Lionel

Someone gave a talk on the potts model on cayley trees this week at rice university. Everytime I come cross mathematical physics work on the potts model I remember that mathematicians saw physicists having fun with their little models and decided they had to take it away... ar…

A dynamical approach to studying the Lee-Yang zeros for the Potts Model on the Cayley Tree arxiv.org

arxiv.org/abs/2608.09641 Lower spectrum of financial correlation matrices: a new perspective on market synchronization Now that's interesting, Rosanna Grassi argues that if you take the bounds of the noise bulk in the spectrum of correlation matrices,

Lower spectrum of financial correlation matrices: a new perspective on market synchronization arxiv.org

arxiv.org/abs/2607.24410 The Fundamental Structure of Risk: From Characteristics to Covariance: Predict the daily vol/cov using low frequency fundamentals which seems to deal with high frequency noise contained in daily returns.

The Fundamental Structure of Risk: From Characteristics to Covariance arxiv.org

adding to my reading list: Dependence-Informed Sparse Neural Architecture for Stock Return Prediction arxiv.org/abs/2608.143... You can match and/or outperform dense neural architecture by leverage the graphical structure in your features.

Dependence-Informed Sparse Neural Architecture for Stock Return Prediction arxiv.org

i hear that just dropping links doesn't get people's attention on here but I just want you to watch this talk Anima Anandkumar: How Do We Build AI That Understands Physics? www.youtube.com/watch?v=z_TZ...

Anima Anandkumar: How Do We Build AI That Understands Physics? youtube.com

One thing that sucks about ai: we are getting more and more single author research papers and decreased collaboration because you can reasonably push out a good number of papers by collaborating with the ai instead of humans.

anybody on here who uses multiple ai chat bots: which of them is more agreeable/disagreeable? I am spending more time using claude (past 2-3 weeks) and I find, at least in chats I've had with it, it is not overly focused on sounding agreeable at the cost of fixing my (il)logical reasoning/mistakes.

if you think ai is going in the wrong direction maybe you can work on those things which you believe would push the field in the right one. Instead of incessantly criticizing and complaining about everybody else doing work. Ship something and publish possible improvements.

I am cursed with new project ideas popping in my mind everyday which is made worse by the ai enabling my impulsive (even compulsive) behavior... 🥲

what I love and hate about ai is that this is basically a sort of complicated autoencoder/nonlinear pca except it's a foundation model:

With the creation of communities and group chats, I think the ML/AI communities on here will be better insulated from much of the vitriol on this platform from now on since those significantly improve privacy and moderation. Exciting stuff.

The way ai is changing software engineering you still test and evaluate engineers/researchers coding ability during hiring when in fact you don't want them to code in their day to day. Coding ability is getting worse not better, it's a bit hard to see this converging to anything positive long term?

Eventually users should be able to locally host frozen version of LLMs. I don't know how you avoid losing your mind suspecting model updates of performance degradation without proper evaluation methods (also why do you even have to do that to begin with?).

quick question for anybody who is into this: in the current environment when it comes to neural network model compression quantization methods are being prefered to pruning ?