This month, Summit Early Science Program users are starting to work on some of the world’s toughest science problems on its most powerful supercomputer: the 200-petaflop, IBM AC922 Summit system at the Oak Ridge Leadership Computing Facility (OLCF). The OLCF is a US Department of Energy Office of Science User Facility located at Oak Ridge National Laboratory (ORNL).
Not only is Summit significantly expanding the capabilities of modeling and simulation—from tracking the elements released in star explosions to virtually testing future fusion reactors—the system is taking researchers into new territory enabled by artificial intelligence (AI). From uncovering unseen patterns in cancer data to creating deep learning networks for scientific image analysis, 2019 is poised to be a groundbreaking year for applying AI algorithms to intractable data science problems. In this article, we preview just 19 of the more than 30 critical science topics and simulations researchers will be tackling on Summit during the Early Science period.
View the complete list of the Early Science projects here.
- Evolution of the universe
To investigate big questions in modern cosmology, including the role of dark energy in the acceleration of the universe and the distribution of unseen dark matter, researchers will simulate a “virtual universe” on Summit that can be combined with observational results.
Project: Frontier Precision Cosmology with HACC
Principal investigator: Salman Habib, Argonne National Laboratory
- Whole-cell simulation
Scientists will use sophisticated experimental data on cellular structures like organelles in atomic detail to simulate the dynamics of what will likely be the first computationally modeled “protocell” that incorporates a cell’s essential features from the atomic to the cellular scale.
Project: Protocell: Petascale Simulation with NAMD and VMD Helps Understanding Cells at the Atomic Level
Principal investigator: Emad Tajkhorshid, University of Illinois-Urbana Champagne
- Inside a nuclear reactor
Researchers will simulate an operating nuclear reactor and compare the results with operational data from a real nuclear reactor plant. Summit will enable the team to simulate details that could not previously be computed, contributing to a national effort to extend the lifetimes of nuclear reactors.
Project: Full Power Simulation of the Watts Bar Nuclear Reactor using the Shift Monte Carlo Transport Solver
Principal investigator: Steven Hamilton, ORNL
- Post-Moore’s Law graphene circuits
Scientists will explore potential metals for components in atomically precise graphene nanoribbon circuits that could offer an alternative to traditional silicon-based computer circuitry, which is predicted to become too small by the 2020s to work reliably.
Project: Nanoscale Design of Contacts to Atomically Precise Graphene Devices
Principal investigator: Jerry Bernholc, North Carolina State University
- A critical point in the formation of matter
To help experimental researchers at Brookhaven National Laboratory’s Relativistic Heavy-Ion Collider find the critical point at which the particle “soup” of quarks and gluons present following the Big Bang coexists with matter as we know it (called hadronic matter), researchers will carry out fundamental physics calculations on Summit that require extreme computational power.
Project: Hot-dense Lattice QCD for RHIC Beam Energy
Principal investigator: Swagato Mukherjee, Brookhaven National Laboratory
- The cell’s molecular machine
Adenosine triphosphate (ATP) synthase is a protein found in photosynthetic cells that is extremely efficient at converting light energy into the cellular fuel ATP through tiny, atomic motions. By simulating ATP synthase on Summit, scientists might help guide the design of bio-inspired solar energy devices.
Project: All-atom simulations of motor proteins for cellular energy metabolism
Principal investigators: Abhishek Singharoy, Arizona State University
- Unpacking the nucleus
On Summit, scientists will use a computational approach based on the strong force that binds subatomic particles, known as lattice quantum-chromodynamics, to make calculations important to experimental searches aimed at uncovering new knowledge about the nucleus, such as the proton radius and potentially undiscovered states of matter at the subatomic level.
Project: Hadrons, nuclei and fundamental symmetries
Principal investigator: Robert Edwards, Thomas Jefferson National Accelerator Facility
- Mars landing
To advance human exploration of the Red Planet, a NASA team will simulate flow between the Martian atmosphere and descending rocket exhaust.
Project: Enabling Human Exploration of the Red Planet
Principal investigator: Eric Nielsen, NASA
- Deep learning for microscopy data
Scientists are using an ORNL-developed AI system called MENNDL to automatically create deep learning networks that can rapidly extract information from electron microscopy data. Electron microscopy is an important tool for nanofabrication (atom-scale manufacturing), which is already being used in the development of new consumer devices, medicines, electronics, and more.
Project: Scalable Machine Learning of Scientific Data
Principal investigator: Robert Patton, ORNL
- Elements from star explosions
To understand the quantity and dissemination of elements expelled from supernovae, astrophysicists are using Summit to model nuclear burning with about 10 times more elements than previous state-of-the-art simulations. These massive simulations require computing at multiple scales, from large-scale fluid motion (hydrodynamics) calculations to small-scale particle interactions.
