⚛️ quantum physics

Microscopic Side Information Controls Ordered Hayden--Preskill Recovery

This paper demonstrates that the loss of microscopic qubit identity in the Hayden–Preskill protocol, leaving only relative order or partial block information, fundamentally alters the recovery threshold by shifting the required output size from linear to a sublinear scale of $\Theta(n^{2/3})$ (or $n^{2/3}B^{-1/3}$ with partial block information) due to rank-aligned coincidences between random subsequences.

Original authors: Joao V. R. Alencar, Allan R. P. Moreira, Joao B. R. Silva

Published 2026-07-20

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Joao V. R. Alencar, Allan R. P. Moreira, Joao B. R. Silva

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to send a secret message across a chaotic, swirling storm. In the world of quantum physics, this storm is called "scrambling." When information falls into a chaotic system—like a black hole—it doesn't just disappear; it gets mixed up so thoroughly that it spreads out across the entire system, hidden inside complex connections between particles. This is the heart of the Hayden–Preskill protocol, a famous thought experiment suggesting that if you wait long enough, you can reconstruct a lost diary from the radiation a black hole emits, provided you have enough of that radiation and know exactly which pieces you are holding.

But here is the catch: to reconstruct the message, you usually need to know which specific particles you caught. It's like trying to solve a puzzle where someone hands you a handful of pieces but forgets to tell you if they are from the sky, the ocean, or the edge of the picture. If you don't know the identity of the pieces, the puzzle becomes infinitely harder. This paper explores a specific, tricky version of that puzzle: what happens if you catch the pieces in the correct order (like a sentence where the words are in the right sequence) but you have absolutely no idea where in the original story those words came from? The researchers ask: how many pieces do you need to catch to successfully read the message again?

The authors, Jo˜ao V. R. Alencar, Allan R. P. Moreira, and Jo˜ao B. R. Silva, investigate this "ordered deletion" scenario. They prove that if you lose the microscopic labels (the "ID tags") of the particles but keep their relative order, the amount of information you need to recover the message changes dramatically. Instead of needing a tiny, constant number of particles, you now need a number that grows with the size of the system, specifically following a rule where the required amount is proportional to the system size raised to the power of 2/3.

Think of it like this: imagine a massive library with $n$ books, all shuffled into a chaotic pile. You want to find a specific short story (a "diary") hidden inside. In the standard version of the game, if you are told exactly which books to grab, you only need a few. But in this new game, the librarian hands you a stack of books in the order they appeared on the shelf, but she doesn't tell you which books they are. You just know, "Book 1 came before Book 2, which came before Book 3."

The paper proves that if you grab too few books (specifically, if the number of books is much smaller than $n^{2/3}$ ), you are stuck. No matter how smart your decoding algorithm is, you cannot recover the story better than random guessing. The information is effectively lost in the noise of the missing labels. However, once you cross a certain threshold—collecting roughly $n^{2/3}$ books—the odds shift. Suddenly, the relative order of the books provides enough "clues" to reconstruct the original story with near-perfect accuracy.

The researchers also explore a middle ground. What if the librarian gives you a little more help? Instead of knowing the exact book, she tells you which section of the library the book came from (e.g., "This one is from the History section"). They found that this partial information acts like a dial. If the sections are large, you still need a lot of books, but if the sections are small (narrowing down the location), you need fewer. The math shows a smooth transition: the more precise your location hints are, the fewer total books you need to catch to win the game.

The core of their discovery lies in a statistical phenomenon they call "rank-aligned coincidences." Imagine two people independently picking random sequences of books from the library. If they pick the same number of books, there's a chance they accidentally picked the exact same physical books just by luck. The paper shows that the probability of these accidental matches depends on that $n^{2/3}$ scale. When you have fewer books than this, the accidental matches are too rare to help you. When you have more, the matches become frequent enough to act as a signal, allowing you to distinguish the real story from random noise.

In short, this paper proves that even a tiny bit of classical information—knowing the order of the pieces, or knowing which "block" they came from—can completely change the rules of quantum recovery. It turns a problem that requires a fixed amount of data into one that scales with the size of the universe you are trying to decode. The authors provide rigorous mathematical proofs for these limits, showing that the "2/3" exponent isn't just a guess, but a fundamental property of how order and randomness interact in quantum systems. They didn't just simulate this; they proved it mathematically, establishing a new "law" for how much information is needed to recover a secret when the labels are missing but the sequence remains.

Technical Summary: Microscopic Side Information Controls Ordered Hayden–Preskill Recovery

Problem Statement
The Hayden–Preskill protocol posits that a "diary" (a $k$ -qubit quantum system) deposited into an old black hole can be reconstructed from a relatively small amount of subsequently emitted radiation, provided the black hole's internal dynamics are sufficiently mixing (scrambling). Standard analyses of this protocol assume the decoder possesses microscopic side information: specifically, the decoder knows the exact identity (indices) of the collected output qubits.

This paper investigates the scenario where such microscopic labels are unavailable. Instead, the decoder receives a subsequence of $\ell$ qubits where only the relative order is preserved, but the absolute positions within the original $n$ -qubit scrambled register are erased. This models an "ordered deletion channel." The central question is how the absence of positional labels affects the amount of radiation ( $\ell$ ) required to achieve high-fidelity recovery of the diary.

