Computer scientists set benchmarks to optimize performance of quantum computers

Two UCLA computer scientists have shown that existing compilers, which instruct quantum computers how to use their circuits to run quantum programs, inhibit the ability of computers to achieve peak performance. Specifically, their research revealed that improved quantum compilation design could help achieve computational speeds up to 45 times faster than currently demonstrated.

Computer scientists have created a family of reference quantum circuits with known optimal depths or sizes. In computer design, the smaller the depth of the circuit, the faster the calculation can be done. Smaller circuits also mean that more calculations can be integrated into the existing quantum computer. Designers of quantum computers could use these benchmarks to improve design tools that could then find the best circuit design.

“We believe in the ‘measure, then improve’ methodology,” said lead researcher Jason Cong, distinguished professor of computer science at UCLA Samueli School of Engineering. “Now that we have revealed the large optimality gap, we are well on our way to developing better quantum compilation tools, and we hope the entire quantum research community will as well.”

Cong and graduate student Daniel (Bochen) Tan tested their benchmarks in four of the most widely used quantum compilation tools. A study detailing their research has been published in IEEE Transactions on Computers, a peer-reviewed journal.

Tan and Cong made the benchmarks, named QUEKO, open source and available on the GitHub software repository.

Quantum computers use quantum mechanics to perform a large number of calculations simultaneously, which has the potential to make them exponentially faster and more powerful than today’s best supercomputers. But there are many issues that need to be addressed before these devices can get out of the research lab.

For example, due to the sensitive nature of how quantum circuits work, tiny environmental changes, such as small fluctuations in temperature, can interfere with quantum computing. When this happens, quantum circuits are called decoherent, that is, they have lost information once encoded in them.

“If we can consistently halve the depth of the circuit through better layout synthesis, we’re actually doubling the time it takes for a quantum device to become decoherent,” Cong said.

“This compilation research could actually extend that time, and it would be the equivalent of a huge advance in experimental physics and electrical engineering,” Cong added. “We therefore expect that these references will motivate both academia and industry to develop better layout synthesis tools, which in turn will help advance quantum computing.

Cong and his colleagues led a similar effort in the early 2000s to optimize the design of integrated circuits in conventional computers. This research effectively pushed two generations of advances in computer processing speeds, using only an optimized layout design, which shortened the distance between the transistors that make up the circuit. This cost-effective improvement has been achieved without any other major investment in technological advancements, such as the physical reduction of the circuits themselves.

“Quantum processors existing today are extremely limited by environmental interference, which places severe restrictions on the duration of calculations that can be performed,” said Mark Gyure, executive director of the UCLA Center for Quantum Science and Engineering, who was not involved in this process. study. “This is why the recent research results of Professor Cong’s group are so important because they have shown that most quantum circuit implementations to date are probably extremely inefficient, and more optimally compiled circuits could allow for run much longer algorithms. This could result in today’s processors. solve much more interesting problems than previously thought. This is a hugely significant breakthrough for the field and incredibly exciting. ”