Quantum computers are the revolutionary next step in the computing world. This technology harnesses the quantum behaviour of atoms to process information. Without the limitations of transistors, quantum computers have the potential to outstrip traditional computers in their problem-solving ability.
Professor Lloyd Hollenberg, Deputy Director of the ARC Centre of Excellence for Quantum Computation and Communication Technology at Melbourne University, is simulating how a quantum computer behaves with the help of Pawsey Supercomputing Centre and breaking a world record in the process.
Quantum computers stores information in qubits. Roughly equivalent to a traditional computer’s bits, except with the potential to exist as both a binary 1 and 0 simultaneously. The number of qubits, their accuracy, and their ability to deal with complex algorithms are at the heart of quantum computer design.
Because a quantum computer can represent and process all possibilities of a binary string simultaneously, the longer the string, the larger the storage space a classical computer needs to match it, increasing exponentially with the length of the string. An accurate 50-qubit quantum computer generally requires several petabytes to simulate on a traditional computer.
“IBM and Google are producing systems around the 50-qubits or larger. The question is when they will get the errors down far enough for these machines to challenge conventional supercomputers,” said Professor Hollenberg.
One way to test this accuracy is to simulate a quantum computer using a supercomputer. Using Pawsey Supercomputing Centre, Professor Hollenberg’s team has created one of these simulations.
One common way to test these simulations is by having a supercomputer and quantum computer work on a random quantum circuit. If the quantum computer beats the supercomputer it has shown quantum supremacy.
Because quantum computer simulations require huge amounts of classical storage, the 50-qubit mark is considered the theoretical limit for simulating random circuits. However, Professor Hollenberg’s team wanted to understand the limits of simulating useful quantum circuits. But such simulations also require significant supercomputing resources.
To go beyond the 50 qubit mark for a quantum algorithm the team developed a sophisticated quantum computer simulator using the Pawsey Supercomputing Centre.
“We didn’t want to just simulate a random circuit because it doesn’t do anything. It’s a nice problem, but we were more interested in what the maximum useful algorithm we can simulate is. The most iconic quantum algorithm is the factoring algorithm,” said Professor Hollenberg.
This pragmatic approach, coupled with the team’s sophisticated simulation framework specifically optimised for the quantum factoring algorithm allowed them to test how a quantum computer might behave running the algorithm up to 60 qubits.
“To put that into perspective – to describe the quantum state generated in a 60-qubit random circuit instance would require some 18 exabytes of memory. It turns out the quantum factoring algorithm, on the other hand, uses quantum states very efficiently.”
Even with Professor Hollenberg’s team focused on a resource-efficient quantum factoring algorithm, they still needed several terabytes of storage space and a large amount of processing power. To solve this problem, Professor Hollenberg’s team were given a grant to carry out their simulation at Pawsey Supercomputing Centre.
Professor Hollenberg said it was a risky move to try and simulate a 60-qubit quantum computer in the finite space they had. The calculations developed and run by MSc student Aidan Dang and postdoctoral fellow Dr Charles Hill, used up the 14 terabytes and most of the CPU allocation available to them, just nudging past the previous record of 56 qubits (albeit for the random circuit problem), set by IBM.
Mr. Dang’s calculation successfully simulated how a 60-qubit quantum computer might solve the problem. It’s an important result, because the factoring algorithm is the basis for almost all our modern encryption software.
“If you take DigiCert who produce SSL encryption keys, they estimate if you had a single core trying to factor a 2048-bit encryption key it would take a million times the age of the universe. With a full-scale quantum computer, it might take days or months depending on the hardware.”
Professor Hollenberg said this was an important space to watch. While the quantum computers aren’t there yet, their potential to solve the factoring algorithm means security systems need to become quantum-proof sooner than later.
“This is one of the problems a quantum computer will gain an exponential speed-up. For a range of other nearer term problems the speed-ups may be polynomial and still very significant. We need to watch this space.”
In the meantime, simulating quantum computers allows a cost-effective way to train programmers and scientists in using the next generation of computer technology.
“There’s a huge role now for education. We’re now running our first subject in quantum computing focusing on the computing aspects rather than the physics.
It’s of great interest to people of many backgrounds. We’re teaching students from physics, maths, computer science, and engineering using our specially developed “quantum user interface” software which allows students to gain hands on experience with a quantum computer simulator. One day we hope to link this QUI to HPC resources such as those in the Pawsey Centre,” said Professor Hollenberg.
“There’s great need to educate people on the logic of quantum information processing and programming so that we have people who can build the applications in various fields and work to make the Australian industry quantum ready.”
While quantum computers still has some way to go before reaching mainstream use, Professor Hollenberg predicts quantum computers with hundreds or even thousands of qubits being utilised for specific problems in as little as five years.
With the future fast approaching, these simulations are vital for preparing for and understanding quantum computer technology. These simulations are only possible through partnering researchers like Professor Hollenberg with the resources of Pawsey Supercomputing Centre.