Venn diagrams labeled “physics major” and “debater” intersect at “people who find the first talk title funny”.

Svako toliko me ovdje u dijaspori ovakav popup podjseti kako je doma bez da moram pitat.

"Snowflake is a word used by sociopaths in an attempt to discredit the notion of empathy." - John Cleese

Classical variational methods are used to simulate the dynamics of 2D quantum models with continuous degrees of freedom using neural networks, pushing the limits of classical computation in simulating these systems. Flatiron Institute, Matija Medvidović go.aps.org/3F333kX

Decided to try and be more active on this platform. A decent-looking profile pic for the occasion. 💀

We have some new work out on quantum dynamics! The question is what happens to quantum matter when you leave it on its own? We show that neural-networks can compress dynamics in cases of practical interest. You can read about our results in PRX Quantum. journals.aps.org/prxquantum/abs…

We made a code library to accelerate AI workflows with density functional theory. Check it out!

Entering the job market, I googled myself to see if the results look sensible. An unexpected find - an introductory talk about neural-network quantum states I gave at Flatiron Institute in June. It’s for ML researchers looking to learn about the topic. 😊 youtu.be/YnFjSyn5a-0?si…

Last week, I had a great conversation with Columbia Quantum Initiative about research around the quantum many-body problem, being a grad student and our recent paper: journals.aps.org/prxquantum/abs…

Anyone at NeurIPS Conference who wants to get a meal and chat? My only condition is that we don’t talk about LLMs.

I wish NeurIPS Conference was structured more like American Physical Society March Meeting.

Density functional theory is difficult. But - testing new functionals can be automatized by using automatic differentiation to interface with the standard Libxc library. I made a little package that uses JAX to quickly try new functionals in PySCF: github.com/Matematija/aut…

New preprint arxiv.org/abs/2402.01558, studying the emergence of confinement in a quenched hexagonal Ising model. Our results offer a physical explanation for the ease with which classical methods performed the IBM Eagle quantum simulation: nature.com/articles/s4158… Flatiron CCQ

Ever needed a pedagogical introduction to neural network quantum states? Javier Robledo Moreno and I have a new arXiv preprint taking you from basic principles to practical calculations. Check it out! arxiv.org/abs/2402.11014

👀

🚨"Can we simulate (ab initio!) the quantum dynamics of strongly correlated electronic systems with variational Monte Carlo?" We answer this question in our latest preprint, arxiv.org/abs/2403.07447 ! (spoiler: yes, we can! 🤯) With Giuseppe Carleo Gabriel Pescia

If I ever open a coffee shop, I will make sure to also sell fudge and call it “The Fudge Factor”.

I defended my PhD at Columbia Physics Department this week! I owe a lot to many people who have had my back, especially Dries Sels and the community at the Flatiron CCQ. I’m genuinely excited for what comes next — More announcements coming soon! 🥳

A personal update - I recently started a new position at ETH Zurich with Juan Felipe Carrasquilla Álvarez. It is nice to be back in Europe!