Thatchaphol Saranurak
@eigx.bsky.social
Assistant Professor at the University of Michigan.
I design fast graph algorithms in dynamic/distributed/local settings.
https://sites.google.com/site/thsaranurak/
I design fast graph algorithms in dynamic/distributed/local settings.
https://sites.google.com/site/thsaranurak/
I hope this is a good step towards "clarity", an essential goal in science.
This is a joint work with the fantastic team: Aaron Bernstein, Joakim Blikstad, Jason Li, and Ta-Wei Tu.
Joakim and Ta-wei coded up the algorithm.
3/3
This is a joint work with the fantastic team: Aaron Bernstein, Joakim Blikstad, Jason Li, and Ta-Wei Tu.
Joakim and Ta-wei coded up the algorithm.
3/3
October 23, 2025 at 4:47 AM
I hope this is a good step towards "clarity", an essential goal in science.
This is a joint work with the fantastic team: Aaron Bernstein, Joakim Blikstad, Jason Li, and Ta-Wei Tu.
Joakim and Ta-wei coded up the algorithm.
3/3
This is a joint work with the fantastic team: Aaron Bernstein, Joakim Blikstad, Jason Li, and Ta-Wei Tu.
Joakim and Ta-wei coded up the algorithm.
3/3
Both low-level implementation and analysis were previously very involved.
But our new paper simplifies both significantly.
Now
- Pretty readable for non-experts.
- Simple enough to code up fully in C++
- I am trying to teach it this semester arxiv.org/abs/2510.17182
2/3
But our new paper simplifies both significantly.
Now
- Pretty readable for non-experts.
- Simple enough to code up fully in C++
- I am trying to teach it this semester arxiv.org/abs/2510.17182
2/3
Combinatorial Maximum Flow via Weighted Push-Relabel on Shortcut Graphs
We give a combinatorial algorithm for computing exact maximum flows in directed graphs with $n$ vertices and edge capacities from $\{1,\dots,U\}$ in $\tilde{O}(n^{2}\log U)$ time, which is near-optima...
arxiv.org
October 23, 2025 at 4:47 AM
Both low-level implementation and analysis were previously very involved.
But our new paper simplifies both significantly.
Now
- Pretty readable for non-experts.
- Simple enough to code up fully in C++
- I am trying to teach it this semester arxiv.org/abs/2510.17182
2/3
But our new paper simplifies both significantly.
Now
- Pretty readable for non-experts.
- Simple enough to code up fully in C++
- I am trying to teach it this semester arxiv.org/abs/2510.17182
2/3
I also enjoy expositions (like some nice papers, surveys, and short textbooks).
But in my case, I put Schrijver's and Frank's combinatorial optimization books into NotebookLM. Honestly, I have only used them by looking up and never tried to read them for fun. Now, I can interact more with them.
But in my case, I put Schrijver's and Frank's combinatorial optimization books into NotebookLM. Honestly, I have only used them by looking up and never tried to read them for fun. Now, I can interact more with them.
June 7, 2025 at 8:34 PM
I also enjoy expositions (like some nice papers, surveys, and short textbooks).
But in my case, I put Schrijver's and Frank's combinatorial optimization books into NotebookLM. Honestly, I have only used them by looking up and never tried to read them for fun. Now, I can interact more with them.
But in my case, I put Schrijver's and Frank's combinatorial optimization books into NotebookLM. Honestly, I have only used them by looking up and never tried to read them for fun. Now, I can interact more with them.
3. editgpt.app has an excellent interface for polishing papers.
4. I failed to use AIs to help me solve any open problems (and even got tricked by their answers once). Have you succeeded? I am interested to hear.
3/3
4. I failed to use AIs to help me solve any open problems (and even got tricked by their answers once). Have you succeeded? I am interested to hear.
3/3
June 7, 2025 at 2:48 AM
3. editgpt.app has an excellent interface for polishing papers.
4. I failed to use AIs to help me solve any open problems (and even got tricked by their answers once). Have you succeeded? I am interested to hear.
3/3
4. I failed to use AIs to help me solve any open problems (and even got tricked by their answers once). Have you succeeded? I am interested to hear.
3/3
1. For the literature search, I enjoy using the Deep Research of both ChatGPT and Gemini. (I still often go through the rabbit hole in Google Scholar.)
2. I upload many textbooks on the same topic to NotebookLM and ask questions. It is a fun way to learn and look up.
2/3
2. I upload many textbooks on the same topic to NotebookLM and ask questions. It is a fun way to learn and look up.
2/3
June 7, 2025 at 2:48 AM
1. For the literature search, I enjoy using the Deep Research of both ChatGPT and Gemini. (I still often go through the rabbit hole in Google Scholar.)
2. I upload many textbooks on the same topic to NotebookLM and ask questions. It is a fun way to learn and look up.
2/3
2. I upload many textbooks on the same topic to NotebookLM and ask questions. It is a fun way to learn and look up.
2/3