Nikos Karalias

DiffCoALG@NeurIPS 2025 About this Workshop Combinatorial algorithms are fundamental across a wide range of domains, owing to their ability to model optimization and decision-making tasks under complex constraints. These alg...

Can LLMs reason about combinatorial problems?
Join DiffCoALG @neuripsconf.bsky.social 2025 — where discrete algorithms meet differentiable ML.

📅 Deadline: *extended* to Aug 31 (AOE)!
🔗 sites.google.com/view/diffcoa...

DiffCoALG@NeurIPS 2025 About this Workshop Combinatorial algorithms are fundamental across a wide range of domains, owing to their ability to model optimization and decision-making tasks under complex constraints. These alg...

🔥 New workshop at @neuripsconf.bsky.social!
DiffCoALG bridges the gap between classic algorithms & differentiable learning.
Think: LLM reasoning, routing, SAT, MIP — neurally optimized.
📌 Submit by Aug 22! 🤖🧠
🔗 sites.google.com/view/diffcoa...
#NeurIPS2025

DiffCoALG@NeurIPS 2025 About this Workshop Combinatorial algorithms are fundamental across a wide range of domains, owing to their ability to model optimization and decision-making tasks under complex constraints. These alg...

🔥 New workshop at @neuripsconf.bsky.social!
DiffCoALG bridges the gap between classic algorithms & differentiable learning.
Think: LLM reasoning, routing, SAT, MIP — neurally optimized end-to-end.

📌 Submit by Aug 22!
🔗 sites.google.com/view/diffcoa...
#NeurIPS2025

Totally admit I may have not kept up with people's opinions though 😅

I feel like there are quite a few top people in TCS with "well adjusted" opinions, no?
Pipeline makes it sound like a regular occurrence.

Does the subfield of TCS may matter? I can't think of many bad examples of top people working in complexity theory right now who are agi-brained.

New paper: Simulating Time With Square-Root Space

people.csail.mit.edu/rrw/time-vs-...

It's still hard for me to believe it myself, but I seem to have shown that TIME[t] is contained in SPACE[sqrt{t log t}].

To appear in STOC. Comments are very welcome!

This sounds like it could take found footage (horror) movies to the next level. I'm definitely sold :D

Yea I thought of Rashomon as well :D

I mean make it in a similar style to Eno. Would be cool to see people collaborating online to piece it together. Seems like there's a lot of potential in this direction 😅

Yea fair. I like the concept quite a bit. Audiences are encouraged to discuss the movie with others to get additional content out of it. Wonder if it's possible to make a "typical" whodunnit movie where you have several versions of the film and you need to watch all of them to figure it out.

Definitely enjoyed the overall vibe. I think that worked. But a lot of things have changed in music over the past 50 years and with the clips they showed I couldn't quite get a feel for how his own approach and attitude towards all the new developments (be it sampling or whatever) may have evolved.

There were a few parts that felt out of place, and I would've liked to see more stories from the 2000s and onward. Most of the stories in my screening were in the 70s-90s range. In any case, I'm really curious about how much the content varies.

I watched it yesterday so here's my update: Overall enjoyable watch. I like the glitchy aesthetic and the way the stories were presented worked better than I would've expected. On the other hand, it was a bit all over the place so I couldn't quite feel any sense of progress in the movie.

Wasn't aware of this so ty for posting. Just got a ticket for the Boston screening!

I hate to be that person, but I've found chatgpt to be useful in some cases (definitely not a consistent thing) that are somewhat like this; i.e., when I'm looking for slightly more 'esoteric' math. Even if incorrect, it may sometimes drop useful terms or sentences to fuel further googling.

Has to be "mischievous". For whatever reason a lot of people pronounce it "mischieveeous".