Google says it has taken a decisive step towards the future use of quantum computers. Their new chip, Willow, achieves enormous speeds with much-needed error correction.
Hartmut Neven, founder and head of Google Quantum AI, writes in a blog post that the new quantum chip, Willow, can solve an algorithm in five minutes that would take the Frontier supercomputer (currently the fastest supercomputer in the world, ed.) more than 10,000,000,000,000,000 billion years to solve.
According to Neven, the Willow chip achieves two milestones in quantum technology research. First is the computing power. The system contains 105 qubits (the specific components that make up a quantum computer) and is said to achieve unprecedented speeds in solving the mathematical problems used as a ‘benchmark’ for these systems.
Now, with less errors
Second, perhaps more importantly, the chip does this with a much lower error rate than many other quantum computers. Qubits are very unstable. The particles can take on the traditional state of 0 or 1 but also a state that spans both. However, this makes them very sensitive to their environment, and they quickly lose their ‘charge’. Where traditional computers make almost no calculation errors, a quantum computer can make an error every hundred or thousand calculations, and that error rate increases as more qubits are added to a system.
Willow implements error correction by joining multiple qubits together. Google links qubits in a square, for example,e 3×3, together to form a ‘logical qubit’. If one is affected by the environment, the system can compare that state with the other qubits in the square and correct where necessary.
15×15
In a paper in the scientific journal Nature, Neven writes that his team managed to reduce the amount of errors as more qubits were added to the system through error correction. If you go from 3×3 linked qubits to 5×5, the system can correct twice as many errors, and from 5×5 to 7×7 qubits, that correction doubles again. This would be the first time that qubits have been built that are stable enough to effectively contribute to error correction as more of them are added.
Neven emphasizes that with the method used and the new chip, scalable quantum computers are possible, although the error margin still needs to be reduced if we want an effectively usable quantum computer. The largest logical qubit that the hardware can handle, 15×15, for example, still only lasts a maximum of an hour before it starts to fail due to an excess of errors.
Quantum computing is a technology that major companies like Google, IBM, and others have been working on for years. When perfected, it could achieve unimaginable speeds in solving specific problems that traditional computing currently struggles with.