Jekyll2026-03-09T20:41:38+09:00https://hanseuljo.github.io/feed.xmlAbout MePhD candidate at KAIST AI. Ex-intern at Google NYC. Working on Opt/ML Theory & LLMs.Hanseul Cho[LaTeX Tips] The Most Elegant Way (Maybe) to Indicate Equally-Contributed Authors Separated with Commas2025-11-26T00:00:00+09:002025-11-26T00:00:00+09:00https://hanseuljo.github.io/posts/latex-tip-equal-contribution

TL;DR

📌 Use the command \rlap{$^*$} when you list multiple equally contributed authors’ names, separated by commas (e.g., ICLR template).

% For example:
[1stAuthor1]\thanks{Equal Contribution.},~~[1stAuthor2]\rlap{$^*$},~~[1stAuthor3]\rlap{$^*$},~~[2ndAuthor],...

What’s the Problem?

I’d like to share a visually appealing LaTeX trick(?) for listing multiple co-first authors that I recently figured out.

Let me introduce three of my “colleagues” (graduate students from KAIST AI) who will assist me in this document: meet Chahun Seol, Sanchul Heo, and Honseul Cha! Of course, they are all imaginary characters I created. (You may notice these names are all anagrams of my name, Hanseul Cho.) But in this document, they and I will play the role of co-authors of a brilliant paper submitted to ICLR 2026, preparing together for a camera-ready version of our manuscript. 😎 (Please don’t get me wrong; I don’t want to advertise an imaginary paper that I haven’t actually taken part in!)

You’re probably familiar with the command \thanks{...} for indicating equal contributions, affiliations, or side information about specific authors. But how should we mark the same footnote symbol (e,g. asterisk(*), 
) to mutiple author names among us (e.g., co-first authors)?

For example, one of the simplest line of code would be 


\author{
% Please be aware that the default template for ICLR main conference papers recommends us putting commas to delimit the author names.
Hanseul Cho\thanks{Equal Contribution.},~~Chahun Seol$^*$,~~Sanchul Heo$^*$,~~Honseul Cha \\ % author names
Kim Jaechul Graduate School of AI, KAIST \\ % affiliation
\texttt{\{jhs4015,chahun.seol,sanchul\_heo,honseul-cha\}@kaist.ac.kr} \\ % e-mail addresses
\And
...[More CoAuthors]...
}

It seems good
 or does it? When we actually compile this, we get the following:

Just Asterisks

Do you notice anything weird? Yes, the gap between the asterisk and the comma!

  • The command \thanks{...}, somehow reduces the gap between the footnote mark and the comma, so that the compiled result shows the comma right after the last letter of an author name; 🙂
  • The command $^*$, however, is just a letter (in math mode of LaTeX)! It has no additional function to shrink the gap between the commas and the author names. 👿

What should we do to match the lengths of the gaps? đŸ€”

Common Failure(?) Cases

đŸ‘ș Failure 1: If the command \thanks is the only answer, how about using the same command several times?:

...
Hanseul Cho\thanks{Equal Contribution.},~~Chahun Seol\thanks{Equal Contribution.},~~Sanchul Heo\thanks{Equal Contribution.},~~Honseul Cha \\ % author names
...

The compiled result:

Multiple Thanks

Obviously, this isn’t the right wayâ€ŠđŸ˜« it applies distinct footnote marks for all co-author names (attached with \thanks).




đŸ‘ș Failure 2: How about manually adjusting the horizontal spaces?

...
Hanseul Cho\thanks{Equal Contribution.},~~Chahun Seol$^*$\hspace{-1.6mm},~~Sanchul Heo$^*$\hspace{-1.6mm},~~Honseul Cha \\ % author names
...

The compiled result:

Multiple Thanks

Okay, now it seems (almost) perfect, at least visually! However, I don’t want to do it this way: the size of shrinkage (e.g., -1.6mm above) needs to be determined carefully through multiple compilations. Not only is it not elegant, but it’s also practically undesirable, especially when your entire LaTeX project is so heavy


The Solution That I’ve Found

To figure out an elegant & practical solution, I probed the ICLR template style file (iclr2026_conference.sty), with the keyword ‘thanks’. And I found the following code lines


\def\maketitle{\par
\begingroup
   \def\thefootnote{\fnsymbol{footnote}}
   \def\@makefnmark{\hbox to 0pt{$^{\@thefnmark}$\hss}} % for perfect author
                                                        % name centering
%   The footnote-mark was overlapping the footnote-text,
%   added the following to fix this problem               (MK)
   \long\def\@makefntext##1{\parindent 1em\noindent
                            \hbox to1.8em{\hss $\m@th ^{\@thefnmark}$}##1}
   \@maketitle \@thanks
\endgroup
\setcounter{footnote}{0}
\let\maketitle\relax \let\@maketitle\relax
\gdef\@thanks{}\gdef\@author{}\gdef\@title{}\let\thanks\relax}

Can you guess how the command \thanks is redefined? I’m not an expert on LaTeX, but I somehow realized that the fourth line above (\def\@makefnmark{\hbox to 0pt{$^{\@thefnmark}$\hss}}) is sort of an important part, which seems to put the footnote mark behind the author name! But what is \hss?

I asked Google Search for help (side note: I “still” prefer to find solutions for debugging my code without using LLM, although I’m a PhD student studying AI
). The Google Search told me: SEE THIS POST ON STACKEXCHANGE (“What are the \hbox \hss \hskip commands?”). ROGER THAT.

