Human Skills · Writing

Using LaTeX

Definition

LaTeX is the field-standard typesetting system for technical and academic publishing — many journals and conferences accept it as the only source format. The fastest way to actually understand it, especially for someone with programming experience, is a single reframe: LaTeX is a programming language whose compiler produces a formatted document instead of an executable. A .tex file has structure, syntax, and includes, in the same way source code does; running it through a LaTeX toolchain “compiles” it to a PDF the way a compiler produces a binary.

Learning Outcome

After working through this, you should be able to read an unfamiliar .tex file and recognize its structure the same way you’d recognize the structure of an unfamiliar program — imports, declarations, named references, and a build step — rather than seeing it as an opaque wall of backslashes.

Core Structure

The mapping to familiar programming concepts:

LaTeX construct Programming analogy
\documentclass{...} Declares the “type” of the whole document (like a base class or template)
\usepackage{...} Import / include — pulls in extra functionality (tables, figures, custom boxes)
\begin{...} ... \end{...} A block or scope — environments like abstract, figure, table, itemize
\label{...} / \ref{...} A named reference / pointer — label something once, refer to it anywhere, and let the compiler resolve the actual number
\cite{...} + a .bib file A dependency on an external database — the bibliography is a separate data file, referenced by key, resolved at compile time
% Comment

The compile model itself is also a build pipeline worth recognizing: your .tex source and a separate .bib bibliography database are processed together (traditionally via pdflatex and bibtex, often run automatically by an editor or online tool) to produce a single output PDF.

Worked Example

A minimal but complete sample document demonstrates the whole structure in one file. Full source, rendered output, and bibliography database are archived here:

The zip is fully self-contained and ready to build: source, bibliography, the IEEEtran class and style files, and the figure it references. Unzip it and run your usual LaTeX/BibTeX toolchain to reproduce the PDF above exactly.

The document declares its type and imports, in the first few lines:

\documentclass{IEEEtran}
\usepackage{wrapfig}
\usepackage{graphicx}
\usepackage[export]{adjustbox}
\usepackage{mdframed}

\title{Sample \LaTeX\ Document}
\author{Mike Heroux}

It then works through the standard sections of a technical report — abstract, introduction, body, conclusion — each as a \section{...} block. Along the way it demonstrates the constructs that come up in almost any technical document: a numbered equation with a \label that’s referred back to later by \ref, a table, a figure wrapped in text via wrapfigure, and both numbered (enumerate) and bulleted (itemize) lists, including a note that lists can nest.

Citations tie the document to its bibliography database. The sample cites two sources:

The former topic is covered very well in~\cite{High93} and
the latter in~\cite{Lamp86}.
...
\bibliography{sample}

Each citation key resolves to a full entry in sample.bib — a plain-text database where each entry has a type, a key, and a set of fields:

@book{High93,
title = "{Handbook of Writing for the Mathematical Sciences}",
author = "Nicholas J. Higham",
publisher = "SIAM",
year = 1993}

@book{Lamp86,
title = "{LaTeX User's Guide and Reference Manual}",
author = "Leslie Lamport",
publisher = "Addison-Wesley",
year = 1986}

At compile time, LaTeX and BibTeX match every \cite{key} in the document against an entry with that key in the .bib file, and generate a formatted bibliography automatically — you never hand-format a reference list.

Common Pitfalls

  • Trying to manually control exactly where a figure or table lands on the page. LaTeX places floats algorithmically based on the surrounding content; you can nudge it, but you don’t have direct control the way you would dragging an image in a word processor.
  • Citing a key with \cite{...} that has no matching entry in the .bib file — this produces an unresolved reference rather than a build error you’d immediately notice.
  • Treating \label and \ref as manual bookkeeping (writing the number yourself) instead of letting the compiler resolve them — which is the entire point of using them.
  • Skipping the “read an existing document as a template” strategy. The source file above is meant to be copied from and modified, not written from scratch each time.

Rubric / Checklist

  • Document has a \documentclass and only the \usepackage imports it actually uses
  • Every figure and table has a caption and a \label that’s referenced at least once via \ref
  • Every \cite{key} has a matching entry in the .bib file
  • A bibliography style is declared and \bibliography{...} is included
  • When in doubt about how to express a construct, an existing document (like the sample above) is used as a template rather than guessing from scratch