IBM develops calculation method that makes quantum computers more accurate

IBM has developed a technique called zero-noise extrapolation that removes noise from quantum calculations. It may seem strange, but by performing multi-level calculations, IBM scientists can estimate what a quantum computer would calculate if there were no distorting data.

The technique improves the results of all algorithms that are run on the IBM Q systems. That’s what IBM says in a research paper it has published in the journal Nature. In this way, quantum computers become more accurate and more useful to use. It is also important that these improvements are achieved by means of software and therefore do not require any hardware improvements.

Quantity calculations

An ordinary computer works with bits, either a 0 or a 1. It is the basic unit in which a computer processes information. A quantum computer uses quantum bits (qbits). A qbit can be either a 0, a 1 or both. A quantum computer can therefore perform more complex calculations than an ordinary computer. But those calculations involve a lot of noise. This leads to errors, as a result of which the applications of quantum computers are relatively limited.

Reducing these errors is crucial to making quantum computers more useful. IBM has been able to achieve this by changing the amount of noise in calculations. By performing a calculation several times, it is possible to estimate how much noise there is. Because many quantum calculations rely on probability calculations, this not only makes the calculations less prone to error, but also increases the chance that they are more accurate.

At the same time, the IBM researchers in their research paper admit that the improvements are not infinite. They are limited by the capabilities of the processor, which means that the noise reduction can even reach a certain level. The more complex the calculations, the less likely it is that this new technique will actually have an effect.