Google is racing to create the first quantum computer capable of solving a problem ordinary computers cannot聽鈥� and it has just made that challenge much harder.
Achieving 鈥�quantum supremacy鈥�, as it is known, involves building a device that can solve a problem faster than any non-quantum computer.
In 2016, a team led by Sergio Boixo at Google published a paper showing that even with state-of-the-art supercomputers, it would be infeasible to simulate the behaviour of a quantum circuit with 48 qubits (quantum bits) at a circuit depth of 40, a measure of the problem鈥檚 complexity.
Advertisement
Computer competition
But now, Boixo and his colleagues have devised a classical, or non-quantum, algorithm that can simulate a quantum circuit with 49 qubits at a depth of 48. In other words, ordinary computers are back on top.
Their new algorithm only gives approximate answers, resulting in errors. That is OK, because current quantum computers also generate small errors, meaning you have to run calculations multiple times to get the correct answer. By matching the size of these errors, Boixo鈥檚 team can help ordinary computers catch up to their quantum competitors.
Boixo and his colleagues found they can simulate the output of a circuit with 49 qubits at depth 48 with a fidelity of 0.5 per cent聽鈥� a measure of the error in the calculation. They estimate this would cost $1 million on Google鈥檚 cloud computing service, which is expensive, but not scientifically impossible.
It is a reasonable approach, says Ciar谩n Lee at University College London. 鈥淚t seems only fair to allow classical computers to simulate the outcome approximately.鈥�
Including cost as a factor in determining whether a system can achieve quantum supremacy brings this race more into the real world, says Lee. 鈥淭his reduces the problem of proving clear聽quantum supremacy from showing quantum computers can do things that are classically聽computationally infeasible, to showing that they can do things that are currently economically infeasible,鈥� he says.
Reference:听
