Berkeley Lab Computing Sciences

03/31/2026

A huge congratulations to Berkeley Lab's Lin Lin for being named a 2026 Society for Industrial and Applied Mathematics (SIAM) Fellow! 🎉 🎉 🎉
This honor recognizes his contributions to numerical analysis, new methods, and software for solving electronic structure problems in computational chemistry and materials sciences.

cc: UC Berkeley

03/17/2026

Last month, Berkeley Lab convened more than 70 researchers and data experts from LBNL, ORNL, SNL, PPPL, UCB, and industry to explore how better data infrastructure, partnerships, and AI-enabled workflows can accelerate scientific discovery across disciplines.

Read more about the event: https://cs.lbl.gov/news-and-events/news/2026/scidata-workshop-2026/

03/02/2026

Did you know that Scientific Data Division researchers, in partnership with ATAP scientists, will lead the American Science Cloud (AmSC) Scientific User Facilities Infrastructure Partnership, which aims to develop the platform infrastructure to host and distribute AI models and scientific data to the broader research community?

“Our partnership is helping to address the unique computing challenges and exciting opportunities encountered by experimental scientists across the DOE’s user facilities...Collaborating across dozens of projects and seven national laboratories is both challenging and rewarding. The teamwork and the opportunity to learn from each other will make it a fun and enriching experience for everyone involved,” says Paolo Calafiura, a senior scientist in Berkeley Lab’s Scientific Data Division and co-lead on the project.

Read more: https://cs.lbl.gov/news-and-events/news/2026/harnessing-ai-for-particle-accelerator-innovation/

02/20/2026

Applications are now open for Berkeley Lab's Deep Learning for Science School (DL4Sci), running from July 20-24, 2026. This intensive five-day program is designed for researchers and engineers looking to push the boundaries of their work.
This year, we're focusing on:
⚛️ Foundation Models in Science
🤖 AI for Reasoning and Autonomous Discovery
🧠 Agentic AI Systems
💻 Practical Guidance for Scaling on High-Performance Computing (HPC)
Join us for a mix of expert-led talks and crucial hands-on tutorials. This is a unique opportunity to connect with pioneers at the intersection of AI and science.
The application deadline is April 10, 2026.
Learn more and apply: https://dl4sci-school.lbl.gov/home

cc: NERSC

02/18/2026

🎉🎉🎉 Congratulations to Aditi Krishnapriyan, a faculty scientist in Berkeley Lab's Applied Mathematics and Computational Research Division, on receiving a prestigious U.S. Department of Energy Early Career Research Program award!

With this award, she will pioneer a new class of machine learning methods designed to learn the fundamental laws of science. This groundbreaking approach aims to solve complex simulation problems with unprecedented speed and accuracy, potentially reducing research timelines from months to just a week.

Learn more about her work: https://bit.ly/ECRPAK

cc: Berkeley Engineering

02/10/2026

Huge congratulations to Berkeley Lab Computing Sciences' Yinheng Tang and his team for being honored as the Overall Winner at the Machine Learning for Microscopy Hackathon! 🏆 Their award-winning project uses AI to solve a major challenge in microscopy: automatically correcting image distortions to get the clearest results possible.

Kudos to the whole team for Berkeley Lab and UCLA: Kang’an Wang, Haozhi Sha, Juhyeok Lee, and Peter Ercius.

See their award-winning work: https://kaliningroup.github.io/mic_hackathon_2/awards/

02/04/2026

Learn more about how Berkeley Lab researchers are harnessing to accelerate discovery across particle accelerators, X-ray and neutron user facilities, biotechnology, and more, building on decades of foundational research and applied expertise.

"Through long-standing AI research, advanced computation, network facilities, and data integration, Berkeley Lab is supporting the U.S. Department of Energy’s Genesis Mission​, a national effort to address challenges in science, energy, and national security" — Jonathan Carter, Associate Laboratory Director, Berkeley Lab Computing Sciences

Berkeley Lab researchers are harnessing to accelerate discovery across particle accelerators, X-ray and neutron user facilities, biotechnology, and more, building on decades of foundational research and applied expertise.

