Anthony Leverrier

Started to curate a table of small quantum Tanner codes aleverrier.github.io/qtanner-search…

@Pronatalism ah, du Pont-Lévêque !

My 3-year-old’s hair is made of spun gold.

if you are able to get 14 papers accepted to ICML, maybe you do not actually need to post about getting 14 papers accepted to ICML

you can outsource your thinking but you cannot outsource your understanding

@dwarkesh_sp Everyone working on postquantum crypto is deeply aware of this. This is the main argument for moving now to quantum-safe cryptosystems.

We don’t talk enough about how any state or group which is harvesting encrypted packets right now will be able to read those contents once quantum computers arrive. There’s a huge espionage and transparency overhang on any information that is currently “secret” and hasn’t been encrypted using post-quantum cryptography.

@ApriiSR @RafaRuizdeLira no need to be a professional player: the state is simply the state of the board. The missing information comes from the fact that it is not possible (for a human or a machine) to explore the full tree of possibilities.

i think there is a stronger sense in which chess is a total information game—a professional chess player could pretty easily actually write down the total game state, especially if you don't make them keep track of threefold repetition

what incomplete data all the pieces are there

@daniel_mac8 had the same one with codex yesterday

WTF???

@RoyRogers_HTMS awesome! where can i get the scores?

Even the flûte was like “bro… I didn’t know I could do this either” 😭🎵 This absolute legend took the plastic recorder we all suffered through in 6th grade and turned it into Dire Straits’ “Sultans of Swing.” Zero shame. Maximum talent. The flute is now in therapy. #SultansOfSwing

@krishnanrohit @Dimillian @thsottiaux same thing here, but now it's running smoothly with 5.5

Codex app bug on Mac. It shows the update button top left. If you click it though, app shuts down, but doesn't reboot. You have to force quit and restart. FYI @Dimillian @thsottiaux

@bryan_johnson wouldn't it be more effective to team up with many partners and try different protocols on different people?

idk if I can keep this up you guys. I just got one more to do: a five min warm eye compress before bed. I do endless things every day. From the moment I wake up to bedtime, and even when I'm sleeping, I'm always doing something and measuring it. It never ends. Help.

@nic_delfosse Congrats! Awesome work!

If you had any doubt that we were taking fault-tolerant quantum computing seriously at IonQ check out our blueprint: 110 pages of compilation, error correction, micro-architecture, and a sufficiently fast decoder, validated with thorough simulations of all the components. Incredibly proud of the work of our team: Felix Tripier Woo Chang Chung, Jacob Young, Safwan Alam, Bryce Bjork, Aharon Brodutch, Finn Lasse Buessen, Nolan J. Coble, Thomas Dellaert, Dmitri Maslov, Martin Roetteler, Edwin tham, Mark Webster, Min Ye, John Gamble, Andrii Maksymov, John Paul Marceaux

@RobinKothari 17 in a row! is that a record?

My strategy of attending the past 17 QIPs in a row has paid off! I'm happy to announce I'm now chair of the QIP steering committee. More seriously, I'd like to thank the committee for selecting me and I look forward to QIP 2027 in Singapore.

useful chart to understand the progress in quantum computing

@krishnanrohit thanks for the tip!

I started a research task on Codex on my laptop and said I'm actually going to be away for several hours so please finish it, so it set a heartbeat every hour for itself to keep the task going, so I came back to something fully done. Fantastic!

@thsottiaux ship the new versions to Europe

Hello builders. What are we getting wrong with Codex, what can we improve?

@thsottiaux will it work in Europe?

1] Open Settings in the Codex app. 2] Go to Personalization and make sure Memories is enabled. 3] Turn on Chronicle below the Memories setting.

We are releasing a *research preview* of Chronicle in Codex. It allows codex to build up memories based on your day to day work on your computer and then refer to these memories to be a lot more helpful. Available for PRO subscriptions and on Mac to start. This is early and consumes quite a bit of tokens, but it has changed how I and many folks at OpenAI use Codex.

@hyharryzhou @ShohamJacoby i agree that you need to find a way to renormalize if you want to compare fairly to codes with O(1) logical qubits.

@letonyo @ShohamJacoby If we compare against 1000 logical qubits idling in surface codes, then I think normalizing by k is still reasonable when we compare against the surface code logical error rate. We do also include the full block's error rate in the main text and will clarify better going forward!

Nice paper showing that rate-1/2 qLDPC codes following Kasai's construction are very promising! arxiv.org/abs/2604.16209 But is it a new trend to normalize the logical error rate per logical qubit?

@hyharryzhou @ShohamJacoby but if you say that what matters is that "a logical error occurred", then you shouldn't normalize by k.

@ShohamJacoby @letonyo We divide by k in our quoted number. The motivation is that in large-scale algorithms, if a logical error occurred, your computation likely failed already, and it is less important to count how many errors occur. We will clarify this in future versions, thanks for the feedback!

@sl2zx either the probability that decoding fails, or the same value divided by the number of syndrome extraction rounds.

@ShohamJacoby this is my experience as well (that many logical qubits are incorrect at the same time)

@letonyo I think this is a bit misleading. From my experience, when the decoder fails, it normally flips several logical qubits in the block. So normalizing by logical qubit kind of gives the impression that the errors are uncorrelated, while in practice they may very well be.