Google Deep Mind in collaboration with Google Quantum put on arxiv their new decoder AlphaQubit 2.
They show high accurancy and fast ML decoding on surface code and Colo(u)r code!
Its exciting to start seeing high accuracy and fast decoder for the color code!
arxiv.org/abs/2512.07737
Overview of RL control. (a) Hierarchy of the feedback loops in control of an error-corrected quantum system. The low-level loop with analog control and readout signals (purple) occurs on a time scale of one QEC cycle; the logical algorithm’s digital feedback loop (green) occurs on a time scale of the decoding latency; the learning feedback loop (pink), presented in this work, is not synchronized with the lower levels, and occurs on a time scale determined by the relevant system drift. The indicated time scales are characteristic of superconducting circuits quantum computing platform.(b) A small space-time chunk of the QEC circuit for the repetition code, highlighting two overlapping detecting regions.(c) One iteration of the learning process. In each epoch, a batch of control policy candidates is sampled from the policy distribution. A certain number of QEC cycles is executed with each policy candidate (shades of red and blue). The acquired QEC data is used to compute rewards by estimating error detection rates for each detector. This information, indicating the relative performance of each policy candidate, is converted by the learning algorithm into a small gradient step of the policy distribution. Then, a new batch of policy candidates is sampled and the process repeats.
We taught a quantum computer to learn from its own mistakes-- without stopping.
Our RL framework repurposes QEC detection events as learning signals to stabilize the system.
Result: Improved Logical Error Rates for both Surface and Color codes on Willow!
see arxiv: arxiv.org/abs/2511.08493
Great work worth a read if you work on superconducting circuit with flux control!
This weekend I tried another home experiment for my toddler. This one was suggested to me by none other than Michel Devoret!
It’s pretty cool. But I was more excited than the kid 😅
Are you a computer scientist and don't know what an OTOC is, but want to understand the problem solved in the recent Nature paper by Google Quantum AI? We wrote a 2-page note that explains the motivation and presents a simplified version of the problem for any input size.
scirate.com/arxiv/2510.1...
The same quantum tunneling honored by this year’s #NobelPrize is driving #URochesterResearch
@machielblok.bsky.social is building quantum systems that do what classical computers can’t—inspired by the same fundamental questions
uofr.us/48o7I0K
#QuantumPhysics #QuantumComputing
Well, so much for my post about learning something new this Nobel week! The Physics prize went to J. Clarke, M. Devoret, and J. Martinis, the pioneers of my own field, superconducting circuits. An absolutely foundational work, and so well-deserved.
www.nobelprize.org/prizes/physi...
Nobel Prize week is almost here. I enjoy the predictions, but the real highlight for me is the forced exposure to science outside my field. It’s the one week I get a proper dive into research I’d otherwise miss.
With my son, we played with vegetal oil, baking powder and vinegar colored in blue. The effect was really cool!
scottaaronson.blog?p=9170
Scott Aaronson on the recent « quantum advantage » claim.
My favorite quote from his comment:
« When they simulate the noiseless quantum computation classically, the advantage disappears? WTF? »
Blog post: "Actually, you can't test if quantum uses complex numbers" algassert.com/post/2501
I doom the concept of that 2021 Nature paper by showing how to compile any distributed quantum protocol into real-only gates while preserving locality.
This work started w/ my struggle to understand fault-tolerant cluster states, and in particular why they always magically implement the same QEC problem as non-MBQC circuits. Understanding this correspondence as chain complex equivalence finally solved it for me, and I hope for other people too :)
Been awhile since I wrote a blog post.
Anyways... why haven't quantum computers factored 21 yet?
algassert.com/post/2500
The QEC25 conference hosted by @yaleqi.bsky.social was really excellent, and videos of all talks are available. So much recent progress on quantum error correction!
qec25.yalepages.org
Amazing !! Thank you!
Congratulation on the manuscript!
Color code decoding with some big claim: arxiv.org/abs/2508.15743
I wish they could try to benchmark this decoder against the experimental color code data from arxiv.org/pdf/2412.14256 zenodo.org/records/1423...
We are releasing experimental data so people can try out their decoding algorithms!
The recording of my talk on the Implementing the surface code using dynamic circuits at QEC2025 is available online!
yale.hosted.panopto.com/Panopto/Page...
Nvidia is releasing their open source tensor network decoder! As a small bonus they used experimental data from the google surface code experiemnt of 2023. I wonder how it would perform on more recent datasets developer.nvidia.com/blog/streaml...
Looks like 3-coupler / hex implementation of the surface code can now tolerate dropout! www.arxiv.org/abs/2508.08116
I will be at the #QEC25 conference at Yale this week! I will be giving a talk on our work on dynamic implementation of the surface code on Monday
I'm often asked if I'll redo the 2019 quantum factoring estimate. Denser storage by yokes, smaller magic factories by cultivation, slimmer approx arithmetic by Chevignard et al… surely the cost is lower now?
Yes, it's lower now.
security.googleblog.com/2025/05/trac...
arxiv.org/abs/2505.15917
Pretty cool work on qudit encoded in GKP.
I think this is also the first experimental demo of a Error Corrected Qudit
www.nature.com/articles/s41...
Let’s be honest. It’s hard to beat the Surface code!
Detecting region with ZX calculus!
Now, if you ask the question the other way of how many circuit constructions would do that, there might be only this one?
My guess as to why the 3-coupler wasn't discovered before is that it shifts around the flavor of the stabilizers.
If you look at the stabilizer at the end of the cycles, for a given measurement qubit in the bulk, the color changes every other round. The 'ear' of the surface code patch shifts a bit.
Tomorrow I'm giving a zoom talk on magic state cultivation at the Simons quantum Colloquium.
Apr 8, 10am PST, see simons.berkeley.edu/programs-eve...
Cultivation is the (current) culmination of a decade of work ending the reign of T gates as the dominant cost in fault tolerant quantum computation.
Largest factorization done on a quantum computer.
Joke aside. That is an extremely instructive work!
In 1h (9am PST or 5pm CET) I will give a talk at the QASAR seminar hosted by @vasmer.bsky.social and @christophe.vuillot.info . I will present our work on implementing 3coupler, walking and iSWAP surface codes!
I really enjoyed the MM session on early history of quantum information. It was really uplifting to listen to.
I really love old lab notes, notes and written mails communications from these days. Look way more fun than emails and chat messages we have these days
