I just published a new blog post!

Types and Neural Networks
( www.brunogavranovic.com/posts/2026-04-20-types-a... )

It's about what comes into view once dependent type systems become a part of neural network training.

Fun twist in Google's zero knowledge proof of quantum circuit.

I never wrote about it here, but as of some time ago I figured out a basic implementation for named axes in TensorType:

https://github.com/bgavran/TensorType

This means that now you're:
a) forced to assign some meaning to all your axes)

scirate.com/arxiv/2604.0...

I’ll admit I was always skeptical of any QML/classical-data-processing advantages in quantum, but this paper will at least give me pause!

everything I hate about science communication in one article. give people a catchy phrase and hide the real science behind it. people leave with the phrase and absolutely nothing else. purple isn't real is a stupid way to discuss what they are talking about. pedantry plus clickbait is not scicomm

The Moon: oh wow you guys decided to come back

Artemis II crew: earth’s haunted

Definition 2.1 (Felinity of a quantum state). For a n-qubit state ρ, the felinity of ρ, F n(ρ) ∈ [0, 1] is defined as...

"We [...] introduce a new measure, felinity."
arxiv.org/abs/2604.02793

Extending the Adiabatic Theorem Like its slowly perturbed counterpart, a rapidly perturbed quantum system stays closer to the ground state than to any excited state.

The quantum adiabatic theorem says that a slowly perturbed system stays in its instantaneous ground state. Now researchers have shown that this principle holds for the opposite limit: The ground state remains the most likely state even for a system subjected to an instantaneous perturbation.

Artemis II lifts off

The forward-looking Oratomic time-efficient architecture claims to "potentially enable runtimes of 10 days for ECC–256 with ≈26,000 qubits, and 97 days for RSA–2048 with ≈102,000." This is based on stitching together several gadgets and plausibly assuming they can work together on an atom platform.

We're making such good progress at reducing the number of qubits needed for factoring that we might accidentally overshoot and bring it down to zero, at which point we would have a classical factoring algorithm.
arxiv.org/abs/2603.28627

Just the coolest plot! I think it’s a testament to the growth of quantum information as a field that we see such a sharp reduction in resources for factoring.

Looks like Oratomic is coming in with a splash!

"our most time-efficient architectures can potentially enable runtimes of 10 days for ECC–256 with ≈ 26,000 qubits, and
97 days for RSA–2048 with ≈ 102,000 qubits" arxiv.org/pdf/2603.28627

Why mathematicians are boycotting their biggest conference More than 1,500 mathematicians are demanding that their field’s most prestigious meeting be moved from the U.S.

More than 1,500 mathematicians are demanding that their field’s most prestigious meeting be moved from the U.S.

hello 👋 i have developed a new tool called blüuümpíi. blüuümpíi will convert every file on your hard drive to a .wav file without asking you, maximize your system volume, and play them on every bluetooth speaker you have ever connected to. if you offer feedback on blüuümpíi i will take legal action

I know I've said this a trillion times but I have to say it periodically to keep from going insane:

The right ran a huge, well-funded, & ultimately successful campaign to completely take over political media & the infosphere more generally. This is by far the most important political development...

Charles Bennett and Gilles Brassard have been named the winners of the A.M. Turing Award, one of the highest honors in computing, for their work establishing the foundations of quantum information theory. The award comes with a $1 million prize. www.quantamagazine.org/quantum-cryp...

The APS Meeting in Denver starts in earnest tomorrow and as sort of an experiment to see if this still works, I've resolved to actually cover the conference on social media (like we did in days of yore), so bear with me for what will be a couple days of posting about physics.

The FeMo-cofactor and classical and quantum computing Recently, my coworkers and I put out a preprint “Classical solution of the FeMo-cofactor model to chemical accuracy and its implications’’ (Zhai et al. 2026). It is a bit unusual to write comm…

Garnet Chan reflects on recent progress using classical heuristics in computational quantum chemistry—and what it means for quantum computing. The lessons he draws can help to steer both classical and quantum approaches in scientifically productive directions.
quantumfrontiers.com/2026/03/12/t...

Chief Executive Officer - New York City, New York (US) job with arXiv | 37961678 arXiv seeks its first CEO to champion open, free scientific discovery and guide the platform’s next chapter as an independent nonprofit.

The arXiv is becoming a nonprofit separate from Cornell, and they're looking for a CEO. Please repost to let good candidates know!
@dabacon.tachyon.institute, @harrowing.bsky.social, @michaelnielsen.bsky.social
jobs.chronicle.com/job/37961678...

Onset of Ergodicity Across Scales on a Digital Quantum Processor Understanding how isolated quantum many-body systems thermalize remains a central question in modern physics. We study the onset of ergodicity in a two-dimensional disordered Heisenberg Floquet model ...

Realised I forgot to add a link to the paper!

scirate.com/arxiv/2603.1...

They introduce a cool metric based on collision entropies of coarse-grained spatial patches of the full system in order to study ergodicity across different length scales in a two-dimensional disordered Heisenberg Floquet model.

There's a whole load of benchmarking with tensor-networks to boot!

New fantastic work from my lovely colleagues at Phasecraft!

"Onset of Ergodicity Across Scales on a Digital Quantum Processor"

by F Alam, M Crichigno, E Crosson, @sflammia.bsky.social F M Gambetta, M Hunter Gordon, M Kreshchuk, @quantumashley.bsky.social , A Nocera & R A Santos

The distinction I would make is that writing is "thinking" in two simultaneous dimensions. One is the attempt to capture an idea in one's head. The other is that during the act of writing the thinking changes the original idea. When we remove the opportunity for that thinking the output is affected.

Student: Can you explain again what a free module is?

Me: Well, it's a

S: And don't say "it's a module that doesn't cost anything"

Me: Look I know my jokes are bad but I wouldn't go for one that obvious

S: OK. So what is it?

Me: It's a module that has time to help you right now

Child’s Play, by Sam Kriss Tech’s new generation and the end of thinking

“For a long time, the tech industry liked to think of itself as a meritocracy: it rewarded qualities like intelligence, competence, and expertise. But all that barely matters anymore.”

Sam Kriss reports on the next generation of AI technologists.

New Tool for Sculpting Single Photons Researchers can adjust the frequency and bandwidth of single photons inside an optical fiber, which will be useful for future quantum networks.

Researchers have shown how the frequencies and bandwidths of individual photons can be tuned over a wide range inside a short length of standard optical fiber. They anticipate that their technique will be useful in future quantum computing and communications networks.

Can the Most Abstract Math Make the World a Better Place? | Quanta Magazine Columnist Natalie Wolchover explores whether applied category theory can be “green” math.

John Baez wants to wrangle one of the most abstract branches of math to better model the natural world. It sounds like a pipe dream. But is it? Read the latest Qualia column:

They have stolen most of the code for their model and vibe coded the rest (happy to share links if needed). Makes me angry when things like this get promoted without verification