Paesani Lab

🎉 Huge congratulations to our 2026 graduates, Cianna, Richa, Henry, Xuanyu, and Ethan! 🎓 We are so proud of everything you have accomplished, and so grateful for all the moments we shared along the way: the milestones and challenges, the discoveries and laughter, and everything in between. 🏄‍♀️ We can’t wait to see all the amazing things you do next! 🚀 #OnceAPirateAlwaysAPirate @UCSanDiego @UCSDPhySci @UCSDChemBiochem @HDSIUCSD @SDSC_UCSD

On the @J_A_C_S cover: A pyrrolidine dimer on Cu(100) turns a bistable switch into a controllable multilevel system. STM reveals 6 conformations & low-energy, two-step switching via intermolecular coupling. Congrats to the Li lab! pubs.acs.org/doi/full/10.10… @UCSanDiego @UCSDPhySci

Celebrating the people who keep our department running every day! 🎉 Congratulations to this year's Staff Award recipients, and thank you to all of our staff for your dedication, teamwork, and commitment to our students, faculty, and community! @UCSanDiego @UCSDPhySci

🎉Congratulations to Prof. Haiwang Yong on receiving an @NSF #CAREER Award for his theoretical and computational research on diffractive imaging of molecular structure and dynamics with vortex beams. 🥳 @UCSanDiego @UCSDPhySci

🚨 Our latest preprint is out at @ChemRxiv! 👉 chemrxiv.org/doi/full/10.26… In this work, we tackle a key software-engineering bottleneck inside #MBX: the repeated evaluation of permutationally invariant polynomial (#PIP) energy/gradient kernels. 🖥️ #PIPs are central to #datadriven #manybody potentials because they represent short-range n-body quantum effects with CCSD(T)-level accuracy. But in molecular dynamics, #PIPs and their gradients are evaluated thousands of times at every timestep, so memory layout and code generation matter enormously. ⏰ We introduce a memory-aware, #SIMD-enabled code-generation pipeline for #MBX. Starting from MB-Fit/Maple-generated #PIP routines, the new workflow: 1⃣ identifies and precomputes duplicate subexpressions 2⃣ represents the straight-line program as a directed acyclic graph (DAG) 3⃣ reorders statements to shrink the live working set 4⃣ reuses temporary memory locations as soon as intermediates become inactive 5⃣ reformats batches of inputs for AVX2 and AVX-512 #SIMD vectorization Using the MB-pol 3-body #PIP as a demanding test case, this reduces the working set by 84.5%. Even in the AVX-512 routine, which evaluates 8 PIPs per call, the optimized implementation uses 64.3% less temporary memory than the original scalar kernel. 😎 The payoff is substantial: #PIP evaluation reaches nearly ideal #SIMD speedups, ~4× with AVX2 and ~8× with AVX-512. In full MB-pol simulations, the optimized kernels improve #MBX throughput by ~1.4–3.2× depending on the instruction set and #OpenMP settings. 🏄‍♀️ This #MBX improvement moves #datadriven #manybody potentials closer to the computational efficiency of conventional force fields while also enabling CCSD(T)-level accuracy for simulations across phases. 🌊 And there is more ahead: these working-set and memory-layout optimizations are directly relevant for #GPUs, where memory traffic is often the limiting factor. Stay tuned! 🚀 Huge congratulations to Shishir and Ethan for making this happen! 🙌 Big thanks to @NSF for funding this software development! 🙏 ✅ SIMD code: github.com/paesanilab/pol… ✅ MBX: github.com/paesanilab/MBX ✅ Google group: groups.google.com/g/mbx-users ✅ Data-driven many-body simulations: mbxsimulations.com ✅ Video tutorials: @paesanilab8913/videos" target="_blank" rel="nofollow noopener">youtube.com/@paesanilab891… @UCSanDiego @UCSDPhySci @UCSDChemBiochem @HDSIUCSD @SDSC_UCSD

🚨 A great start to the @GordonConf on Molecular Interactions and Dynamics! 👉 grc.org/molecular-inte… Francesco opened the first evening session with an overview of our group's work on #ions in #water, spanning molecular clusters, bulk hydration, and aqueous interfaces. 🌊 The talk also included our latest results on the distribution of halide ions at the air/water interface, highlighting how our physics-aware #datadriven #manybody simulations help disentangle the physical mechanisms that govern ion hydration and interfacial behavior. 🫧 Preprint: 👉 chemrxiv.org/doi/full/10.26… @UCSanDiego @UCSDPhySci @UCSDChemBiochem @HDSIUCSD @SDSC_UCSD

