AMD technology is helping Cornell University unlock discoveries in the world of physics.
Dr. Lawrence Gibbons, a professor in the department of physics at Cornell University, has had an interest in science from a very early age. A self-proclaimed "Trekkie," Gibbons says he was really fascinated by physics as a child and was reading about black holes in early junior high. "I was interested in physical phenomenon and started pursuing it," he said.
While in college at the University of Chicago, Gibbons said he was greatly influenced by one of his professors who was involved in particle physics. He was exploring one aspect of a phenomenon known as "charge parity (CP) violation" -- an area of study that compares a mirrored universe (parity) and the charge conjugation of matter to the physical universe and looks for measurements that could distinguish that universe from ours.
"Almost all lab experiments consistently produce equal amounts of matter and antimatter, so it would stand to reason that the Big Bang should have created equal amounts of matter and antimatter," Gibbons explained, "yet we live in matter-dominated universe. We don't see a corresponding amount of antimatter anywhere." Trying to understand and explain the reason for this phenomenon has been at the heart of Gibbons' work.
In 1967, Andrei Sakharov, a Soviet nuclear physicist, published a paper proposing certain conditions at the time of the Big Bang, including some amount of charge parity (CP) symmetry violation, that could have resulted in a universe that slightly favors matter over antimatter.
"That whole question fascinated me and got me involved in particle physics," Gibbons explained.
"I've spent my career trying to understand more about the structure of the weak interactions in general, and how CP violation fits into the picture. We have a model that can accommodate what we see in the lab, but it turns out that the level of violation we get in a lab is nowhere near good enough to explain why we have a matter-dominated universe."
Measuring Magnetic Moments
The project that Gibbons and his team at Cornell University is working on is part of a long-standing series of experiments that calculate the "magnetic moment" of either electrons or "muons" (particles formed in the upper atmosphere that look like electrons but are 200x heavier).
"Electrons and muons are charged and have momentum with their internal spin that resembles the behavior of a tiny bar magnet," Gibbons said. "Our goal is to determine the strength of those bar magnets."
Gibbons said that when radar technology emerged, it made a whole class of precision experiments technologically feasible. There were experiments done to measure the magnetic moment of electrons.
And this work led to Julian Schwinger's invention of quantum field theory -- what we use today to describe particle interactions at the fundamental level.
"What is happening now is purely a quantum mechanical effect," Gibbons said. "Muons can, for a brief moment, emit and reabsorb a particle like a photon, or have loops of charged particles that blink in and out of existence. The whole process shifts the magnetic field a bit, and there have been multiple experiments to measure the size of those shifts."
Gibbons said that in around the year 2000, the Brookhaven National Lab measured the deviation of the magnetic moment of muons from two to about 500 parts per billion. The research suggested a significant discrepancy between Brookhaven’s results and the theoretical prediction.
"If you can imagine there are new kinds of particles that could eventually couple back to the charge of muons, you can imagine they would, indeed, shift things in the direction that we see. We had great interest in knowing if Brookhaven's result was right. Also, we wanted to significantly improve precision so we could make an even finer comparison with theory. That's what launched us," Gibbons said. "Today, we are trying to do a factor of 4X improvement over what Brookhaven has done. We have spurred a worldwide effort among our theory colleagues along the way."
More than 150 researchers and 100 theorists are currently involved in the project, Gibbons said.
The A-Ha! Moment
Cornell University's particle research is performed at Fermilab, the Fermi National Accelerator Laboratory run by the U.S. Department of Energy, and it is aimed at measuring the strength of the magnetic field of the muon. There are 191 collaborators from 35 institutions in seven countries involved in the project.
Last Spring, the team rocked the scientific world when its research findings confirmed the results of the earlier Brookhaven research, and at the same level of precision that had been previously observed. "This finding is important because it's telling us that the magnetic field is a bit stronger than theory is telling us it should be, given the particles we know about," Gibbons said. "The jury is still out as to whether there is a statistically significant effect or not."
Gibbons said that sometime later this year, the team should be coming out with its next result, based on a factor of 4X more data than its previous findings, and higher levels of precision. The final results will come a couple years from now.
"If in the end we can demonstrate -- once we get to our final precision-- that we disagree with the theoretical prediction, it means that we have definitive evidence that there has to be a new kind of fundamental force (beyond known forces like gravity or electromagnetism) that we had no inkling of before, and there are particles associated with it -- dark matter -- that we don't know the characteristics of."
This could fundamentally change our understanding of the universe and laws of physics.
Gibbons said the moments leading up to the reveal of test results was tense.
"To avoid biasing ourselves, we had someone not involved in the experiment slightly change the clock frequency and record it in a sealed envelope," Gibbons said. "We did all of our calculations and analysis with digitizers using a placeholder frequency that we ran the experiment at. It was only until everyone was 100 percent committed to publishing what we had that we asked the people holding the true frequency value tell us what it was so we could see where our results would land. We had no idea if we were going to be in the ballpark of the old measurement or of the theoretical prediction or be somewhere else way off the map. We were confident we had solid results, but where it would land was unknown. Those seconds waiting for them to type in the true frequency value were nerve-wracking. Everybody was cheering when they saw the result. It was fun and exhilarating but also nerve-wracking leading up to it."
How did AMD Help in this Discovery?
"AMD technology has been at the heart of our research for many years," Gibbons said. "We were designing digitizers with a lot of ADCs (analog-to-digital converters) and needed traffic 'cops' to manage the data flow. We needed to be able to move a lot of data around at 800 megasamples per second, and there were about 1,300 channels we were instrumenting. We needed FPGAs to move the data out of the ADCs and into data buffering so we could asynchronously read out the data. We also needed a master FPGA on each board that could control data flow and deal with the triggering and monitoring information we were passing in and out of the digitizers. AMD was able to meet these requirements and more."
Today, there are close to 2,000 AMD Kintex™ 7 FPGAs being used in the experiment, the majority of which control data acquisition, and others that manage data flow and communication with the outside world. Kintex 7 devices provide cost-effective access to high transceiver counts in a small package. Gibbons said that the longevity and reliability of AMD products has been crucial to their work.
"AMD's FPGAs are at the heart of the digitizers we use to take in all of our research data," Gibbons said. "They have been running continuously for more than seven years and are working flawlessly. The Kintex series turned out to be a nice match for price and came with a lot of horsepower on top of it. The Vivado™ tools have also worked out well and have provided a nice training ground for students and even me. I very much enjoy working with AMD. We have a very nice working relationship all around."
©2023 Advanced Micro Devices, Inc. All rights reserved. AMD, the AMD Arrow logo, Epyc, Ryzen, Radeon, Xilinx, Zynq, and combinations thereof are trademarks of Advanced Micro Devices, Inc. Other product names used in this publication are for identification purposes only and may be trademarks of their respective companies. PID #1671659. All performance and cost-savings claims are provided by Dr. Lawrence Gibbons at Cornell University and have not been independently verified by AMD. Performance and cost benefits are impacted by a variety of variables. Results herein are specific to Cornell University and may not be typical. GD-181.
2 Comments
