iPhone 17 Pro is the ugliest iPhone ever

A very exciting and lucrative force in the Indian economy in the next decade is going to be the formalisation of many sectors...

Stability is an illusion

The Bootstrap of Points and Lines We study two-dimensional conformal field theories (CFTs) with boundaries via the conformal bootstrap. We derive a positive semi-definite program from crossing symmetry of three observables: the annulu...

It’s up on arXiv! It took 3.5 years of work

Man I had forgotten how rad Riemannian geometry is!

It’s upload to arXiv day 🥳

No. Models are just matrices. So pure numbers. Even though it's a PITA, you don't need to use their code to run the model if you have the weights available. You can write your own code

La fameuse fenêtre d’Overton

Should I be worried that 2025 literally starts with WTF?

Looking back, great teachers are really important early on in one’s career.

I chose physics because I had a great HS physics teacher. I chose theoretical physics because I had a great LinAlg/group theory instructor in first year of UG. These two combined made me want to pursue very mathy physics

Now that I have been at Caltech for 2 months, I can confidently say that I have been a part of the academic community in the US, Europe, and India.

And I can see why the US is leaps and bounds ahead in many areas of research

Time and time again, I keep saying the same thing - people will always catch up to your model’s performance quickly. The only real differentiator will be the quality of the data you train on