The publication of a new study, which took place in Nature, referred to the achievement of so-called “quantum superiority” by Google engineers after the controversy of recent weeks that resulted from the fact that part of the document had leaked and had previously been published online. The document itself was deleted.
According to the study, an experimental quantum processor developed by Google engineers would have performed a very complex calculation in minutes that a traditional supercomputer would take thousands of years to perform. These results show, as stated in the same document, that “quantum acceleration is achievable in a real system and is not excluded by any hidden physical laws.”
Google’s quantum processor, Sycamore, would actually have done a calculation in three minutes and 20 seconds. A supercomputer, even the fastest in the world, would take about 10,000 years to do the same calculation. This would indeed represent the achievement of “quantum superiority.”
In particular, as reported in the abstract of the study, Google engineers used “a processor with programmable superconducting qubits to generate quantum states on 53 qubits,” which takes about 200 seconds to scan an instance of a quantum circuit millions of times: a task that would take about 10,000 years for an avant-garde supercomputer.
IBM immediately denied the new document on the grounds that Google would not consider the currently largest and most powerful non-quantum supercomputer called Summit, which could do the same calculation in just 2.5 days.
Summit was developed by IBM and is located at the Oak Ridge National Laboratory in Tennessee. Google has not yet commented on IBM’s statements.
However, according to John Preskill, professor at Caltech who coined the term “quantum superiority,” this would be a milestone and a fundamental step toward building practical quantum computers, a new phase that would “have a significant impact on society.” According to Preskill, the actual milestone will be a useful application developed with a quantum computer. In fact, at the moment we are still confronted with results achieved in what appears to be a more competitive competition between experienced computer engineers, far from having any practical impact.
Background info and sources:
Latest posts by Martin Hill (see all)
- Artificial intelligence can predict future discoveries by analyzing millions of scientific articles - November 15, 2019
- New soft imitation leather with tactile feedback that can also be used in virtual reality - November 12, 2019
- New aluminium battery with organic molecules offers double density - November 9, 2019