agentsclimarketplace

Latex translation guide

Skill brycewang-stanford/Auto-Empirical-Research-Skills/skills/43-wentorai-research-plugins/skills/tools/ocr-translate/latex-translation-guide

🔬 A curated collection of 23,000+ agent skills for empirical research across 8 social science disciplines. | 精选 23,000+ AI Agent 技能库,覆盖8大社会科学学科的实证研究。CoPaper.AI 20分钟完成一篇可复现的规范实证论文,并支持用户上传 Skills。-- Maintained by CoPaper.AI from Stanford REAP.

Install
npx -y skills add brycewang-stanford/Auto-Empirical-Research-Skills --skill latex-translation-guide

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.

What its author says it does

Copied from the file, not written here

Translate LaTeX documents preserving math formulas and structure

SKILL.md

4.6 KB, as published. Nobody here has run it

LaTeX Document Translation Guide

Overview

Translating LaTeX academic documents requires preserving mathematical formulas, cross-references, citations, and formatting while converting the text between languages. This guide covers tools and techniques for translating LaTeX papers — from command-line utilities to full document pipelines. Particularly useful for making research accessible across language barriers.

LaTeXTrans Approach

# Install LaTeXTrans
pip install latextrans

# Translate a LaTeX file
latextrans translate paper.tex --from en --to zh --output paper_zh.tex

How It Works

  1. Parse: Extract text segments while preserving LaTeX commands
  2. Protect: Shield math environments ($...$, \[...\], equations)
  3. Translate: Send text segments to translation API
  4. Reconstruct: Reassemble with original LaTeX structure

Python Usage

from latextrans import LatexTranslator

translator = LatexTranslator(
    source_lang="en",
    target_lang="zh",
    engine="google",  # or "deepl", "openai"
)

# Translate a file
translator.translate_file("paper.tex", "paper_zh.tex")

# Translate a string
result = translator.translate(
    r"The loss function $\mathcal{L}(\theta)$ is minimized "
    r"using gradient descent with learning rate $\eta$."
)
# Output preserves $\mathcal{L}(\theta)$ and $\eta$ untouched

MathTranslate Tool

# Install MathTranslate (specialized for math-heavy papers)
pip install mathtranslate

# Translate arXiv paper directly
translate_arxiv 2301.00001 -o translated.tex

# Translate local file
translate_tex paper.tex -o paper_translated.tex

MathTranslate Features

# Configuration
import mathtranslate

# Set translation backend
mathtranslate.config.set_translator("google")  # free
mathtranslate.config.set_translator("openai")  # higher quality

# Translate with customization
mathtranslate.translate(
    input_file="paper.tex",
    output_file="paper_zh.tex",
    source_lang="en",
    target_lang="zh-CN",
    threads=4,  # parallel translation
)

Manual Translation Tips

Protecting Math Environments

import re

def extract_and_protect(latex_text: str) -> tuple:
    """Extract math environments before translation."""
    math_pattern = r'(\$\$[\s\S]*?\$\$|\$[^$]+\$|\\begin\{equation\}[\s\S]*?\\end\{equation\}|\\begin\{align\}[\s\S]*?\\end\{align\})'

    placeholders = {}
    counter = [0]

    def replace_math(match):
        key = f"__MATH_{counter[0]}__"
        placeholders[key] = match.group(0)
        counter[0] += 1
        return key

    protected = re.sub(math_pattern, replace_math, latex_text)
    return protected, placeholders


def restore_math(translated: str, placeholders: dict) -> str:
    """Restore math environments after translation."""
    for key, value in placeholders.items():
        translated = translated.replace(key, value)
    return translated

Commands to Protect

% Always protect these:
\ref{...}      % Cross-references
\cite{...}     % Citations
\label{...}    % Labels
\eqref{...}    % Equation references
\url{...}      % URLs
\texttt{...}   % Code/monospace

% Math environments to protect:
$...$          % Inline math
$$...$$        % Display math
\[...\]        % Display math
\begin{equation}...\end{equation}
\begin{align}...\end{align}
\begin{theorem}...\end{theorem}  % Custom environments

Bilingual Output

% Create side-by-side bilingual document
\usepackage{paracol}

\begin{paracol}{2}
\switchcolumn[0]
The transformer architecture has become...

\switchcolumn[1]
Transformer架构已经成为...

\switchcolumn[0]
Self-attention computes $\text{Attention}(Q,K,V) = \text{softmax}(\frac{QK^T}{\sqrt{d_k}})V$

\switchcolumn[1]
自注意力计算 $\text{Attention}(Q,K,V) = \text{softmax}(\frac{QK^T}{\sqrt{d_k}})V$
\end{paracol}

Translation Backends

BackendQualityCostSpeed
Google TranslateGoodFreeFast
DeepLBetterFreemiumFast
OpenAI GPT-4BestPaidSlower
ClaudeBestPaidSlower

References

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.