LaTeX for Academic Publishing: What Journals Actually Require (And What's Optional)

#latex #devjournal #programming #tutorial

A practical guide based on real journal submission experience - not the textbook version.

"Submit in LaTeX" covers a lot of ground. Some journals require it strictly. Others accept Word but strongly prefer LaTeX. A few say "LaTeX preferred" and mean "we'll accept either but we'll reformat your Word document and it will look different from what you submitted."

After handling LaTeX formatting for academic papers across IEEE, ACM, Springer, Elsevier, MDPI, and arXiv at The LaTeX Lab, here's what we actually know about what different publishers require, where submissions break, and what matters most.

Which Publishers Actually Require LaTeX

Effectively mandatory:

IEEE - all Transactions, Letters, and conference proceedings. IEEEtran is the required class. Submitting Word gets your paper reformatted by the editorial office, often with formatting that doesn't match what you intended.
ACM - all venues using the acmart class (SIGCHI, SIGCOMM, PLDI, etc.). Their submission system accepts PDFs but the camera-ready requires the LaTeX source.
arXiv - requires LaTeX source for most submissions. PDFs without source are accepted but flagged as lower confidence for reproduction.
Most mathematics journals (AMS, SIAM) - LaTeX is the de facto standard; Word submission is unusual enough that editorial staff often struggle with it.

Strong preference, not technically required:

Springer - LNCS conference proceedings require LaTeX. Journal articles accept Word but the typesetting quality is noticeably different.
Elsevier - accepts Word for most journals but the elsarticle LaTeX class is the production format. Authors who submit LaTeX go through less reformatting in production.

Genuinely either/or:

MDPI - accepts both. Their mdpi.cls template is well-maintained and produces clean output, but Word submissions are handled with similar quality.
PLOS ONE - accepts Word or LaTeX. No strong preference either way.

The Three Most Common Journal Template Mistakes

1. Loading packages the class file already loads

Every journal class file loads a set of packages internally. When you add them again in your preamble, you get option clashes or redefinition warnings.

The common offenders with IEEEtran:

% Don't load these separately with IEEEtran - it handles them:
\usepackage{cite}        % IEEEtran has its own citation handling
\usepackage{graphicx}    % already loaded
\usepackage{amsmath}     % load this one is fine, but check first

% Same issue with elsarticle:
\usepackage{natbib}      % elsarticle loads natbib internally

Read your class file's documentation before adding packages.

2. Wrong citation style for the journal

Every publisher has a required citation style. Getting it wrong is a desk rejection risk at some venues, and always requires a revision round.

% IEEE: numeric citations, sorted by appearance
\bibliographystyle{IEEEtran}

% ACM (acmart class): handles citation style internally via class options
\documentclass[format=acmsmall, review=false, screen=true]{acmart}

% Springer LNCS: numeric, sorted by appearance
\bibliographystyle{splncs04}

% Elsevier (author-year for most journals):
\bibliographystyle{elsarticle-harv}  % Harvard author-year
% or
\bibliographystyle{elsarticle-num}   % numeric

Check the journal's author guidelines - not the template README, the actual guidelines page. They often contradict each other.

3. Two-column float handling

IEEE and ACM use two-column layouts. Figures and tables that span both columns need figure* and table* environments. Using figure for a wide figure produces output that overflows into the second column.

% Single-column figure (fits in one column):
\begin{figure}[t]
  \includegraphics[width=\columnwidth]{fig1}
  \caption{Single-column figure.}
\end{figure}

% Full-width figure spanning both columns:
\begin{figure*}[t]
  \includegraphics[width=\textwidth]{fig2}
  \caption{Full-width figure spanning both columns.}
\end{figure*}

The * variants can only be placed at the top or bottom of a page ([t] or [b]), not inline. This causes figures to float further than expected - usually fixed by adjusting the figure's position in the source relative to where you want it to appear.

What "Template Compliance" Actually Means in Practice

Journal editorial offices run their own checks on submitted LaTeX. What they're looking for:

Margins. The class file handles this - don't override \geometry settings unless the template documentation says to. Any margin change that makes the paper shorter (easier to read for reviewers) will be caught.

Font size. Some conferences enforce 10pt strictly. \small or \footnotesize in the main body text is a flag.

Reference format. This is the most common reason for revision requests. A missing DOI field, an inconsistent journal name abbreviation, or conference papers formatted as journal articles.

Figure resolution. PDFs submitted to most venues should have figures at 300dpi minimum for raster images. Vector figures (PDF, EPS, SVG) don't have this issue - use them wherever possible.

arXiv-Specific Notes

arXiv has a few requirements that trip up first-time submitters:

All files in one directory. arXiv's compiler expects your .tex, .bib, all figure files, and your .cls/.sty files in a flat directory or a simple zip. Nested subdirectories for figures sometimes cause compilation failures.

No \pdfoutput=1. arXiv sets this itself. Including it in your preamble can cause conflicts.

BibTeX, not BibLaTeX. arXiv's compilation environment supports BibTeX with bibtex. BibLaTeX with biber is supported but less reliable. If you're targeting arXiv, test with the BibTeX backend.

The .bbl file. If arXiv can't run BibTeX on your submission, it won't compile the bibliography. Submit the pre-compiled .bbl file alongside your .tex to guarantee bibliography output.