Three Remarkable Quantum Events at the Simons Institute for the Theory of Computing in Berkeley

In the picture (compare it to this picture in this  post) you can see David DiVincenzo’s famous 7-steps road map (from 2000) to quantum computers, with one additional step “quantum supremacy on NISQ computers” that has proposed around 2010. Step 4 of logical memory with a substantial longer lifetime than physical qubits has not yet been achieved (Craig Gidney expects it in five years); However, Dolev Bluvstein et al.’s paper makes a bold move toward higher steps in DiVincenzo’s plan and toward some kind of quantum fault tolerance even before reaching very stable logical qubits.

I would like to report on three remarkable quantum computing events at the Berkeley Simons Institute. Two events are future events. In a few hours, Oded Regev will talk about his new (now famous, see here and here ) quantum algorithm for factoring. On Monday March 25 there will be a workshop in honor of Michael Ben-Or. Update: Here is the link to Oded Regev videptaped lecture; Here is the link (livestreaming and recording) to the lectures of the Michael Ben-Or workshop. Here is my toast to Michael.

The event that already took place is a lecture by Dolev Bluvstein. The lecture described a remarkable recent work that draws much attention in the quantum computing community and gives hope for further experimental breakthroughs. In fact, Michael Ben-Or told me about the lecture, and I also read about the paper in Scott’s blog. Scott did a good job of mentioning various concerns and limitations of the work and at the same time expressed (in his characteristic style) hope that this progress will eventually make it harder to justify quantum computer skepticism; be that as it may be, let me take off my skeptical hat for this post and tell you what I learned from Dolev. A few days after the lecture Dolev and I had a zoom discussion where Dolev clarified some matters for me and told me a lot of interesting things.

The neutral atom approach dates back to early proposals and experiments in optical lattices, and in particular the theoretical work of Rydberg blockade led by Misha Lukin, Peter Zoller, and Ignacio Cirac. It is different from ion-trapped quantum computing and from superconducting quantum computing; (Among these two it is perhaps closer to ion-trapped QC).

Dolev explained how prior to the very recent paper described in the talk there were earlier breakthrough papers. Over the years these ideas were explored, among others,  by Antoine Browaeys, Mark Saffman, and Harvard / MIT team, led by Lukin, Vladan Vuletic and Markus Greiner. Several years ago, analog processing with neutral atom arrays began to show great promise. For example, a 2017 U.S.-Russia Harvard based 51 qubit experiment by Bernien et al. (At the time, I reported about this paper in this post.) Increasing the fidelity of 2-gates from 80% or so to 97.5+%, and very recently 99.5%, and building useful programmable devices brought neutral atoms (at least on some fronts) to the big-league of quantum computing methods.

Dolev started his lecture with the following statement:

Fighting decoherence is the central challenge in large scale quantum computation. Quantum error correction is the only known realistic route to suppress gate errors to the required levels for useful algorithms  .

Here is what was achieved in the very recent paper (as far as I understand)

1. They created a large variety of quantum error-correcting codes including surface code, color code, and certain high-dimensional codes.
2. To check that the state they created is indeed what they aimed for, they measured the state against a large number of product states.
3. Using these coded qubits, they created a variety of logical states such as GHZ state (with 4 logical qubits)  and IQP state.
4. For this purpose  they developed a method  of parallel control: applying simultaneously 1-qubit logical state or 2-qubit logical state for all the corresponding physical qubits.
5. Once a logical state was created they measured it and (repeating the process) obtained a large sample of bitstrings corresponding to the measurements of physical qubits.
6. At this point the following steps used (classical) computation on the samples. Each “physical” sample represents a “logical sample”.
7.  Removing bitstrings that had “many” errors leads to higher quality (yet smaller) samples for the logical state. (for the IQP circuits they do “error detection” or postselection as the codes are distance 2, but for the other circuits they do error correction without relying on postselection).
8. (Normalized) XEB (linear cross-entropy) is used as a measure of quality. For noiseless IQP states XEB=2.

