A new article introduces a continuous-variable analogue of the Pauli propagation algorithm, called Displacement Propagation. Surprisingly, non-Gaussianity and symplectic coherence can make the system easier to simulate when noise is present.
arxiv.org/abs/2510.07264

Oh yeah. Strange. It was working (for me) earlier but not now.

Thanks for highlighting.

Let‘s try this: arxiv.org/abs/2510.01154

This is the paper: scirate.com/arxiv/2510.0...

Thanks for the fun collaboration Sasha (
@sheffield-qc.bsky.social
) and Chiddy !

FYI, our results here don’t contradict arXiv:2312.09121 which focus on loss estimation, proofs, and continuous protocols

Of these, the most intriguing/significant is the switch to discrete optimization

& maybe the path to adaptive quantum advantage is all about finding those discrete sweet spots 😉

And we have found one… we provide numerical evidence that our problem lives in the Goldilocks zone:

- trainable (no exponential concentration)

- not classically surrogatable (thanks to high entanglement + magic)

Of course, this being quantum, we face two extra demons…

- Exponential concentration (barren plateaus + shot noise)

- Classical surrogation (can classical shadows fake the landscape?)

For a real separation, we need a sweet spot that dodges both.

More concretely, we show that for a range of moderate entangling strengths the landscape is unimodal but non-separable landscapes.

Then numerics show adaptive hill-climbing converges efficiently

But non-adaptive approaches - blow up exponentially.

We translate that logic into a quantum recompilation task

The hidden string = the placement of T-gates between layers of semi-random unitaries

Goal = uncover the T gates positions

As in LeadingOnes, identifying early T-gates helps you make progress, but you can’t optimize each gate independently

This is the canonical “adaptivity pays” task

Its unimodal (no local minima) but non-separable (each bit cannot be trained independently)

- Adaptive strategies can flip one bit at a time, use the feedback, and find the string in O(n) queries.

- Non-adaptive strategies need exponentially many.

Our task is a quantum twist on the classic LeadingOnes-OneMax problem.

In this problem you're trying to learn a hidden bitstring.

Your score = how many leading bits match the target before the first mismatch.

So 1110… matches 1101 better than 1011… even if they have the same Hamming weight.

@joachimfavre.bsky.social also has a bunch of cool projects being written up - and is currently looking for a PhD - so I recommend following him :)

thanks! very helpful - will add.