Physics theses: equations, units, and multi-file sanity

Physics manuscripts carry more math per page than most fields. That is a feature until a half-finished chapter breaks the whole compile because one label moved.

Multi-file from the start

Even a short MSc thesis benefits from a root file and chapter includes:

\documentclass{report}
\begin{document}
\include{chapters/intro}
\include{chapters/methods}
\include{chapters/results}
\end{document}

Use \includeonly{chapters/methods} while you draft so you are not rebuilding three chapters to fix one equation. Details: one root file, many chapters.

Units are not text

Write $v = 3.0\,\mathrm{m\,s^{-1}}$ by hand if you must, or load siunitx and use \qty{3.0}{m.s^{-1}}. Journals differ on spacing; pick one package path and stick to it. Mixed styles look like two authors who never met.

Dirac, vectors, and bold conventions

Quantum and condensed-matter groups often need \braket for Dirac notation. Classical mechanics groups argue about bold vs arrow vectors. Document the convention in the preamble and refuse ad-hoc exceptions mid-chapter. See Dirac notation and bold math.

Two-column journal traps

APS, IOP, and similar classes change float behavior. A figure that looked fine in article may refuse to sit beside the paragraph that cites it. Compile against the real class early, not the weekend before submission. Keep a minimal figure width (0.45\textwidth in two-column) as the default.

Keep the source offline when the lab network dies

Beamlines and cleanrooms are not known for stable Wi‑Fi. A local compile with a bundled engine (as in Oleafly) means the PDF still builds when the browser editor does not load. Store the project under Git so a bad equation edit is one checkout away from undo.