A quantum computer powerful enough that its circuits can’t be simulated by any classical computer, no matter how powerful, could be released as early as next year, a quantum vendor claims. UK-based Quantinuum is close to making the breakthrough, according to its founder.
Quantinuum was formed when UK start-up Cambridge Quantum merged with engineering giant Honeywell’s quantum computing division in December 2021. Based in Cambridge, it produces quantum hardware, software and security solutions. One of those products is the H1 quantum computer, a trapped-ion quantum computer. Since its inception, it has broken quantum volume records and run the first error-free algorithm on logical qubits. Its successor is the H2 with higher fidelity and more qubits. The H2 has already helped bring topological qubits closer to reality.
The current-generation machines are still very much in the noisy era of quantum computing, Ilyas Khan, who founded Cambridge Quantum out of the University of Cambridge in 2014 and now works as chief product officer, told Tech Monitor that we’re moving into the “mid-stage NISQ” where the machines are still noisy but we’re seeing signs of logical qubits and utility. Thanks to error correction, detection and mitigation techniques, even on noisy error-prone qubits, many companies have been able to produce usable outcomes. But at this stage, the structure and performance of the quantum circuits could still be simulated using classical hardware. That will change next year, says Khan.
“We think it’s important for quantum computers to be useful in real-life problem solving,” he says. “Our current system model, H2, has 32 qubits in its current instantiation, all to all connected with mid-circuit measurement.” Those circuits can still be simulated using a classical supercomputer. But that is changing. “At some point in 2024, there will be a quantum computer that simply cannot be simulated, no matter how many GPUs or core clusters or concatenation of high-performance computers you use. It literally will not be capable of simulating,” he says.
Quantum computers in physics and medicine
One use case example he gave of this next generation of quantum computers is in physics. “When people are smashing particles in the Large Hadron Collider, these applications are vast users of classical computing resources,” he says. But a quantum computer could potentially do this more efficiently, and even derive more insights from the data that’s generated. This is the point at which Khan believes “we enter a wonderful new area of science”.
He also says medicine is an area where the capabilities of true quantum computers could come into their own. Talking about the point when quantum computers have enough logical qubits to outperform supercomputers consistently, he said one use case might be a machine “big enough to do drug discovery at command”, allowing for personalised and tailored medicine for a specific health problem or patient.
Khan didn’t specify what the ‘unsimulatable’ machine would be like, but improvements in error mitigation and correction middleware are likely behind the increase in logical qubits and the higher levels of complexity. He suggested that for a machine to be unsimulatable it might need 58 or more physical qubits with all-to-all connections. That is where each qubit is connected to every other qubit to maximise the efficiency of the quantum circuits. This significantly increases complexity but allows for logical qubits to emerge.
“So 58 qubits in a system that is all to all connected, where you have mid-circuit measurement that is not post-processed, then the quantum computer cannot be simulated,” he says. At this point, we’d start to enter a new era of logical quantum computing and a point where quantum computers have a viable place in the market.
Businesses should prepare for a quantum future
Khan says his company plans to update its quantum machine, with greater numbers of logical qubits and improved error handling. It is the error solutions that allow Khan and colleagues to achieve logical qubits and operate error-free algorithms.
They will also use the next generation to verify the findings and results of calculations run on the previous-generation computer. For example, the recent use of logical qubits on the H1 quantum computer to calculate the ground energy state of a hydrogen atom and simulate the atom was verified and peer-reviewed, with findings checked by the H2.
The point where quantum computers are able to consistently perform calculations not viable or possible on a classical computer will come by the end of this decade, Khan argues. He says the reason for his prediction is that launching each new generation of quantum computer improves on the last and can help unlock more information on how to get to the point of advantage.
Talking about the unsimulatable machine, he adds: “I think that achievement allows us to discover so much more about these computers that we currently don’t know. And that then allows us to design control systems that are calibrated” and can perform ever more complex calculations.