(Details via first link in comments)

📌 OPAL, the Orchestrated Platform for Autonomous Laboratories to Accelerate AI-Driven BioDesign, is using robotic systems, AI agents and models, and standardized data-sharing platforms to accelerate the biotechnology pipeline, from gene discovery to commercialized technology.

📌 SYNAPS-I, Berkeley Lab’s new AI platform, transforms petabytes of imaging data from advanced light and neutron scattering facilities into discoveries across energy, microelectronics, medicine, and more.

📌 MOAT, the Multi-Office Particle Accelerator Team, led by Berkeley Lab, is adding AI to make particle accelerators even more impactful and to help revolutionize how we do science.

Learn more about the and projects from Lawrence Berkeley National Laboratory in the links in the comments below and stay tuned for our project-specific stories, rolling out all this week.

Argonne National Laboratory
Pacific Northwest National Laboratory
Oak Ridge National Laboratory

02/02/2026

From designing life-saving drugs to building smarter AI, progress often depends on our ability to simulate incredibly complex systems. But these large-scale probabilistic models—specifically a type known as Markov chains—are often a computational bottleneck. Now, researchers at Berkeley Lab have developed a framework that directly addresses this by building a smaller, faster model that operates by the same fundamental rules as the original, ensuring accuracy is never sacrificed for speed.
This method mathematically identifies the core dynamics of a system, creating a compact model that faithfully mirrors the original's long-term behavior and output. This provides equally reliable predictions with significantly faster run-times and improved interpretability, accelerating the entire cycle of simulation, analysis, and discovery.

Learn more: https://bit.ly/MCcompression

________________________________________

01/13/2026

“Machine learning is game-changing for materials discovery because it saves scientists from repeating the same process over and over while testing new chemicals and making new materials in the lab,” said Berkeley Lab's Kristin Persson, the Materials Project Director and Co-Founder. “To be successful, machine learning programs need access to large amounts of high-quality, well-curated data. With its massive repository of curated data, the Materials Project is AI ready.”

What used to take months can now happen in days. Used more than 5,000 times a day, the Materials Project gives researchers immediate access to machine-learning ready materials data, helping speed up discovery for energy storage, quantum technologies, and advanced manufacturing.

“Accelerating materials discoveries is the key to unlocking new energy technologies.” — Anubhav Jain, Materials Project Associate Director (Link in first comment below)

NERSC
Berkeley Lab Energy Technologies Area

01/12/2026

As AI transforms the scientific landscape, high-quality, structured data is essential to unlocking its full potential. Our Scientific Data Division treats data as a first-class asset, managing the entire data lifecycle to ensure it is clean, curated, and ready for analysis. With the exponential growth of scientific data, our work has never been more crucial.

We invite you to watch our latest video and meet some of our incredible staff members.

Also, check out our new website, reflecting our commitment to innovation.

Join us in shaping the future of scientific data management and empowering AI-driven discovery! hashtag hashtag hashtag

Video: https://lnkd.in/gz_4uKVq
Website: https://scidata.lbl.gov/

01/08/2026

Can an AI generate an image that’s not just realistic, but scientifically true?
While anyone can generate a photorealistic cat, creating a valid microCT scan of a plant root that encodes real biology is a monumental challenge.
Berkeley Lab researchers have published one of the first in-depth evaluations comparing how different generative AI models tackle this very problem. Their findings are a crucial step toward a future where AI can bridge experimental gaps, reveal patterns too costly to find in the lab, and accelerate breakthroughs in materials science, biology, and energy research.
🔗Learn more: https://bit.ly/GenAI_LBNL
🧫 Read the full evaluation in the Journal of Imaging: https://www.mdpi.com/2313-433X/11/8/252

cc: U.S. Department of Energy