[ad_1]
October 1, 2021 – Dr. Giuseppe Barca and his team have set a new world record for quantum chemical calculations by using a supercomputer to predict the chemical reactions and physical properties of molecular systems made up of tens of thousands of atoms.
This new chemical modeling technology is used on supercomputers around the world and is leading to new technological leaps in the fields of renewable energies, medicine and advanced manufacturing.
The previous record was also held by Barca and his team. “The 2020 record was mainly about reaching unprecedented molecular sizes,” said Barca, a lecturer at the ANU School of Computing.
“This new algorithm can not only model much larger molecular scales than before, but also with a much higher accuracy than its predecessor.”
The previous code was 99% accurate, which may sound pretty good.
“Unfortunately, it has been known for over 50 years that an accuracy of 99% is by no means sufficient to predict chemical reactions!” Said Barca. “If we want to predict a chemical conversion, we need the additional 1%.”
Barca achieved “the extra 1%” by using electronically correlated methods, which are so computationally sophisticated that they weren’t possible on a large scale until last year when the latest supercomputers went online.
The higher accuracy means that scientists can model chemical bond breaking, which will have a major impact on advances in chemical, physical, biological, and engineering sciences.
The record was achieved as part of the exascale computing project launched in 2016 as part of US President Barack Obama’s National Strategic Computing Initiative, a “nationwide effort” to maximize the benefits of exascale computing systems that will be launched in 2022 to be expected.
Exascale supercomputers will be able to process more than 10¹⸠(1 trillion) floating point operations per second, which is 5 to 8 times more powerful than today’s fastest supercomputers.
Barca is acting as partner investigator for the General Atomic and Molecular Electronic Structure System (GAMESS) team at the ECP, which includes quantum chemists and HPC specialists tasked with developing algorithms and methods to power the upcoming exa-scale supercomputers to take full advantage of it.
“These types of projects require highly multidisciplinary expertise with contributions from materials science, chemistry, biology, quantum mechanics, supercomputing and artificial intelligence,” said Barca. “This record would not have been possible without my partners.”
The GAMESS team is led by the renowned quantum chemist Dr. Mark S. Gordon of Iowa State University and includes his students Jorge L. Galvez Vallejo, David L. Poole, and Melisa Alkan. In Australia, Barca is supported by his student at the ANU School of Computing, Ryan Stocks, and Barca’s mentor at the ANU, Dr. Alistair P. Rendell, who is now the Dean of Engineering at Flinders University in Adelaide.
Barca is responsible for software development for graphics processing units (GPUs), which are the types of processors that exascale machines and many supercomputers use today. He describes their mission as “building general tools for the advancement of science using the most advanced computer technology in the world”. They will present their results in November at the Supercomputing 2021 Conference, the outstanding event for HPC.
The record-breaking calculation was carried out by the Supercomputer Summit in the Oak Ridge laboratory in Tennessee – currently the second fastest supercomputer in the world. With 27,600 GPUs, it took Summit just over 11 minutes to simulate a protein with over 45,000 atoms and 180,000 electrons. It is the largest ab initio-correlated quantum chemistry calculation to date, and its speed, accuracy and resolution surpass all previous computer experiments.
Practical applications for “The additional 1%”
The paradigm shift will come when computer simulations replace traditional chemical experiments while maintaining the same level of accuracy.
These calculations are expected to solve some of the world’s toughest problems and streamline technological advancement by reducing the time and cost of R&D processes or, if that is not possible, replacing them entirely, Barca said.
Applications include (1) the development of more efficient catalysts for the production of second generation biofuels, (2) the design of materials with high thermal efficiency for low-carbon buildings, (3) the design and production of new nanotechnologies to cure bacterial infections and to improve cancer treatments, and (4) low-cost strategies for the industrial production of nanomaterials.
Nanomaterials are chemical substances or materials that are used on a very small scale. Barca said nanomaterials have “myriad uses” including nanocrystals and quantum dots, battery materials, drug carriers and medical devices, supramolecular assemblies and soft fabrics.
A 99% accurate computational approach is already being used to test and screen drugs, including drugs used to treat COVID-19. This would be “incredibly expensive, inefficient and impossible without using high-performance computational chemistry,” Barca stated.
“Usually a drug can have an effect when it binds to a biological receptor or an enzyme. Now you can imagine that an astronomical list of drug candidates with different chemical structures can be compiled for a particular disease. Testing in the laboratory that each of these drugs binds to a receptor would be impractical, â€he said.
HPC enables simulations instead of experiments. “Because we know the physical and chemical equations that determine binding, we can simulate binding and sort out the drugs that are unlikely to be of use.”
After such a screening, “the additional 1%†becomes crucial for a safe and effective innovation.
“If the prediction is wrong, not only have we wasted money, time, equipment and our reputation, but it can lead to much worse, especially when the R&D process in the laboratory is automated. It could start a chain reaction causing an explosion or the release of a deadly gas, â€said Barca.
“To predict the correct outcome of a chemical reaction, we need to be ~ 99.99% accurate. This is because whether or not a chemical reaction actually takes place depends on the energy difference between reactants and products. If my starting materials have an energy of 1000 and my product has 999.9, the energy difference between the products and starting materials is only -0.1. Unfortunately, this difference is enough to determine the outcome of our response and how quickly it will happen. “
Cross-disciplinary collaboration in a pandemic
Before the pandemic, Barca had traveled to the US an average of 8 times a year and stayed for at least 2 weeks each time. When the number of cases skyrocketed and travel restrictions were imposed, Barca were forced to remotely manage the project.
“I have weekly meetings with my ECP partners and students in the US and meet with the PaCER staff every two weeks,” he said, referring to a related project in conjunction with the Pawsey Center for Extreme-scale Readiness in Western Australia.
ANU is the only Australian partner in the ECP project. In the US, Barca is coordinating with experts from the Ames National Lab in Iowa, the University of Texas El Paso, the Georgia Institute of Technology, Old Dominion University and EP Analytics in addition to the aforementioned Iowa State University and the Oak Ridge National Lab. ECP also uses the expertise of America’s leading HPC companies, NVIDIA, Intel, AMD, Cray.
To facilitate communication, Barca has set up two Slack channels through which information can be exchanged across institutions, disciplines and continents. Participants are invited to “post messages, chat and interact, discuss aspects of the project and anything they want to chat about”.
“Bringing these people from different academic backgrounds and backgrounds together, giving them insight into the latest developments in the projects and encouraging them to have a say in future development directions is a key aspect of our success,” said Barca.
Source: Australian National University
[ad_2]