Methodology
The authors model the system using an $n$ -qubit register $C = MH$, where $M$ is the $k$ -qubit diary maximally entangled with a reference $R$ , and $H$ represents the remaining black hole degrees of freedom. A scrambling unitary $U$ acts on $C$ , followed by an emission process.

Channel Definition: The primary channel is the ordered deletion channel $\mathcal{D}_{n \to \ell}$ . It selects a random subset of $\ell$ positions, retains the qubits in their original relative order, and discards the indices of the selected positions.
Partial Information Extension: The authors generalize this to block-resolved channels, where the register is divided into $B$ consecutive blocks. The decoder learns which block each surviving qubit originated from, but not its specific position within that block.
Analytical Tools:
- Converse (Lower Bound): The authors derive an exact identity for the normalized Choi purity of the ordered deletion channel. This purity is shown to be the exponential moment of the number of rank-aligned coincidences ( $M_{n,\ell}$ ) between two independent random subsequences. By bounding the factorial moments of these coincidences using negative-binomial and Dirichlet-multinomial estimates, they establish that below a certain threshold, the Choi state becomes indistinguishable from the maximally mixed state.
- Achievability (Upper Bound): For the achievability proof, the problem is mapped to a classical "planted subsequence" problem. The authors define a weighted alignment statistic $T(X, Y)$ based on the order statistics of the survivors. Using martingale concentration inequalities (Freedman's inequality) and the Hanson–Wright inequality, they demonstrate that this statistic can distinguish the planted distribution (where the subsequence is correlated with the input) from a product distribution (where it is not). This classical separation is then lifted to the quantum regime using Sibson information and the joint state–channel decoupling theorem (Cheng, Dupuis, and Gao), utilizing sandwiched Rényi divergences.

Key Results

Ordered Deletion Threshold:
- Subcritical Regime: If $\ell = o(n^{2/3})$ , the optimal entanglement fidelity $F_e^{\text{opt}}$ converges to $4^{-k}$ uniformly over all scrambling unitaries. This is the baseline fidelity achievable by discarding the output and preparing a maximally mixed state. No decoder can asymptotically exceed this value in this regime.
- Supercritical Regime: If $\ell = \omega(n^{2/3})$ and $\ell = o(n)$ , and the scrambling unitary is Haar-random, $F_e^{\text{opt}}$ converges to 1 with probability approaching 1.
- Recovery Scale: The operational recovery length scales as $\ell_{\text{rec}} = \Theta(n^{2/3})$ . This contrasts sharply with the standard labeled setting, where $\ell_{\text{rec}} = O(k)$ (constant for fixed $k$ ).
Mechanism of the $2/3$ Exponent:
The exponent arises from the statistics of rank-aligned coincidences. For two independent random subsequences of length $\ell$ , the probability that their $a$ -th survivors occupy the same microscopic site scales as $\sqrt{\ell}/n$ . Summing over the bulk of ranks yields a total alignment strength $V_{n,\ell} \asymp \ell^{3/2}/n$ . The transition occurs when this strength becomes order unity, i.e., $\ell \asymp n^{2/3}$ .
Block-Resolved Side Information:
When partial position information is available via $B$ blocks:
- If blocks are small ( $B \ge \sqrt{n}$ ), the recovery scale is $\ell_{\text{rec}} \asymp n/B$ .
- If blocks are large ( $B \le \sqrt{n}$ ), the recovery scale is $\ell_{\text{rec}} \asymp n^{2/3} B^{-1/3}$ .
- The crossover at $B \asymp \sqrt{n}$ marks the transition from relying on the accumulation of weak rank information within blocks to the blocks effectively localizing individual survivors.
Comparison with Chronological Emission:
The paper contrasts ordered deletion with a "chronological emission" channel (where the input is permutation-twirled before partial tracing). The latter retains information based on Pauli degree rather than rank-aligned coincidences, resulting in a different spectral structure and recovery properties, highlighting that the definition of "order" fundamentally alters the quantum channel.

Significance and Claims
The paper claims to demonstrate that microscopic side information is a critical resource in the Hayden–Preskill protocol. The identity of the collected subsystems is not merely a technical detail but fundamentally changes the scaling of the required output.

Theoretical Impact: It establishes that without microscopic labels, the recovery problem shifts from a constant-size requirement to a mesoscopic scale ( $n^{2/3}$ ), diverging with the system size.
Mechanism Identification: The work identifies "rank-aligned coincidences" between random subsequences as the governing mechanism for both the converse and achievability arguments in the unlabeled setting.
Information-Theoretic Insight: The results quantify the operational value of classical side information (block labels) in quantum recovery, showing how coarse-grained metadata can bridge the gap between the unlabeled and fully labeled regimes.

The authors emphasize that their achievability results are existential (proving the existence of a recovery map) and rely on the decoupling framework for random dynamics, rather than providing an explicit, efficient decoding algorithm. The work connects quantum scrambling recovery to classical deletion channels and synchronization-error models, specifically linking the analysis to planted subsequences in random deletion channels.

More like this