theorem + proof, a derivation, a figure/table + exposition, or a short passage — I can lift out without needing pages of prior setup.
- I have a clean citation (authors, title, year, stable link or DOI) saved in
references.bib.
If any of the five is shaky, choose a different source now — it is far cheaper than discovering the problem mid-project.
See also
- Picking a Paper to Replicate — how to scope the project once you have a source.
- Data guidelines — the same know-your-provenance discipline, applied to datasets.
- AI use guidelines — the course’s position on AI assistance and verification.