It basically tells us:

  • \hbox to <WIDTH>{<SOMETHING>} puts <SOMETHING> inside a horizontal box of width <WIDTH>.
  • You can adjust the alignment within the box using \hss inside <SOMETHING>.
    • If <SOMETHING> is \hss <STUFF>\hss, it aligns <STUFF> at the center.
    • Also, \hss <STUFF> corresponds to the right alignment of <STUFF>.
    • Similarly, <STUFF>\hss corresponds to the left alignment of <STUFF>.
  • For the cases where <WIDTH> equals 0pt, TeX introduces their shorthand commands: clap, llap, and rlap.
    • \hbox to 0pt{\hss <STUFF>\hss} == \clap{<STUFF>}.
    • \hbox to 0pt{\hss <STUFF>} == \llap{<STUFF>}: the right end of <STUFF> comes to center, so the text <STUFF> goes left of the center.
    • \hbox to 0pt{<STUFF>\hss} == \rlap{<STUFF>}: the left end of <STUFF> comes to center, so the text <STUFF> goes right of the center.

So, my conjecture is: Since the style file contains \hbox to 0pt{$^{\@thefnmark}$\hss}, if \@thefnmark is substituted with the asterisk(*), then it must be equivalent to \rlap{$^*$}!

Therefore, my FINAL TRIAL is as follows:

...
Hanseul Cho\thanks{Equal Contribution.},~~Chahun Seol\rlap{$^*$},~~Sanchul Heo\rlap{$^*$},~~Honseul Cha \\ % author names
...

And the compiled result is


rlap is GOAT

TADA! An elegant code (without laborious decision of numerics) and visually satisfying result! 🎉

Conclusion

I hope this document help some researchers who are still novice in LaTeX writing (like me). Lastly, say goodbye to my awesome imaginary colleagues, Chahun Seol, Sanchul Heo, and Honseul Cha! 👋

]]>
Hanseul Cho
[Paper Reading] PENCIL: Long Thoughts with Short Memory2025-04-04T00:00:00+09:002025-04-04T00:00:00+09:00https://hanseuljo.github.io/posts/pencil-erasable-cot

For mobile: View PDF

Main References

Supplementary References

]]>
Hanseul Cho
[Ph.D. Thesis Proposal] Unraveling and Overcoming Challenges in Machine Learning: Generalizability, Adaptability, and Multifacetedness2025-01-24T00:00:00+09:002025-01-24T00:00:00+09:00https://hanseuljo.github.io/posts/phd-thesis-proposal

*This slide was presented in OptiML’s internal workshop. (Latest Update: 2026.02.13.)

]]>
<b>Hanseul Cho</b>
[Paper Reading] Viewing Log-Depth Transformers via the Lens of Distributed Computing2024-10-10T00:00:00+09:002024-10-10T00:00:00+09:00https://hanseuljo.github.io/posts/log-depth-transformer

For mobile: View PDF

Main References

]]>
Hanseul Cho
[Paper Reading] StableSSM: Alleviating the Curse of Memory in State-space Models through Stable Reparameterization2024-07-16T00:00:00+09:002024-07-16T00:00:00+09:00https://hanseuljo.github.io/posts/stable-ssm

For mobile: View PDF

Main References

]]>
Hanseul Cho
[Paper Reading] Convex and Non-convex Optimization under Generalized Smoothness2024-06-03T00:00:00+09:002024-06-03T00:00:00+09:00https://hanseuljo.github.io/posts/generalized-smoothness

For mobile: View PDF

Main References

]]>
Hanseul Cho
[Paper Reading] Understanding Gradient Descent on Edge of Stability in Deep Learning2024-03-20T00:00:00+09:002024-03-20T00:00:00+09:00https://hanseuljo.github.io/posts/edge-of-stability-gd

For mobile: View PDF

Main References

]]>
Hanseul Cho
[Paper Reading] A U-turn on Double Descent: Rethinking Parameter Counting in Statistical Learning2024-03-07T00:00:00+09:002024-03-07T00:00:00+09:00https://hanseuljo.github.io/posts/u-turn-double-descent

For mobile: View PDF

Main References

]]>
Hanseul Cho
[Coursework Report] An Overview on Optimal Transport and its Application to Model Fusion2023-12-16T00:00:00+09:002023-12-16T00:00:00+09:00https://hanseuljo.github.io/posts/optimal-transport-final

For mobile: View PDF

Abstract

In this report, we briefly overview the theory of optimal transport (OT), entropically regularized OT, and the Sinkhorn algorithm, and their application to the deep learning model fusion technique. For the theory of OT, we delve into the derivation of the dual OT problem and the proof of the no-duality-gap result (i.e., strong duality). Through the lens of duality, we can analyze the entropic OT and its dual problem, and derive the Sinkhorn algorithm. We then provide a short survey of convergence results of the Sinkhorn algorithm and its variants. Lastly, we turn our attention to model fusion, which combines the power of several differently trained deep learning models (i.e., neural networks) into a single powerful model. We illustrate OTfusion [Singh and Jaggi, 2020], a method of aggregating several neural networks via optimal transport, and we offer a short discussion of further applications of it.

Main References

]]>
Hanseul Cho
[Paper Reading] Implicit Bias of Large Depth Networks: a Notion of Rank for Nonlinear Functions2023-09-08T00:00:00+09:002023-09-08T00:00:00+09:00https://hanseuljo.github.io/posts/implicit-bias-large-depth

For mobile: View PDF

Main References

]]>
Hanseul Cho