Great to see this finally out!! Congrats to all coauthors!🍾

Metals owe their properties to how local defects emerge & amplify in collective dislocations under stress.🛠️ 👉We show how tracking local atomic fluctuations & their space&⏱️correlations allows tracking metals' behavior through the elastic&plastic phases🚀 pubs.aip.org/aip/jcp/article…

How much complexity is needed in self-assembling molecular systems to observe non-trivial emergent behaviors typical of more complex, higher-scale systems?🤯 Not much!😲 See @NatureComms our work on the collective resilience of supramolecular polymers!🚀 nature.com/articles/s41467-0…

#LEAP is out @PNASNexus!🚀 Building on abstract concepts of local fluctuations & their correlations in space & ⏱️, #LEAP provides info on the physics of complex dynamical systems from the atomic- to the macro-scale in agnostic & purely data-driven way!🤩 academic.oup.com/pnasnexus/a…

When studying complex systems, common belief is that high-dimensional analyses are desirable to prevent losing important information... but to what extent this is really needed/beneficial remains often unclear.😵‍💫 Now @arxiv we challenge this assumption: arxiv.org/abs/2412.09412v2 🚀

Big congrats to @MatteoCioniMC on his PhD defense! 🎉✨ After years of hard work @PoliTOnews, Matteo's brilliant research in molecular dynamics & complex systems has pushed boundaries. 🌌💻 We're proud to celebrate this huge milestone with him! 🚀👨‍🎓 #PhDDefense #Science

The choice of descriptors is key for extracting information from data.🧐 Look at @SimoneM118 work! arxiv.org/abs/2411.12570 While advanced descriptors may rely on higher signal-to-noise, we show how even the simplest descriptor may become super efficient upon local denoising! 🚀

Finally out in @PNASNews “Onion 🧅 Clustering” 🚀: an efficient & essentially parameter-free unsupervised clustering method that can classify statistically relevant fluctuations & microscopic dynamical domains in noisy timeseries data of any kind!!🤩🥳 pnas.org/doi/10.1073/pnas.24…

Recent simulations showed non-trivial dynamics in metals. But these often remain hard to prove experimentally. Here we combine for the 1st time experimentally-resolved Au-NP structures with MD simulations & #ML, proving their real-time atomic dynamics! onlinelibrary.wiley.com/doi/…

😎: Onion Clustering does not attempt to fit all data in a time-series into clusters! Depending on the resolution, dt, it classifies all the information that can be classified in statistically robust way & store the "undetermined" information in the ENV0 cluster! 🚀🤯🔥

Tracking statistically-relevant fluctuations in noisy time-series data is key in many fields, from #MachineLearning, to the study of signals & complex systems! Use Onion #Clustering🧅: easy, unsupervised, physically-interpretable, statistically-robust!🚀🤯 arxiv.org/abs/2402.07786

Fresh @ J. Chem. Eng. Data the new work of @mattiaperr! 🥳🤩 We tested SwarmCG to automatically optimize rigid 3-site water models onto multiple microscopic (exp. g(r))+macroscopic thermodynamic properties, hitting the physical limits of such models!🧐🤯 pubs.acs.org/doi/10.1021/acs…