Researchers at Google have created an algorithm that can translate complex physical problems into the language of quantum mechanics, which could make quantum computers able to tackle more tasks.
Once they become powerful enough, quantum computers might become useful for specific jobs, such as breaking encryption or modelling quantum mechanics, but it is still largely unknown how useful they will be for many other scientific problems that classical computers can鈥檛 help with.
Certain complex problems, such as how best to distribute power over an electrical grid or how a bridge responds to an earthquake, are still best understood using classical physics and computers.
Advertisement
Now, at Google and his colleagues have developed an algorithm to translate a large range of classical problems so that they can be run on quantum computers. 鈥淭here鈥檚 an important class of classical systems in which we can get an exponential speed-up in simulating the dynamics of those systems on a quantum computer,鈥� says Babbush.
Any system that is in a steady state and is then disturbed by some external force, like a Kevlar vest suddenly being hit by a bullet, can be mathematically described by a system of balls and springs, bouncing back and forth, commonly known as Hooke鈥檚 law.
Babbush and his team realised that the mathematics of these classical spring systems, however complex, could always be expressed as a version of the Schr枚dinger equation, which describes how any quantum system changes through time.
By looking at similarities between the two equations and using symmetries in the problems, the researchers then worked out an algorithm to translate how far and fast the springs move into the language of the Schr枚dinger equation and the quantum bits, or qubits, used by quantum computers.
Many physical problems can be described using this description of balls and springs, says Babbush, including most wave-like systems such as maps of neuronal activity or light bouncing off a surface.
The researchers also showed that the problems their algorithm can solve all belong to a group 鈥� closely related to the 鈥淧鈥� of the famous P versus NP problem聽鈥� that consists of problems that quantum computers can solve in relatively short amounts of time.
This group contains every possible problem that can currently be solved on a quantum computer, which means that any other algorithm can also be expressed in the language of the ball-and-spring algorithm, although not necessarily made faster.
This is important, says at the University of Oxford, because it means this is firmly outside the purview of classical computers. 鈥淏ecause it鈥檚 solving some problem about the physics of a classical system, you might at first think, well, maybe there is just some efficient classical algorithm. And they gave very strong evidence that there isn鈥檛.鈥�
Babbush and his team haven鈥檛 calculated how many qubits would be needed to run their algorithm, but it is likely to be beyond today鈥檚 machines. However, it could be one of the first applications that a 鈥渞elatively modest鈥� error-corrected quantum computer is used for, says Babbush.
Physical Review X
