Alessandro Caruso

Postdoctoral researcher in AG Clementi at FU Berlin.

[Alessandro Caruso @acaruso0]

Big thanks to Jacopo for the great collaboration. Thanks to Lorenzo, Edoardo, and @ZakMachachi for their precious help, and to @FrankNoeBerlin and @CecClementi for their support. Funded by DFG, @MATHplusBerlin, @NSF, @Einstein_Berlin and BMBF. 3/3

[Alessandro Caruso @acaruso0]

Excited to have RANGE published in @NatureComms 🥳 Oversquashing kills long-range information in GNNs. RANGE extends the reach via attention-based virtual nodes, consistently outperforming local MPNNs, with minimal overhead and linear scaling. Paper: https://t.co/yXm4q9CJ2P 1/3

[Alessandro Caruso @acaruso0]

Check out the new extended results: equivariant baselines (PaiNN, SO3krates, MACE), long-range extrapolation tasks, comparison with Ewald-based methods, and stable MD simulations on systems with up to 370 atoms. Code: https://t.co/Gfhubrg3hb 2/3

[Alessandro Caruso @acaruso0]

@atAndreasBurger @CecClementi @FrankNoeBerlin Ciao Andreas! I would argue that you couldn't get the same result by 'just' adding capacity to the node vectors in the message-passing step. RANGE is applied after and independently from the MP step, and it represents a re-wiring of the network topology beyond the MP reach.

[Alessandro Caruso @acaruso0]

We just shared our latest work with
@CecClementi and @FrankNoeBerlin!
https://t.co/nk6gRfhHss
RANGE addresses the "short-sightedness" of MPNNs via virtual aggregations, boosting the accuracy of MLFFs at long-range with linear time-scaling.
Next-gen force-fields are here! 🧠🚀

[Alessandro Caruso @acaruso0]

@CecClementi @FrankNoeBerlin It was a pleasure to work with Jacopo on this. I thank Lorenzo and Edoardo for their precious help, and @Einstein_Berlin, DFG, and BMBF for funding!

[Alessandro Caruso @acaruso0]

@sabrinaesaquino @ThePrimeagen @ThePrimeagen Berlin mentioned!!

[Cecilia Clementi @CecClementi]

We are looking for a talented and motivated postdoc to work in a collaboration between @FrankNoeBerlin and my group at Freie Universität Berlin on the development of machine-learned coarse-grained protein models. Please email [email protected] if interested.

[Cecilia Clementi @CecClementi]

I have several positions available in my group at Freie Universität Berlin. Come work with us on the combination of ML and stat mech for the development of protein models at different resolutions and cool applications in collaboration with experimentalists!

[Alessandro Caruso @acaruso0]

Campus is beautiful! 🌳

[Alessandro Caruso @acaruso0]

Thrilled to embark on my new journey as a #PostDoc @FU_Berlin, joining the fantastic team of @CecClementi!
Ready to explore the unexplored, pushing the frontier of #coarsegraining and #machinelearning... Let's begin!

[Alessandro Caruso @acaruso0]

I am so grateful to @PaesaniLab and all my friends @UCSanDiego for the incredible journey towards my #PhD! I will soon leave #SunDiego 🌅 to start a new adventure in #Berlin. Excited to explore new horizons, and push the boundaries of #theochem and #machinelearning!
#pirates 🏴‍☠️

[Paesani Lab @PaesaniLab]

🎉 Congratulations to Alessandro! He's navigated the #compchem seas & defended his #PhD like a true #pirate. Time to pass #Quelo, our symbol of camaraderie & nontraditional exploration, to Hank. We're all thrilled to see what uncharted seas Alessandro will explore next!
#PhDone🎓