Tuesday, July 18, 2017

Lukin's 51-qubit Russian-American analog quantum computer

My former ex-colleague Mikhail Lukin has been doing some cool things in atomic physics and the most recent one is a 51-qubit quantum computer of a sort. Read the story
Russian, US Scientists Team Up to Create World's Most Advanced Quantum Computer
at the semi-official Russian Sputnik News server (or an original story in Russian or Google News).

You may see the different attitudes of the media in the two countries. While the U.S. media are obsessed with witch hunts against a son of Donald Trump who may have touched the hand of a Russian woman in the recent year, and they wouldn't dare to mention a Russian quantum computation advance at Harvard, Russian media are celebrating the cooperation of Russian and American scientists.

This Snap's hit from the 1990s, "Power", is started by a Russian announcement that the U.S. company Transceptor Technologies began to produce the computers Personal Sputnik. ;-)

Think twice who is the aßhole here. Yes, Lukin is Putin's agent according to a Democratic Party's lawmaker.

Their results were summarized in a fresh quant-ph preprint
Probing many-body dynamics on a 51-atom quantum simulator
The authors are mostly affiliated with Harvard's physics department, 1.5 of them with MIT, 1 with Caltech, and 0.5 of them with Harvard-Smithsonian. So despite Misha's ethnicity, that would probably make the paper an American one. But Lukin recently co-founded the Russian Quantum Center. I suppose that the plan is to do lots of things over there.

While IBM and Google compete for the largest quantum computer with quantum supremacy – but their encoding of the quantum information is not quite clear – Lukin et al. are very transparent about the technology. They use strongly interacting atom arrays. In particular, it is rubidium-87 atoms. All of the atoms may be either in the ground state \(\ket g\) or the excited state \(\ket r\), the Rydberg state. Those excited states are meant to have hydrogen-like energies multiplied by \(Z^2\).

The Hamiltonian of their array is\[

\frac{{\mathcal H}}{\hbar} = \sum_i \frac{\Omega_i}2 \sigma^i_x - \sum_i \Delta_i n_i +\sum_{i\lt j} V_{ij}n_i n_j

\] where \(\sigma_x\) is the usual Pauli \(2\times 2\) matrix in the basis \(\ket g,\ket r\), treated just like if it were "up" and "down", and \(n_i\) is \((1-\sigma_z)/2\) in the same basis – it is equal to \(\ket r\bra r\). The interaction constants \(V_{ij}\) seem constant in their gadget but the Rabi frequency \(\Omega_i\) may be changed by adjusting their laser intensities and detunings, much like the parameters \(\Delta_i\).

So they may and they do manipulate the lasers to achieve some instructions – and something is happening with the quantum information stored in the question "whether this atom or the next one is excited". I did expect Lukin to work with excited and unexcited atoms and lasers – it's been his strength for quite some time.

The parameters are adjusted "continuously" and for this reason, their computer is a quantum computation counterpart of "analog computers" that existed around a century ago. They have already computed some difficult tasks using their quantum computer. But it's not some Shor-like algorithms that would use the quantum supremacy. They have simulated some quantum evolution or computed optimization etc.

I think that computers like that could be turned into full-blown quantum computers that run the usual "purely digital" quantum algorithms. What is analog about Lukin's computer, after all? Perhaps the fact that the adjustable parameters are continuous? It looks very likely to me that a priori, all similar computers may adjust their parameters continuously. You just need to adjust them accurately enough to pretend that your quantum computer is "digital". And it would also be helpful to make some of the parameters \(V_{ij}\) adjustable.

There have been many other proposals – and people are working on proposals – how to store the quantum bits. But the excited/unexcited atoms seem so straightforward. I would be surprised if some large companies that work on this stuff didn't try to put Lukin's approach on steroids.

Lukin's device is said to beat John Martinis' 49-qubit quantum computer prepared at Google in some respects. Well, one advantage is that 51 is greater than 49 LOL. The Google's 49-qubit computer is sometimes promised by the end of the year. IBM keeps on claiming that they're still leaders, too.

No comments:

Post a Comment