Project: Modeling Stellar Explosions and Their Nucleosynthesis with an Optimized FLASH Code
Principal investigator: Bronson Messer, ORNL
- Cancer data
Using AI, researchers are training computers to extract important information from large volumes of clinical text and biomedical documents on cancer. Such information can be used to help doctors determine the best treatment for each patient and improve overall population health outcomes.
Project: Exascale AI to Advance Health Using Big Heterogeneous Biomedical Data
Principal investigator: Georgia Tourassi, ORNL
- Earthquake resilience for cities
To improve earthquake prediction for cities, researchers will use the computational power of Summit to couple the shaking of the ground with building structures in the same simulation while also modeling new physical features.
Project: Low-order Unstructured Finite-element Earthquake Simulation on Summit
Principal investigator: Kohei Fujita, University of Tokyo
- The nature of elusive neutrinos
Scientists worldwide are studying the properties of the neutrino—a neutral subatomic particle that is nearly massless yet abundant in the universe and, once better understood, that could help answer unsolved problems in physics. A team of nuclear physicists are using Summit to generate an important benchmark for neutrino studies by computing a hypothetical type of nuclear decay known as neutrino-less double-beta decay in a calcium-48 nucleus.
Project: The neutrino-less double beta-decay of calcium-48
Principal investigator: Gaute Hagen, ORNL
- Extreme weather detailed with deep learning
To study extreme weather patterns like hurricanes at new levels of detail, researchers will use Summit to explore the application of deep learning to climate data analysis, which involves finding meaningful patterns in massive datasets.
Project: Exascale Deep Learning
Principal investigator: Prabhat, Lawrence Berkeley National Laboratory
- Flexible, lightweight solar cells
On Summit, researchers will model energy-converting processes in organic materials, which could guide designs for highly efficient organic photovoltaic devices that could be competitive with traditional solar cells.
Project: Organic Photovoltaic Materials Design Using the GronOR Non-Orthogonal Configuration Interaction Software
Principal investigator: Remco Havenith, University of Groningen
- Virtual fusion reactor
Plasma physicists are using Summit as a “virtual fusion reactor” to model the behavior of plasma—the hot gas medium in which particles generate fusion energy. Understanding plasma behavior is critical for fusion experiments like ITER, which will explore how fusion can help sustainably meet growing global energy demand.
Project: Using XGC to predict ITER’s boundary plasma performance and its impact on fusion efficiency
Principal investigator: C. S. Chang, Princeton Plasma Physics Laboratory
- Unpredictable material properties
Materials scientists will simulate coupled structural and electronic phases using, for the first time, the quantum mechanics-based method Quantum Monte Carlo. By combining these two phases in transition metal oxides, known for their unpredictable yet useful properties, researchers aim to demonstrate structural optimization of these materials.
Project: Structurally complex oxides with Quantum Monte Carlo
Principal investigator: Paul Kent, ORNL
- Genetic clues in the opioid crisis
In 2017, opioids contributed to more than 49,000 overdose deaths in the United States. In midst of this national public health crisis, researchers will use Summit to study complex genetic interactions that lead to physical traits, such as how people develop chronic pain and respond to opioids. The results could help inform treatment for patients predisposed to substance abuse and other conditions.
Project: Attacking the Opioid Epidemic: Determining the Epistatic and Pleiotropic Genetic Architectures for Chronic Pain and Opioid Addiction
Principal investigator: Dan Jacobson, ORNL
- Turbulent environments
On Summit, researchers will explore how combustion takes place in turbulent environments such as gas turbines or car engines using a fluid dynamics solver that incorporates the multiscale and multiphysics processes at play in these systems.
Project: First Principles Investigation of Turbulent Scalar-Mixing and Combustion in Supercritical Fluids
Principal investigator: Joseph Oefelein, Georgia Tech
And that’s not all
Beyond the Summit Early Science Program, scientists from around the world are already beginning to access Summit through the 2019 DOE INCITE program, and there is still an opportunity for researchers to submit proposals for the 2019–2020 Advanced Scientific Computing Research (ASCR) Leadership Computing Challenge (ALCC) program, which focuses on projects that align with the DOE mission to ensure America’s security and prosperity by addressing its energy, environmental, and nuclear challenges through transformative science and technology solutions.
The ALCC call for proposals closes February 13, 2019. Visit the DOE Advanced ScientificComputing Research website to learn more.