🚨 New preprint alert! 🚨 Can deep learning redesign a protein that controls ice growth? 🧊 Our latest work is now online at @biorxivpreprint: 👉 biorxiv.org/content/10.648… We used #ProteinMPNN to redesign AFPIII, a compact fish antifreeze protein , while keeping the established ice-binding residues fixed. 🐠 Below is a short summary of the main takeaways: ✅ We generated AFPIII-like sequences predicted by #OmegaFold and #AlphaFold2 to preserve the native fold. ✅ We selected 5 designs for experiments, and all expressed efficiently in E. coli, with high soluble expression. ✅ One design, MPNN5, showed moderately improved thermostability relative to WT. ✅ All 5 designs retained measurable thermal hysteresis activity. An important result is that none of the redesigned variants reached WT antifreeze activity, even though the known ice-binding residues were preserved. 🤔 This suggests that AFPIII’s function depends on more than the canonical ice-binding site. Residues outside the established IBS may help tune hydration water, backbone rigidity, ice affinity, or resistance to engulfment by ice. Stay tuned! 😎 Big congrats to Cianna for leading this great collaboration with @BoiseState and @USNRL! @UCSanDiego @UCSDPhySci @UCSDChemBiochem @HDSIUCSD @SDSC_UCSD

🎉 A proud day for our group at the @UCSDChemBiochem End-of-Year Awards Ceremony! Congratulations to our fantastic #PhD students who were recognized for their achievements this year: 🏅 Henry — Teaching Assistant Excellence Award 🏅 Suman — Teddy Traylor Award for Research Excellence 🏅 Etienne — Chancellor's Dissertation Medal We are incredibly proud of all they have accomplished and excited to see what comes next! 🏄‍♀️ @UCSanDiego @UCSDPhySci @HDSIUCSD @SDSC_UCSD

A special lunch today! 🎉 It was wonderful to welcome back former group members Hilliary Frank (now a program manager at @NYASciences), Yuanhui Pan (now a #postdoc at @UF), and Shreya Gupta (now a #PhD student at @UW) and hear about their journeys since leaving the group. 🏄‍♀️ No matter where life takes us, we'll always be part of the Paesani Lab #pirate crew! 🏴‍☠️⛵️ #OnceAPirateAlwaysAPirate @UCSanDiego @UCSDPhySci @UCSDChemBiochem @HDSIUCSD @SDSC_UCSD

That's a wrap for Zoe and Alison! 🎓 Today we celebrated their last day in the group before they set sail for their next adventures in graduate school: 🌊 Zoe → @UMich 🌊 Alison → @UChicago Over the years, they became an important part of our pirate crew 🏴‍☠️, contributing to our research, bringing energy and creativity to the group, and making the journey more fun for all of us. We are incredibly proud of both of them and can't wait to see all the amazing things they will accomplish in the years ahead. Fair winds and following seas, Zoe and Alison! ⛵️ #OnceAPirateAlwaysAPirate @UCSanDiego @UCSDPhySci @UCSDChemBiochem @HDSIUCSD @SDSC_UCSD

What a week for our group! 🏄‍♀️ Richa and Suman represented us at the #CECAM workshop on Physics-aware Machine Learning for Molecules and Materials in NYC (cecam.org/workshop-detai…) 🖥️, Jiwon presented at #ICQC2026 in Berkeley (icqc2026.org) 🧬, and Francesco wrapped up the week in Telluride, sharing our latest advances in #datadriven #manybody simulations for #vSFG spectroscopy of aqueous systems at the Nonlinear Optics at Interfaces workshop (…-optics-at-interfaces.raiselysite.com) 🌊. It’s been exciting to connect with scientists across different scientific communities and share our latest research. 🙂 Stay tuned... it’s shaping up to be an exciting summer! ☀️🚀 @UCSanDiego @UCSDPhySci @UCSDChemBiochem @HDSIUCSD @SDSC_UCSD

And that's a wrap for #ICQC2026! 🎉 The conference ended on a high note with Jiwon (@huhji123) receiving an @AmerChemSociety PHYS Poster Award 🏆 for her research on improving the treatment of dispersion interactions, paving the way for the next generation of #datadriven #manybody potentials! 🚀 Huge congratulations, Jiwon! 🏄‍♀️ @UCSanDiego @UCSDPhySci @UCSDChemBiochem @HDSIUCSD @SDSC_UCSD

We are at #ICQC2026 in UC Berkeley! Our amazing #PhD student Jiwon (@huhji123) presented her work on a general dispersion damping scheme that recovers DLPNO-CCSD(T) dispersion without any empirical parameters. A preprint is currently under way. Stay tuned for more updates! @UCSanDiego @UCSDPhySci @UCSDChemBiochem @HDSIUCSD @SDSC_UCSD

We are at the #CECAM workshop on Physics-aware Machine Learning for Molecules and Materials cecam.org/workshop-detai… on Roosevelt Island in New York City! 🗽✨ Our amazing #PhD students #Suman and #Richa presented our latest work on #datadriven #manybody simulations for computing hydration free energies of alkali metal and halide ions 🧂💧and for probing the heterogeneity of free O–H groups at the air/water interface using #vSFG spectroscopy. 🌊⚡️ Great poster session and exciting science all around! 🎉👏 Check out their papers to learn more: 👉 doi.org/10.1021/acs.jc… 👉doi.org/10.1021/acs.jp… @UCSanDiego @UCSDPhySci @UCSDChemBiochem @HDSIUCSD @SDSC_UCSD