Training quantum computers: Physicists win prestigious IBM award

Overview

Quantum challenge completed: A team of five, headed by quantum physicist Professor Ronny Thomale of the Cluster of Excellence ct.qmat based at two universities in Würzburg and Dresden, secured second place in the international IBM Quantum Open Science Prize. The Würzburg research group managed to solve this year’s competition challenge on quantum magnetism. They devised an algorithm enabling IBM’s 16-qubit quantum chip to be trained to outperform conventional computing capabilities. This achievement could pave the way to unraveling a four-decade-long physics mystery in the realm of frustrated magnetism.

 

Theoretical physics conquers quantum computing

The global race to master quantum computing is heating up. Both the tech industry and a huge scientific community are navigating multiple avenues with one common objective: the creation of an extremely powerful and universally functional quantum computer. Quantum computing holds the potential to propel advances in areas like artificial intelligence and machine learning. Unlike conventional computers that operate using bits limited to logical states of just 0 or 1, quantum computers utilize quantum bits, or qubits for short. A qubit can exist in countless intermediate states – a phenomenon known as superposition. Until now, however, these intermediate states have been extremely fragile and decay rapidly. Researchers hope to change this by achieving stability in these quantum states and harnessing the power of numerous qubits.

 

Training with new code

In competitions such as the Quantum Open Science Prize, industry giants like IBM are on the hunt for refined algorithms to enhance their quantum tech’s efficacy. The goal of IBM’s competition this year was to code a 16-qubit quantum chip to match the reliability of its traditional counterparts. The hope is that once this benchmark is met, it could lead to quantum chips capable of computations previously deemed impossible. To aid in this endeavor, IBM provided competitors with access to the 16-qubit Falcon quantum chip. “We designed a novel algorithm and tested it again and again on the IBM chip. We were able to reserve computing time and relay the code online,” explained Dr. Pratyay Ghosh, the project lead of the award-winning five-member team from Professor Ronny Thomale’s Chair of Theoretical Physics I and a postdoc at the Würzburg-Dresden Cluster of Excellence ct.qmat – Complexity and Topology in Quantum Matter.

 

A leap forward for physics

“Normally, IT giants like IBM have their quantum technologies optimized by computer scientists due to their expertise in algorithmic matters. Yet, out of over 130 submissions worldwide, a team of budding theoretical physicists from my chair have clinched the second spot. We’re delighted! This achievement underscores the idea that the evolution from traditional to quantum computing isn’t just the purview of computer science, but should also be supported by foundational physics,” said Professor Ronny Thomale, commenting on the IBM accolade. “My team employed numerous physics-oriented coding strategies to tailor the algorithm to the challenge set. At ct.qmat, we research and develop quantum materials, and also study kagome lattices and quantum magnetism – key aspects of this year’s task. Our advantage was that we are physicists.” Currently, the team is preparing a publication that sheds light on their research from a physics perspective.

 

A rapid algorithm and two months of error correction

For the 2023 IBM Quantum Open Science Prize, participants were tasked with employing the 16-qubit IBM Quantum Falcon to precisely determine the energy of a magnetic quantum material’s ground state. The material in question is based on a kagome lattice comprising 12 atoms. “With a 16-qubit chip at our disposal to discern the ground state of a kagome star made up of 12 atoms, we promptly identified an effective algorithm for the quantum circuit,” stated Pratyay Ghosh. And he added: “The real challenge lay in sifting the genuine signal from the prevalent quantum noise, which translates to error correction. That kept us busy for two full months.”

 

A traditional computer can easily compute this year’s IBM challenge, but the outcome it produces serves as a critical benchmark for assessing the quantum computer’s coding accuracy, as Thomale points out. Inherent noise (“quantum noise”) arises when a quantum chip processes a problem. This noise stems from the characteristic fragility of superposition states, and it needs to be eliminated if quantum computing is to outperform conventional computing. “My team succeeded in minimizing the divergence of our quantum algorithm from the reference value achieved with conventional computing to below 1 percent. In this endeavor, our algorithm was fine-tuned to accurately ascertain the 12-atom kagome lattice’s ground state, despite the presence of quantum noise.” 

 

Solution on the horizon?

IBM not only evaluated the efficiency of the code, but also its scalability. First, the team had to find an algorithm that described the tiny 12-atom quantum system as well as a conventional computer. “Having accomplished that, we hope to transition our approach to larger quantum systems soon,” declared Thomale. “One of the unsolved problems in frustrated magnetism is how to determine the properties of the ground state of a kagome Heisenberg magnet.” Quantum magnetism is one of the central research areas at the Cluster of Excellence ct.qmat. As Thomale explained: "We theorize the existence of a spin liquid hidden within the kagome magnet. However, solving this quandary will require the calculation of a much larger kagome lattice.” A spin liquid describes the phase with no magnetic order of a quantum magnet. It’s a novel state of matter that could open up entirely new starting points for technologies. However, there’s a snag: “Twelve qubits aren’t enough to detect a spin liquid. We would probably need at least a thousand.”