A quantum version of a computer algorithm widely used in finance, engineering and scientific modelling shows promising signs of operating much faster than existing methods. Experts say there are many hurdles to overcome before it overtakes versions running on ordinary computers, but that the potential gains could be dramatic.
In recent years, several research groups have claimed to achieve 鈥渜uantum advantage鈥� 鈥� the point at which a quantum computer can complete a task that would be impossible for ordinary machines. But the benchmark problem typically used in these experiments, involving simulating quantum circuits, has little real-world application.
Now, at the University of Zurich in Switzerland and researchers from IBM have demonstrated that quantum computers can speed up an extremely common algorithm that is already used in a wide range of scientific fields and industries. Monte Carlo simulations 鈥� named after the gambling destination in Monaco 鈥� are used to predict systems with a lot of randomness in their variables. Models are run many times with differing inputs and an average output is calculated.
Advertisement
The team demonstrated that, while the effort required to calculate a Monte Carlo problem increases as the size of the input increases, just like with classical computers, it does so much more slowly. This means that if larger problems can be successfully run on a quantum machine, they could be solved much faster than the classical equivalent.
Mazzola stresses that the team isn鈥檛 yet claiming quantum advantage 鈥� the result demonstrates future potential, rather than current ability. A quantum computer running the algorithm would be likely to need at least 1000 qubits, or quantum bits, to pull ahead of a classical machine. And experts expect that even upcoming more powerful quantum computers will only perform up to thousands of operations a second, whereas a classical computer can perform billions. That means a speed-up of several orders of magnitude may be necessary just to catch up with classical computers, let alone overtake them, he says.
鈥淚f this works, it鈥檚 going to enhance, by a lot, the way in which we model systems and that, in turn, will allow us to make better predictions in a wide range of fields,鈥� says Mazzola. 鈥淸But] we cannot exclude that our classical friends can devise something even better.鈥�
at quantum computing firm Orca Computing says that previous claims of quantum advantage have involved problems that are 鈥渕anifestly useless鈥�, but this Monte Carlo algorithm is different.听鈥淚t鈥檚 a clever application to a really widely used type of optimisation problem that could be relevant to lots of commercial applications,鈥� he says. 鈥淚t鈥檚 definitely promising.鈥�
at the University of Texas at Austin says the work is potentially exciting because the speed-up promises to be significant enough to overcome the downsides of quantum computers, such as the need for large amounts of qubits to be devoted to fixing errors.
鈥淚f true, this would give the speed-up a much better chance of mattering in practice after accounting for the overheads from quantum fault-tolerance,鈥� he says, though he cautions there is always a chance of someone coming up with a better classical algorithm to once again put the quantum version in second place.
Nature
