Uploading papers to ArXiV

Quick overview

  1. Make sure arxiv_latex_cleaner is installed (see section 2)
  2. Copy all the relevant files into a folder, say public_relevant_files
    1. Rerun the latex to make sure the new set of files produce the same pdf.
  3. Run python -m arxiv_latex_cleaner /path/to/public_relevant_files to generate a folder /path/to/public_relevant_files_arXiv. This folder will contain the comments stripped tex files and a filtering of the files relevant for arXiv upload.
  4. Rerun the latex in the folder public_relevant_files to make sure the new set of files produce the same pdf as in public_relevant_files.
    1. You may have to copy over all the figures again, since arxiv_latex_cleaner may do some unnecessary image manipulation.
  5. Make arXiv-specific edits (see section 3)
  6. Produce the arXiv-ready PDF
  7. Check the arXiv-ready PDF (see section 4)
  8. Perform latexmk -C in the new folder to remove irrelevant figures.
  9. Zip the folder arxiv_generic_arXiv and upload.
  10. Delete the public_relevant_files folder if needed

Setup

  1. Clone the git repository https://github.com/google-research/arxiv-latex-cleaner
  2. Do python setup.py install to make it like a command line`

arXiv-specific edits in public_relevant_files_arXiv folder

  1. Get the generic template for arxiv from https://github.com/kourgeorge/arxiv-style.
    1. Note that you can get the sty file directly by using the link https://github.com/kourgeorge/arxiv-style/raw/master/arxiv.sty.
    2. Copy over the arxiv.sty into the public_relevant_files_arXiv folder.
  2. Preamble: Instead of the conference/journal specific commands, use the following preamble
    \documentclass{article}    %
    \usepackage{arxiv}
    \usepackage{bookmark}
    
  3. Document heading
    \title{TITLE OF THE PAPER\thanks{Funding blurb}}
    \author{
        Author 1\thanks{Author 1 info} , 
        % Spacing helps the symbol
        Author 2\thanks{Author 2 info} , and
        Author 3\thanks{Author 3 info} 
    }
    \date{}
    \chead{SHORT TITLE}    % Otherwise thanks will ruin it
    \renewcommand{\headeright}{}    % To remove the Preprint
    \renewcommand{\undertitle}{}    % To remove the Preprint
       
    \begin{document}
    \maketitle
    
  4. Keyword
    \keywords{KEYWORD1 \and KEYWORD2 \and KEYWORD3}
    
  5. Bibliography
    1. Copy over original_tex_file.bbl into the public_relevant_files_arXiv folder as arxiv.bbl.
    2. Add \input{arxiv.bbl} to the end of the file instead of the bibliography commands.
    3. We avoid using the same file name as the original_tex_file.bbl to avoid latexmk from overwriting it.

Checks

  1. Make sure linebreaks that were built-in for two-column format are now removed
  2. Make sure equations are single lined, when possible, to reduce the page length.
  3. Figures:
    1. Tikz-based figures might have to replaced with screenshots of the original paper.
    2. Make sure figures are placed correctly. Sometimes figures that were forced to below each other can now be placed side-by-side.
    3. Remove all .fig and .svg files
    4. Remove any extraneous files.
  4. Use \adjustbox{width=DESIRED_WIDTH}{BLAHBLAH} to resize tables to the desired shape.

Creating a diff pdf

Use latexdiff --type=BOLD initial_subm.tex revised_subm.tex > diff.tex to create a diff file that only highlights the new changes, and ignores deletion.

To change from bold to color, use

\providecommand{\DIFadd}[1]{{\protect\color{blue}#1}}
%instead of
%\providecommand{\DIFadd}[1]{{\bf #1}} %DIF PREAMBLE

instead of the generated \DIFadd command in the diff.tex.

Similarly, to see the removed parts, use

\providecommand{\DIFdel}[1]{{\protect\color{red}#1}} %DIF PREAMBLE
%instead of
%\providecommand{\DIFdel}[1]{} %DIF PREAMBLE

Biography photos

Using GIMP 2.8 to satisfy IEEE photos

  1. Image > Mode > Grayscale
  2. Image > Scale image (Correct for dpi and photo size)
  3. Image > Canvas size (Crop and reposition)

Source: https://tex.stackexchange.com/questions/315914/customizing-latexdiff-default-underline-output-styling

Research Scientist

Researcher with experience in optimization, control, stochastic modeling, and reinforcement learning