There are often interesting reports about quantum computers in the media. The world is on several places busy building quantum computers with very generous budgets, funded by governments and software giants such as Microsoft and Google. They are also working very intensively on this in The Netherlands, Delft – QuTech. Instead of writing blogs on it, spread over different moments, I have devoted a special page on my website to it. It is also made visible in the main menu. This page will – in addition to some information on the subject – contain also actual links to articles on this subject that are interesting to me and – important – readable for laymen.
In september 2019 the Financial Times reported: ‘Google claims to have built the first quantum computer that can carry out calculations beyond the ability of today’s most powerful supercomputers, a landmark moment that has been hotly anticipated by researchers.’
The quantum processor of Google, with 54 Qubits – of which one failed – managed to produce a random sequence of 53 bits with a certain distribution within 200 seconds. That’s something even a supercomputer can’t do, since the processes of a classical computer with bits that are either 1 or 0 are fundamentally not random. Random output is even undesirable. Each Qubit of a quantum computer, on the other hand, can be in both states ‘simultaneously’. If you can succesfully entangle those 54 Qubits together without ‘disrupting’ their entanglement, you can in principle perform 254 (~250 million) calculations in parallel.
Entangling so many Qubits is a technical achievement of the first order. Qubits are very unstable, which means that they can ‘decay’ to a ‘hard’ 1 or 0 bit after a very short time, a few milliseconds. Entanglement of unstable Qubits more or less multiplies that instability per added component. Unfortunately, the article doesn’t say what that particular distribution in which those random numbers had to be generated, but I assume that you can produce a huge amount of random number series in 200 seconds, while you have to pick out those that meet your special criterion.
The article gives no further details, which allows me to give my own thoughts a little free range here. A QRNG you can purchase on the Internet has a processor around 45 Mhz, so I think it produces random zeros and ones at that rate, 45 million per second. With 53 QRNGs connected in parallel, you have generated 9 billion random sequences of 53 bits after 200 seconds. Then you still have to be able to find the series that meets that special condition, which could possibly be a tough task even for a supercomputer. But when you can impose this special condition on those 53 Qubits in advance, then you have immediately the right outcome after just one operation.
I am very curious about more details and especially how people managed to impose the desired restrictions on the Qubits in Google’s quantum computer in advance. And also why they still needed 200 seconds.