PhD student in Machine Learning at University of Geneva and University of Northwestern Switzerland (FHNW). Collaborating with @PolymathicAI
Italia
Joined August 2016
- Tweets18
- Following168
- Followers26
- Likes98
We explore diffusion models for forecasting the spatial evolution of high-latitude ionospheric electric potential patterns from solar-wind & IMF measurements.
A first step toward modeling the Sun–Earth connection with generative AI. 🌍☀️
arxiv.org/abs/2609.31804
4/ Our results highlight the potential of diffusion models for spatio-temporal forecasting in space weather and, more broadly, for generative modeling of complex physical systems.
3/ We also evaluate the models on the 2015 St. Patrick’s Day geomagnetic storm, completely outside the training period, to study how they behave under challenging, out-of-distribution conditions.
2/ The goal is not only to predict a single future state, but to investigate whether generative models can better represent the complex and uncertain spatio-temporal evolution of the ionosphere.
This becomes particularly relevant during highly dynamic space-weather conditions.
1/ We compare three approaches:
• Conditional diffusion
• Unconditional diffusion
• Deterministic U-Net
using SuperDARN electric potential maps together with DSCOVR measurements at the L1 Lagrange point: Bx, By, Bz and solar-wind velocity.
Francesco Ramunno retweeted
If you're at #ICML2026, come chat about foundation models for physics!
Posters:
• Test-time generalization for physics, led by @LouisSerrano31 (Wed 8, 5pm)
• Walrus, led by @mikemccabe210 (Wed 8, 10:30am)
Both are collaborations with @PolymathicAI and the @FlatironInst
Francesco Ramunno retweeted
Want to develop ML methods for Physics?
Check out our summer internships in the Center for Computational Mathematics at the Flatiron Institute @FlatironInst
Location: Manhattan, New York
Apply here: apply.interfolio.com/177778
Deadline: Jan 16, 2026
Francesco Ramunno retweeted
quick update: i started a blog!
my first post is live & explains the unstable learning dynamics caused by the "deadly triad" of RL through the lens of basic optimization theory. this was so much fun to learn/write about, hope yall enjoy ✨
helenqu.com/blog/posts/deadl… 🧵
Francesco Ramunno retweeted
🚀We’re looking for amazing scientists and engineers to join @PolymathicAI (NYC)!
Want to work on scientific foundation models + ML for physics, biology, astronomy, solar physics & more?
Want to contribute to frontier research in AI with the most brilliant and fun crowd?
Please sign up on our interest form:
👉forms.gle/dGvq1dhZuu1AmS646
#Hiring #MachineLearning #Science
Francesco Ramunno retweeted
Partial observability is a key challenge in predicting physical systems, where only part of the state is observed.
Check out our poster #2213 at #neurips2025 on Thu, Dec 4, 4:30pm! We propose a multiscale inference scheme for diffusion models to better predict these systems.
Francesco Ramunno retweeted
🚀We’re looking for 2026 interns at @PolymathicAI (NYC)!
Want to work on scientific foundation models + ML for physics, biology, astronomy, & more?
Want to contribute to frontier research with a brilliant, fun, and friendly team?
Please sign up on our interest form 👉 forms.gle/hxXPCv3qssEkEQJT7
#Hiring #Internship #MachineLearning
Francesco Ramunno retweeted
So proud to see the culmination of effort led by @mikemccabe210 @PayelMukhopadh3 @__tm__157 @BrunoRegaldo @FrancoisRozet along with the amazing team at @PolymathicAI to produce
The first "Polymathic"/cross-disciplinary AI model for fluid dynamics!
We started with creating The Well, the first internet scale of fluids dataset co-led by @mikemccabe210 and @oharub with the goal of creating the first cross-disciplinary fluid model!
Now it is here, and we are lovingly naming it the Polymathic Walrus 😁!
1/ Today with my colleagues @PolymathicAI, I'm excited to release our latest project, Walrus, a cross-domain foundation model for physical dynamics, into the world.
polymathic-ai.org/blog/walru…
Paper: arxiv.org/abs/2511.15684
Git: github.com/PolymathicAI/walr…
HF: huggingface.co/polymathic-ai…
Francesco Ramunno retweeted
Wanna really fundamentally change science using ML?
Come join the @PolymathicAI initiative at our Cambridge (UK) location!
Francesco Ramunno retweeted
Does a smaller latent space lead to worse generation in latent diffusion models? Not necessarily! We show that LDMs are extremely robust to a wide range of compression rates (10-1000x) in the context of physics emulation.
We got lost in latent space. Join us 👇
Francesco Ramunno retweeted
The final days of summer are upon us, and it is bittersweet to say goodbye to our great group of @PolymathicAI interns! 😭 @JacopoTeneggi @cskokgibbs @astro_nolan @CristianaD2202 @LouisSerrano31 @rachelczhang
Here are a few pics to remind us all the fun we had! (and hold your breath for our amazing list of papers led by our interns out soon!)
Francesco Ramunno retweeted
This is the first step in a direction that I am very excited about! Using LLMs to solve scientific computing problems and potentially discover faster (or new) algorithms. #AI4Science #ML4PDEs
We show that LLMs can write PDE solver code, choose appropriate algorithms, and produce stable implementations.
Please read @Shanda_Li_2000's thread for more details on our work. 🧵👇
Can LLM solve PDEs? 🤯
We present CodePDE, a framework that uses LLMs to automatically generate solvers for PDE and outperforms human implementation! 🚀
CodePDE demonstrates the power of inference-time algorithms and scaling for PDE solving. More in 🧵:
#ML4PDE #AI4Science
