Picking a Paper to Replicate

Starting the LaTeX Project — choose a source, lock your scope, outline, and scaffold

A short conceptual reading to start the LaTeX Project, the two-week project that runs across Weeks 5 and 6. This week is about planning and setting up the project. The exact assignment prompt, any source-check or approval steps, due dates, and submission details live in the Assignments/LMS space.

Weeks 2–4 built the pieces of mathematical writing in Quarto: notation, structure, and the supporting apparatus of figures, tables, and citations. Week 5 stops adding new tools and turns those pieces into the start of a real short paper.

The LaTeX Project is a short replication of an open-access mathematics paper or excerpt — a polished short paper, a few pages long. It runs across two weeks: this week you plan and scaffold it; next week you write and polish it. There is no new tool this week. You work in Quarto .qmd and render to PDF exactly as you have since Week 1. (If you are curious what raw .tex looks like underneath, you are welcome to look — but it is entirely optional and not part of the project.)

What “replication” means here

Replication means reproducing a small, self-contained piece of an existing source in your own typeset document, citing that source. It is not retyping a whole paper, and it is not inventing new mathematics. You take one well-chosen unit, write it up clearly in your own words and notation, and point back to where it came from.

A good Week 5–6 unit is one of these:

  • a theorem and its proof;
  • a derivation worked through step by step;
  • a figure or table and the exposition that explains it;
  • a short passage or section — a self-contained definition with an example, a lemma, a single worked argument.

That is the whole project. One unit, reproduced well, in a few clean pages.

Scope is the main risk — not syntax

By now the syntax is not the hard part; you can already typeset math, structure writing, and add a figure, a table, and a citation. The thing that makes a short replication project go wrong is almost always scope: choosing a whole paper, a hard theorem, or several results at once, and then running out of time.

So the most important decision this week is how much you take on. Pick small. A project that is scoped well nearly writes itself next week; one that is too big stalls. If you are unsure whether something is small enough, it probably is not — cut it down to one unit.

Choosing a realistic source

Prefer a source that is short, self-contained, and readable at your level over an ambitious one. Concretely, look for:

  • a paper or excerpt where one unit can be lifted out without needing ten pages of prior setup;
  • mathematics you can follow and explain — if you cannot follow the proof, you cannot replicate it;
  • a source that is genuinely open-access, so you can read all of it and cite it cleanly.

The companion resource — Finding open-access mathematics sources — shows where to look (arXiv, open-access journals, public-domain texts) and how to check a source before you commit to it.

Verifying a source is genuinely open-access

“Open-access” here has a practical meaning: you can read the full text without a paywall, and you are allowed to read and cite it. Before you build a project on a source, confirm three things yourself:

  • you can open the full text (not just an abstract) without paying;
  • the source is real — it actually exists and is what you think it is;
  • you can capture a clean citation for it (authors, title, year, and a stable link or DOI).

This is the same provenance discipline the course applies to data: know where it came from, and be able to cite it. The open-access sources resource has a short checklist you can run through.

From source to annotated outline

Once you have your source and your one unit, outline before you write. An outline is the list of section headings your finished paper will use, with a sentence under each saying what will go there. A workable shape for a replication is:

  • Setup — the definitions and notation the reader needs.
  • Statement — the theorem, result, or claim you are reproducing.
  • Replication — your own write-up of the proof, derivation, or exposition.
  • Discussion — a short note on what it means or why it is interesting (optional, brief).
  • References — the source, cited.

Those headings become the section headings in your document, and the one-sentence notes become your annotated outline — placeholders that tell next-week-you exactly what to write.

Scaffolding the project folder

The project gets its own permanent folder, latex-project/, which you build over two weeks and finish in place. Setting it up now — and getting it to render — removes almost all of the painful surprises from next week.

A scaffold is not a draft. It is the skeleton of the paper that already compiles: a title block, your outline as headings, a placeholder sentence under each, and your source already cited so the References section appears. It should look like an outline of a paper that builds cleanly — mostly empty, but working.

The Week 5 workflow

You can do all of this on your own machine; bring anything that won’t render to one of the scheduled studio times in the MAC.

  1. Create the project folder. In math-software-portfolio/, make a subfolder named latex-project/ (next to hw01/–hw04/).
  2. Create latex-project.qmd with a YAML title block: a project title and author: "Your Name", and format: pdf.
  3. Create references.bib in the same folder and add a real entry for your chosen source.
  4. Add bibliography: references.bib to your YAML.
  5. Cite the source once in your plan with [@yourkey].
  6. Add your headings. A short project-plan section (source, why it’s open, your unit, your scope) and the annotated outline headings, each with a placeholder sentence.
  7. Render to PDF. Open latex-project.qmd in VS Code and run Quarto: Preview (Ctrl/Cmd + Shift + P), press Ctrl/Cmd + Shift + K, or run quarto render latex-project.qmd in a terminal opened in the folder.
  8. Confirm the citation and References section resolve — the in-text [@yourkey] renders, the References list appears at the end, and there is no ?? on the page.
  9. Confirm the skeleton is clean but not content-complete. It should compile and look like an outline of a paper — that is exactly right for Week 5.

If the skeleton renders cleanly now, Week 6 is mostly writing.

What Week 6 looks like

In Week 6 you finish the project in this same latex-project/ folder. You fill the annotated outline with the actual replication of your unit, add any figure or table you need (the Week 4 apparatus skills), tighten the prose, and render the polished short paper. The better your plan and scaffold this week, the calmer next week is.

AI can help you plan — but cannot verify the source

AI assistants are useful for the planning work: they can suggest candidate topics or papers, summarize a source so you can judge its scope, sanity-check whether a unit is small enough, draft a .bib entry, or propose an outline you then edit.

What AI cannot do is tell you whether a paper exists, whether it is actually open-access, or whether you have understood it correctly. Assistants routinely recommend papers that do not exist, claim a paywalled paper is “freely available,” and invent DOIs and publication details. The course rule from earlier weeks applies with a Week-5 edge: AI is a first pass, not a verdict. You open the source yourself, confirm you can read it without paying, confirm it is real, and confirm it says what you think.

The three-line AI Use Note (Tool / Purpose / Verification) applies. This week the Verification line should describe how you confirmed your source is real and genuinely open-access, for example:

Verification: found the paper on arXiv, opened the full PDF and confirmed it is free to read, confirmed the title and authors, and rendered my skeleton to confirm the citation and References section resolved with no ??.

See the AI use guidelines for the course’s full position.

What you’ll do this week

In one paragraph: you will choose an open-access mathematics paper or excerpt, verify it is genuinely open and real, pick one self-contained unit to replicate, write a short scope statement and an annotated outline, and scaffold a latex-project/ folder that renders a clean skeleton PDF — the start of the polished short paper you finish in Week 6. The exact assignment prompt, any source-check or approval steps, due dates, and submission details live in the Assignments/LMS space.

See also