Researchers at the University of Waterloo have made advances in quantum devices, including a fork-like silicon grating structure that generates twisted neutrons with angular momentum.Dusan Sarenac
The University of Waterloo last month announced the creation of a minuscule device – so tiny it can fit more than a million times over on a grain of sugar – that twisted and expanded neutron beams into 10-centimetre-wide doughnut shapes.
While most people are unlikely to care about the experimental manipulation of subatomic particles, for those immersed in quantum science the novel device was another step toward one of the ultimate goals in the field: to build a full-fledged, all-purpose quantum computer.
These machines would have abilities superior to the world’s best supercomputers, and they would be able to solve problems that are currently unsolvable. The global race to build one has been less of a sprint and more of a series of small but significant steps, combined with a few leaps.
“It’s still a long way before we have a full-functioning, full-scale computer,” says Norbert Lütkenhaus, executive director at the University of Waterloo’s Institute for Quantum Computing (IQC). “In Canada and around the world we have an emerging quantum industry full of companies that are exploring the challenges and solutions in quantum computing.”
The potential solutions are highly complex. Building the machine does not follow a standard set of instructions. Over the years researchers in different countries – including Canada, the United States, Germany, Australia, Japan and China – have been looking at various quantum computing platforms based on either trapped atoms, trapped ions, superconducting circuits or nanoscale spintronics.
Each of these approaches comes with its own set of advantages and problems.
Superconducting systems, for example, are good at mass-producing quantum bits – also called qubits, the quantum equivalent of the classical computer bit – but they must be kept at extremely low core temperatures, Mr. Lütkenhaus says. By comparison, ion-trapped qubits are highly stable and they need less cryogenics but they are harder to scale.
Regardless of the type of platform they’re based on, qubits are fragile and vulnerable to disturbances or “noise” from electromagnetic fields or material defects in the hardware. Quantum systems are also prone to errors – the result, in some cases, of qubits getting entangled when they’re not meant to be, and “talking over” each other.
While various teams of scientists continue to work through the hardware-related challenges of quantum computing, other groups have been working on algorithms that can correct errors and discern the character of system noise.
“You need to know what is really going wrong with your quantum computer and then you’ll have a better idea of how to design the most optimal algorithm,” says Tony Cubitt, a physicist and co-founder of Phasecraft, a British company that has been developing algorithms to analyze the noise in quantum hardware. “If we can figure out the character of the noise, then we can figure out the best algorithms that will work with the noise.”
Despite the hurdles, a number of companies have taken a figurative quantum leap and built small-scale machines. IBM counts 20 superconductor-based quantum systems to date and it recently released its latest model, called Osprey. At 433 qubits, it’s the world’s biggest quantum computer.
Google has one, called Sycamore, which houses a comparatively modest 53 superconducting qubits but, in 2019, it demonstrated “quantum supremacy” by performing a 200-second calculation of a problem the company said would have taken a classical supercomputer more than 10,000 years to solve.
Other companies, including Microsoft, Hewlett-Packard, Fujitsu and Honeywell are either working on research to advance their goal of building a commercially available quantum computer or they have announced plans to launch over the coming months.
In Canada, Burnaby, B.C.-based D-Wave Systems Inc. is in its fifth generation of systems that use a process called quantum annealing to solve optimization problems. Accessible through the cloud, D-Wave’s Advantage quantum computer boasts more than 5,000 qubits, it is highly specialized, and it is used by companies such as BASF, ArcellorMittal and Caixabank to optimize processes in such areas as manufacturing and investment portfolio risk management.
“We are seeing particularly compelling use cases across a wide variety of industries, including financial services, manufacturing, logistics, retail and life sciences,” says Murray Thom, vice-president of product management at D-Wave. “Our customers are seeing value right now in quantum hybrid applications from peptide design to employee scheduling to shipping container logistics to financial risk reduction.”
Roman Lutsiv, co-founder of Toronto-based financial solutions developer Adaptive Finance, says his company combines the power of quantum with artificial intelligence and machine learning to design investment portfolios that consistently outperform the S&P 500.
“We’ve been running these portfolios [with D-Wave quantum computers] for three-and-a-half years now – we have a live track record, with financial data and performance available on our website,” Mr. Lutsiv says.
Despite the growing commercial presence of quantum computers, many scientists in the field share Mr. Lütkenhaus’s view that a fully scaled, universal computer that can run any quantum algorithm is still years away.
Michele Mosca, one of IQC’s founders, says the answer to the question of “are we there yet” depends on the respondent’s definition of a quantum computer.
“If we’re talking about a fully scalable, fault-tolerant quantum computer that can break RSA 2048, then we’re not in that phase yet,” he says, referring to a currently uncrackable encryption key invented by Massachusetts Institute of Technology computer scientists Ron Rivest and Adi Shamir, and mathematician Leonard Adleman.
His opinion is shared by other top quantum experts. Mr. Mosca co-led a 2021 survey by the Global Risk Institute, a Toronto-based think tank focused on risk management for financial services. In it, a third of 47 quantum scientists from around the world expressed optimism that a quantum computer capable of cracking RSA 2048 in 24 hours would be up and running in a decade, while almost two out of three were positive it will happen in 15 years.
“Some people who say we have quantum computing today may just be talking about the physical qubit, and yes, we’ve built thousands and thousands of them on a chip,” adds Mr. Mosca, who is also president and CEO of EvolutionQ Inc. The Waterloo, Ont., startup helps companies prepare for the security risks presented by quantum computing.
“There’s also annealers – like what D-Wave has, and I think it’s a great thing what they’ve done – but they’re designed for limited applications.”
Compared with a fully scaled quantum computer, which does not yet exist, the systems around today are also small and, consequently, deliver results that are either achievable or are just slightly better than what can be generated using a classical supercomputer. Using such computers is like using a spaceship to get to a destination that can be reached on a bicycle, Mr. Mosca says.
That was the experience of the team at ProteinQure, a Toronto startup that uses computational tools to discover protein-based drug treatments. CEO and co-founder Lucas Siow says his company initially used D-Wave’s systems to run applications using a hybrid of quantum and classical computing.
“But we haven’t managed to find an advantage using quantum computers, so we’re now using classical supercompute,” he says. “When we started, we determined that we would need about 100,000 times improvement in total performance to make it worthwhile to use quantum, but it’s roughly been something like 10-times improvement.”
Mr. Mosca at IQC offers his quantum timeline prediction: a less-than-5-per-cent likelihood that the world will see a fault-tolerant, RSA 2048-cracking quantum computer within the next five years.
“Our critical infrastructure is not yet ready, so it’s more like 25-per-cent likely in 10 years and 50- per-cent likely in 15 years,” he says.