Dolev responded to a question asked in his lecture that I also repeated whether their “transversal gates” globally acting on several pairs of physical qubits would not yield correlated errors: No, Dolev explained, globally acting on many physical qubits (to apply a logical 1-gate) or on many pairs of physical qubits (which form a logical 2-gate) does not lead to correlated errors. The laser just indicates on which qubits (or pairs) we need to apply a gate but the action of the gates themselves are statistically independent.

Besides that, Dolev briefly told me about various interesting things. He explained the Yale group approach to error correction and hailed them for making many of the innovations for superconducting quantum computing that we see today, he mentioned ion-trapped quantum computers and the different approaches of  Ion-Q and Quantinuum, he mentioned some other quantum error correcting experiments, including the recent Google experiment, and various other things.

Another thing that I learned is that (as I suspected) Dolev is an Israeli name and indeed Dolev (joined by his family) moved from Israel to California at the age of four.

Here are links to some relevant papers:

a) This is the paper Dolev talked about

Bluvstein, D., Evered, S.J., Geim, A.A. et al. Logical quantum processor based on reconfigurable atom arrays. Nature 626, 58–65 (2024).

b) Here are two papers (that appeared back to back) one from the Lukin Group and one from the Saffman group on recent successful neural atoms quantum computers.

Bluvstein, D., Levine, H., Semeghini, G. et al. A quantum processor based on coherent transport of entangled atom arrays. Nature 604, 451–456 (2022).

Graham, T.M., Song, Y., Scott, J. et al. Multi-qubit entanglement and algorithms on a neutral-atom quantum computer. Nature 604, 457–462 (2022).

Dolev mentioned some earlier breakthroughs for 2-gate fidelities for neutral atom qubits and especially the work of Harry Levine.

A few additional comments, some with a bit of skeptical flavor:

1. In my view the community of quantum computing should carefully scrutinize experimental claims and especially claims of substantial progress. Of course, the primary responsibility of scrutinizing scientific claims lies at the hands of the researchers themselves, and it is also the responsibility of the researchers to document the experiment and to give the community access to the raw data.
2. Over SO Craig Gidney made the following thoughtful comment (December 8th, 2023 at 12:58 pm) regarding how close we are to logical qubits of a substantially higher quality than physical qubits:
   

   “With respect to the question of “who showed/will-show the first logical>physical advantage”… I’ve surveyed the QEC experiment papers over the past two decades. Something that caught me off guard was how every single one of them, even the old NMR ones, managed to find some way to claim logical > physical. Doing single rounds instead of multiple rounds, doing rep codes instead of quantum codes, using detection instead of correction, postselecting out effects like leakage… there’s lots of ways to scale the difficulty to your current experimental capabilities.IMO, if you focus on the benchmarks that actually matter, then 5 years ago the field was obviously physical>logical. Currently the field is arguably physical ~= logical. And within 5 years I expect the field to be obviously logical>physical across multiple architectures.”

3. My “argument against quantum computers” asserts that logical qubits and gates of a substantially higher quality than (current) physical qubits and gates are inherently infeasible.
4. We had an earlier post over here about an interesting (theoretical) paper by Gao et al. from the Lukin group. 
5. There were several interesting questions and comments during Dolev’s lecture and later during the panel discussion with a lot of prominent quantum computer researchers present in the audience. The entire event was terrific.

Update (29/3): A Quanta Magazine article by Philip Ball.  It mentions another pioneering paper by Ivan Deutsch. For a more complete picture of the history, see Ivan Deutsch’s comment that mentions the 1999 paper by Gavin K. Brennen, Carlton M. Caves, Poul S. Jessen, and Ivan H. Deutsch, and the follow up 2000 paper by D. Jaksch, J. I. Cirac, P. Zoller, S. L. Rolston, R. Côté, and M. D. Lukin.