21/06/2023

Folha: Viana talks about advances in quantum computing

Reproduction of Marcelo Viana's column in Folha de S. Paulo.

Four years ago, the computing world was shaken by an article in the journal Nature in which Google researchers announced they had achieved "quantum supremacy": using a quantum computer, they had performed in just 3 minutes and 20 seconds a calculation that, according to them, the fastest conventional supercomputer in the world would take 10,000 years to complete.

While acknowledging the achievement, some experts downplayed its magnitude and relevance. In 2021, a group in China showed that, in fact, this calculation can be performed on a conventional computer in just 5 minutes. And other researchers, notably from IBM, pointed out that the problem solved by Google was handpicked for the experiment, having no practical interest in itself.

Now IBM is also proclaiming, in Nature, another important advance in the field, which would be opening the "age of utility" in quantum computing.

The idea of quantum computing dates back to the 1980s, when it was proposed by physicist Paul Benioff and other scientists. Conventional computers store and process information in the form of bits, tiny units capable of assuming only two states: 1 ("on") and 0 ("off"). Quantum computers, on the other hand, use quantum bits, called qubits, which take advantage of the strange properties of matter described by quantum mechanics to perform calculations at breakneck speed, far beyond the reach of classical computers.

The computer used by Google in 2019 had only 53 qubits, while IBM's has no more than 127 qubits. For comparison, the chip in any smartphone contains billions of conventional transistors. But what quantum computers lack in size, they make up for in speed.

However, qubits are difficult to maintain because they are easily affected by their interactions with the environment. Therefore, quantum calculations are unreliable: if we repeat the same calculation, we will likely obtain different answers each time.

This is the problem that IBM claims to be solving: its researchers have found a method to increase the reliability of calculations, in order to obtain truly useful results.