You can double its use as a poster for conferences

We beat Google’s zero-knowledge proof of quantum cryptanalysis Comments

I have to say, publishing results as zero knowledge proof does have one benefit: it tricks security researchers into learning about quantum circuits. This is some amazing work!

blog.trailofbits.com/2026/04/17/we-beat-googl...

Thanks!

Congrats! Very naive question but: Why is preparing a Gibbs state an important task? I would have thought that preparing ground states may be, since there is probably some entanglement resource you could use for e.g.. However I don't immediately see the use for thermal states

It should be |ψ⟩ in the first panel

It should be |ψ⟩ in the first panel

Is the W_g related to Weingarten calculus by any chance?

Tailor Bird sews leaves together with plant fibre to form its nest!

Awesome! What algorithm (if any) did you use to find these?

Actually, I think there are infinite, but are all variations of that 7 hex case with a fully connected middle tile.

Removing that, not convinced a non-trivial solution configuration exists (eg. swapping a straight line 180 degrees). Would love to be proven wrong!

Here's a 9 hex case I found.

I can come up with some contrived examples (See other reply), but not convinced if there are other examples

But not sure how to do it in general

My friend came up with this configuration: if you rotate it, you get three triangles instead of one

Was it vibecoded?

I wonder if it's easy or hard to find configurations with multiple allowed solutions

For some reason I thought the left was a giant apple

A lot of analytic work can be in the former too, I think. Things that could be automated in practise once you've done it once that were cumbersome to to before (unless you want to rewrite Mathematica). How much of the such derivations by LLMs should we trust? Even for simulations we do many checks.

From a pen & paper theorist point of view, I find it much harder to fall into the trap described here, since I find that it *is* the point to understand every detail of the proof (as you say). In contrast, for coding, often I just want a function without understanding every detail and I start vibing

From a pen & paper theorist point of view, I find it much harder to fall into the trap described here, since I find that it *is* the point to understand every detail of the proof (as you say). In contrast, for coding, often I just want a function without understanding every detail and I start vibing

UK greenhouse gas emissions are now down 54% since 1990. They fell 2% last year.

It's something the country doesn't boast about enough. Yes, some of it is switch from industry to services. But the biggest reductions are from changes to electricity supply.

👇
www.gov.uk/government/s...

amazing! can you make the pointer also curved?

don't forget sepia mode

Quantum Suicide in Many-Worlds Implies P=NP In this paper we propose a totally serious algorithm to solve NP problems in polynomial time provided one is willing to wager the fate of all observers in the universe on the many-world interpretation...

what got me was modelling the observer's death with a replacement channel to the global thermal state ρ(β)

scirate.com/arxiv/2603.2...

woah. i thought it was when you apply a phase of -1 to the qubits |ψ⟩ to obtain -|ψ⟩

I might owe you a lot more than just a qubit if it goes like -1e3, it'll be like the big short except quantum

Thanks. I think I was tripped up by the phrases 'bulk' and 'area' entropy in the abstract. Sorry to bug you again but: if so, then what does adding 'mass' correspond to in the QEC? Is it like adding local errors, or changing the logical state? And can I understand how it would 'bend' the metric?

idea is that*

I'm guessing idea that the bulk to boundary map is some isometry encoding logical quantum information into a quantum error correcting code. How does gravity play into all this? I'm a PhD student in quantum info/many body physics but no background in quantum gravity.

what if it goes negative

If you have agents who are perfect Bayesians, surely they have a prior and posterior modelling the other agents. But then they have to model that they have a prior and posterior etc... how does one stop the infinite recursion?