{"id":2828,"date":"2023-01-04T16:59:02","date_gmt":"2023-01-04T16:59:02","guid":{"rendered":"https:\/\/salarydistribution.com\/machine-learning\/2023\/01\/04\/lights-cameras-atoms-scientist-peers-into-the-quantum-future\/"},"modified":"2023-01-04T16:59:02","modified_gmt":"2023-01-04T16:59:02","slug":"lights-cameras-atoms-scientist-peers-into-the-quantum-future","status":"publish","type":"post","link":"https:\/\/salarydistribution.com\/machine-learning\/2023\/01\/04\/lights-cameras-atoms-scientist-peers-into-the-quantum-future\/","title":{"rendered":"Lights! Cameras! Atoms! Scientist Peers Into the Quantum Future"},"content":{"rendered":"
\n

Editor\u2019s note: This is part of a series profiling people advancing science with high performance computing.<\/i><\/p>\n

Ryan Coffee makes movies of molecules. Their impacts are huge.<\/p>\n

The senior scientist at the SLAC National Accelerator Laboratory (above) says these visualizations could unlock the secrets of photosynthesis. They\u2019ve already shown how sunlight can cause skin cancer.<\/p>\n

Long term, they may help chemists engineer life-saving drugs and batteries that let electric cars go farther on a charge.<\/p>\n

To make films that inspire that kind of work, Coffee\u2019s team needs high-performance computers, AI and an excellent projector.<\/p>\n

A Brighter Light<\/b><\/h2>\n

The projector is called the Linac Coherent Light Source (LCLS). It uses a linear accelerator a kilometer long to pulse X-rays up to 120 times per second.<\/p>\n

That\u2019s good enough for a Hollywood flick, but not fast enough for Coffee\u2019s movies.<\/p>\n

\u201cWe need to see how electron clouds move like soap bubbles around molecules, how you can squeeze them in certain ways and energy comes out,\u201d said Coffee, a specialist in the physics at the intersection of atoms, molecules and optics.<\/p>\n

So, an upgrade next year will let the giant instrument take 100,000 frames per second. In two years, another enhancement, called LCLS II, will push that to a million frames a second.<\/p>\n

Sorting the frames that flash by that fast \u2014 in random order \u2014 is a job for the combination of high performance computing (HPC) and AI.<\/p>\n

AIs in the Audience<\/b><\/h2>\n

Coffee\u2019s goal is to sit an AI model in front of the LCLS II. It will watch the ultrafast movies to learn an atomic dance no human eyes could follow.<\/p>\n

The work will require inference on the fastest GPUs available running next to the instrument in Menlo Park, Calif. Meanwhile, data streaming off LCLS II will be used to constantly retrain the model on a bank of NVIDIA A100 Tensor Core GPUs<\/a> at the Argonne National Laboratory outside Chicago.<\/p>\n

It\u2019s a textbook case for HPC at the edge, and one that\u2019s increasingly common in an era of giant scientific instruments that peer up at stars and down into atoms.<\/p>\n

\"LCLS<\/a>
A look at part of the LCLS instrument. (For more details, see this blog.)<\/figcaption><\/figure>\n

So far, Coffee\u2019s team has been able to retrain an autoencoder model every 10-20 minutes while it makes inferences 100,000 times a second.<\/p>\n

\u201cWe\u2019re already in the realm of attosecond pulses where I can watch the electron bubbles slosh back and forth,\u201d said Coffee, a core member of SLAC\u2019s overall AI initiative.<\/p>\n

A Broader AI Collaboration<\/b><\/h2>\n

The next step is even bigger.<\/p>\n

Data from Coffee\u2019s work on molecular movies will be securely shared with data from Argonne\u2019s Advanced Proton Source, a kind of ultra-high-resolution still camera.<\/p>\n

\u201cWe can use secure, federated machine learning to pull these two datasets together, creating a powerful, shared transformer model<\/a>,\u201d said Coffee, who\u2019s collaborating with multiple organizations to make it happen.<\/p>\n

\"Ryan<\/a>
Coffee in the \u2018projection room\u2019 where the light in his next molecular movies will first appear.<\/figcaption><\/figure>\n

The transformer will let scientists generate synthetic data<\/a> for many data-starved applications such as research on fusion reactors.<\/p>\n

It\u2019s an effort specific to science that parallels work in federated learning<\/a> in healthcare. Both want to build powerful AI models for their fields while preserving data privacy and security.<\/p>\n

\u201cWe know people get the best results from large language models<\/a> trained on many languages,\u201d he said. \u201cSo, we want to do that in science by taking diverse views of the same things to create better models,\u201d he said.<\/p>\n

The Quantum Future<\/b><\/h2>\n

The atomic forces that Coffee studies may power tomorrow\u2019s computers, the scientist explains.<\/p>\n

\u201cImagine a stack of electron bubbles all in the same quantum state, so it\u2019s a superconductor,\u201d he said. \u201cWhen I add one electron at the bottom, one pops to the top instantaneously because there\u2019s no resistance.\u201d<\/p>\n

The concept, called entanglement in quantum computing<\/a>, means two particles can switch states in lock step even if they\u2019re on opposite sides of the planet.<\/p>\n

That would give researchers like Coffee instant connections between powerful instruments like LCLS II and remote HPC centers training powerful AI models in real time.<\/p>\n

Sounds like science fiction? Maybe not.<\/p>\n

Coffee foresees a time when his experiments will outrun today\u2019s computers, a time that will require alternative architectures and AIs. It\u2019s the kind of big-picture thinking that excites him.<\/p>\n

\u201cI love the counterintuitiveness of quantum mechanics, especially when it has real, measurable results humans can apply \u2014 that\u2019s the fun stuff.\u201d<\/p>\n<\/div>\n","protected":false},"excerpt":{"rendered":"

https:\/\/blogs.nvidia.com\/blog\/2023\/01\/04\/hpc-ai-quantum-coffee\/<\/p>\n","protected":false},"author":0,"featured_media":2829,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[3],"tags":[],"_links":{"self":[{"href":"https:\/\/salarydistribution.com\/machine-learning\/wp-json\/wp\/v2\/posts\/2828"}],"collection":[{"href":"https:\/\/salarydistribution.com\/machine-learning\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/salarydistribution.com\/machine-learning\/wp-json\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/salarydistribution.com\/machine-learning\/wp-json\/wp\/v2\/comments?post=2828"}],"version-history":[{"count":0,"href":"https:\/\/salarydistribution.com\/machine-learning\/wp-json\/wp\/v2\/posts\/2828\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/salarydistribution.com\/machine-learning\/wp-json\/wp\/v2\/media\/2829"}],"wp:attachment":[{"href":"https:\/\/salarydistribution.com\/machine-learning\/wp-json\/wp\/v2\/media?parent=2828"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/salarydistribution.com\/machine-learning\/wp-json\/wp\/v2\/categories?post=2828"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/salarydistribution.com\/machine-learning\/wp-json\/wp\/v2\/tags?post=2828"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}