Shoham Jacoby

Ph.D. candidate at HUJI, researching quantum computing and quantum error correction.

My hebrew account at @shohamjac

[Shoham Jacoby @ShohamJacoby]

So this is how I imagine our magic state factory now (@DulwichQuantum am I doing this right?)

[Yotam Vaknin @yotamvaknin]

The commits on ecdsa-fail keep referencing "Fiat-Shamir Island." It sounds a lot like what you'd do if you wanted to cheat this test.
Are the AIs winning by simply cheating? Let's find out:

[Shoham Jacoby @ShohamJacoby]

@kasai_lab Also, I guess we are missing the other part of the calculation-the logical error rate and measurement rate during surgeries.

Because, if you plan on an algorithm that has the volume of 1e12, and get LER of 1e-12 per round, but require 100 rounds per surg, that's not good enough.

[Shoham Jacoby @ShohamJacoby]

@kasai_lab The Pareto-frontier is a great.

Normally, you would expect the distance to also come into consideration (might try metrics such as kd^2/n which does not make sense beyond 2D), but when you set a target LER and than compare encoding rates, that seems like the right way to do so.

[Shoham Jacoby @ShohamJacoby]

@yotamvaknin @letonyo @QueraComputing @TeamOratomic I asked one of them in qec26 and they denied it 🤷🏻‍♂️

[Shoham Jacoby @ShohamJacoby]

@CraigGidney Very cool find.

Regarding keeping stim working on native windows: a lot of new students I work with do not have easy access to linux and might get lost working with wsl
(after all, they are phyisicst without a
lot of sw bckgrnd yet).

So this helps them hit the ground running.

[Shoham Jacoby @ShohamJacoby]

@CraigGidney Just done my first review. I guess after hearing how difficult it is for them to find reviewers I'll do more.

[Shoham Jacoby @ShohamJacoby]

@hyharryzhou @letonyo Thank you for your comment!

So just to be clear:
When using LER per logical, you are not just dividing by k? You actually count the number of observable flips?

Meaning that if x logical flipped in the same memory experiment, this count as x errors?

[Yotam Vaknin @yotamvaknin]

New paper today!

Neutral atom quantum computers are scaling fast, but they're ~500× slower than superconducting qubits. Can we close the gap?

[Shoham Jacoby @ShohamJacoby]

@letonyo Ohh totally missed this. Thanks for the insight!

[Shoham Jacoby @ShohamJacoby]

@AndiGu00 @letonyo Very cool, thank you for the explanation!

[Shoham Jacoby @ShohamJacoby]

@AndiGu00 @letonyo Ohh totally missed this. Thanks for the insight!

[Shoham Jacoby @ShohamJacoby]

@AndiGu00 @J_Pablo_Bonilla Nice work! If that's the LER for the gross code, maybe the LER of the two-gross code is good enough for shor's algorithm.

[Shoham Jacoby @ShohamJacoby]

@letonyo @AndiGu00
Hi Andi, care to drop by? 😄

[Shoham Jacoby @ShohamJacoby]

@letonyo Ohhh, that's important to note. So by 144 relative to bravyi et al.? (12 for logicals, 12 for d rounds).

What I find weird is that the bposd plot looks the same as Bravyi et al., so I am probably still missing something.

[Shoham Jacoby @ShohamJacoby]

@TeamOratomic @preskill @RobertHuangHY @DolevBluvstein

[Shoham Jacoby @ShohamJacoby]

In their new paper, Oratomic use a bivariate bicycle code with parameters [[248,10,≤18]] (2603.28627).

I thought this is a nice test for my script, and indeed it run for about half an hour and it confirmed that the distance of the code is exactly 18.

[Shoham Jacoby @ShohamJacoby]

Wrote a (technical) blog post about finding the circuit-level distance of a quantum error correction code using integer linear programming, repo included.

Let me know if you find this helpful!

Link in a comment.

[Shoham Jacoby @ShohamJacoby]

Pushed the example to the repo, check it out!
https://t.co/CHcbXkxGNp

[Shoham Jacoby @ShohamJacoby]

Proud to have contributed to this work.

Because neutral-atom and ion platforms operate much slower than other modalities, high-rate magic-state generation is a problem.

This work presents a way for them to generate magic states at a much higher rate. Super important.

[PRX Quantum @PRX_Quantum]

By implementing transversal gates directly on surface codes in systems with nonlocal connectivity, new cultivation protocols increase magic-state-generation rates by over a factor of 20 and achieve ultralow error rates. @yotamvaknin @ShohamJacoby

https://t.co/xeNET0slqH