PDF with Formulas to Markdown — Equations as LaTeX
Equations are images of glyphs, not text. The default conversion of an equation-heavy paper is gibberish — half the symbols missing, the rest garbled. We detect equation regions, run them through a math-aware recogniser, and emit valid LaTeX that any Markdown viewer with MathJax or KaTeX renders correctly.
Inline vs display math
Inline equations (the kind that sit in a sentence: "the loss is L = ...") come out wrapped in single dollars: $L = ...$. Display equations (the centred, numbered ones) come out wrapped in double dollars on their own line: $$\nabla \cdot E = \rho/\epsilon_0$$. Equation numbers, when present, are preserved as a parenthetical comment after the closing dollars.
Where this LaTeX renders
Anywhere with MathJax or KaTeX support: Obsidian (built-in), GitHub README rendering (since 2022), MkDocs Material with the math extension, Jupyter notebooks, Quarto, and most Markdown editors with a math plugin. For tools without native math support, the LaTeX still reads correctly as plain text and round-trips cleanly into proper math software.
Standard LaTeX: arithmetic, calculus, linear algebra, set notation, summations, integrals, fractions, sub/superscripts, Greek letters, common operators. Specialised notation (commutative diagrams, complex tensor indices) may degrade — review the LaTeX manually for the rare cases.
Inline vs display: how are they distinguished?
Display equations (centred, often numbered, set off from text) come out in double-dollar blocks on their own line. Inline equations stay inside their paragraph in single dollars. The distinction is detected from the source layout, not configured by the user.
Will the LaTeX render in Obsidian and GitHub?
Yes — both have native MathJax-compatible rendering. Obsidian supports both inline and display math out of the box; GitHub has rendered math in Markdown since June 2022 ($$ blocks and $...$ inline).
Can I convert physics or engineering papers?
Yes — those are the most common use case. Vector calculus, partial derivatives, tensor notation, and standard physics symbols (Planck constant, electron charge, etc.) all convert with high fidelity. Greek letters render natively in LaTeX.
Are equation numbers preserved?
Yes — when the source PDF labels equations (e.g. "(3.14)"), we emit them as comments after the closing $$. Cross-references in the body that say "see equation (3.14)" remain unchanged so they still match.