⚛️ quantum physics

Constraints on recovering quantum information after erasure

This paper proves M.B. Hastings' conjecture by establishing new no-cloning constraints that demonstrate it is impossible to perfectly recover quantum information with a probability exceeding $1-t$ when each of $n$ physical carriers is independently erased with probability $t \geq 1/2$ .

Original authors: Mohammad A. Alhejji, Noah Lordi, Omkar Baraskar, Ariel Shlosberg, Emanuel Knill

Published 2026-07-21

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Mohammad A. Alhejji, Noah Lordi, Omkar Baraskar, Ariel Shlosberg, Emanuel Knill

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 the universe as a giant, cosmic library where information isn't written on paper but is stored in the delicate, invisible states of tiny particles. This is the world of quantum information, a field that promises computers so powerful they could solve problems in seconds that would take today's supercomputers millions of years. But there's a catch: quantum information is incredibly fragile. It's like trying to keep a soap bubble intact while a hurricane is blowing; the slightest touch, or "erasure," can make the information vanish forever.

To protect this fragile data, scientists use "quantum codes," which are like magical safety nets. Instead of storing a secret in one place, they spread it out across many carriers (particles), similar to how you might hide a treasure map by tearing it into pieces and scattering them across a forest. If you lose a few pieces, you can still reconstruct the map. However, there's a fundamental rule of the quantum world called "no-cloning." It means you cannot make a perfect copy of a quantum state. If you could, you could cheat the system and recover the map even if you lost too many pieces, which physics simply doesn't allow. This rule sets a hard limit on how much information we can protect and how much loss we can survive.

This brings us to a big question that has puzzled scientists: If we lose a specific set of pieces from our scattered map, what are the odds we can still perfectly rebuild the original secret? In 2026, a team of researchers led by Mohammad A. Alhejji and colleagues tackled this by looking at the strict mathematical limits of these recovery chances. They were testing a bold guess made by physicist M.B. Hastings, who wondered if there's a "tipping point" where, once you lose more than half your pieces, it becomes mathematically impossible to recover the information with high probability, no matter how clever your code is.

The team proved that Hastings was right. They discovered that for quantum information, there are strict "no-cloning" boundaries that act like invisible walls. If you try to recover information from too many different, non-overlapping sets of lost pieces at once, the math breaks down. Specifically, they showed that if each piece of your quantum data is erased with a probability of 50% or higher, you cannot recover the information with a success rate better than the chance that it wasn't erased. They didn't just guess this; they provided a rigorous mathematical proof using a concept called the "Lovász number," which acts like a complex ruler measuring the geometry of how these pieces can or cannot overlap.

For small numbers of carriers (four or fewer), the rules are relatively simple: the only limit is that you can't recover more than 100% of the information across all possible scenarios. But once you get to five or more carriers, the rules get weird and much stricter. The researchers found that the sum of your recovery probabilities for certain groups of lost pieces is capped by a specific number derived from the shape of their relationships, not just by simple addition. They even showed a specific example with five carriers where a seemingly possible recovery plan turns out to be a logical impossibility, like trying to walk through a wall that physics says doesn't exist.

Ultimately, this paper settles a long-standing debate by confirming that the quantum world has a hard ceiling on how well we can simulate a "cleaner" channel from a "noisier" one. If your quantum channel is too noisy (erasing more than half the data), you cannot magically simulate a cleaner channel that erases less. The authors proved this by showing that any attempt to do so would violate the fundamental laws of quantum mechanics. While they solved this specific puzzle, they also highlighted that for larger systems, there are still many unknowns about exactly how these recovery probabilities behave, leaving the door open for future explorers to map out the rest of this strange, quantum landscape.

Technical Summary: Constraints on Recovering Quantum Information after Erasure

Problem Statement
The paper addresses the fundamental limits of quantum error correction regarding the probabilistic recovery of quantum information after the erasure of physical carriers. While the minimum distance of a quantum code provides a worst-case guarantee (perfect recovery for any erasure of size $d-1$ or less), it offers a coarse characterization of performance. The authors investigate the fine-grained set of realizable recovery probabilities, denoted as $Q_n$ . Specifically, given a tuple of probabilities $q = (q_S)_{S}$ , where $q_S$ represents the probability of perfectly recovering quantum information from a specific subset of carriers $S$ , the paper asks: which tuples $q$ can be realized by a quantum code of length $n$ ?

This inquiry is motivated by the "erasure-simulation conjecture" proposed by M.B. Hastings. The conjecture posits that for a quantum erasure channel $E_t$ with erasure probability $t \ge 1/2$ , it is impossible to simulate an erasure channel $E_r$ with $r < t$ using asymptotically large block lengths. Proving this requires establishing rigorous constraints on the sum of recovery probabilities for specific sets of carriers.

Methodology
The authors employ a combination of convex geometry, quantum information theory, and graph theory:

Convexity and Downward Closure: The set of realizable tuples $Q_n$ is shown to be convex and downward closed (if $q \in Q_n$ and $q' \le q$ coordinate-wise, then $q' \in Q_n$ ). This allows the authors to characterize $Q_n$ by examining its extreme points.
No-Cloning Constraints: The core physical principle utilized is the no-cloning theorem. The authors demonstrate that for any disjoint carrier sets $S_1$ and $S_2$ , the sum of their recovery probabilities cannot exceed 1 ( $q_{S_1} + q_{S_2} \le 1$ ). This is derived by considering parallel recovery operations; if the sum exceeded 1, it would imply a non-zero probability of cloning the quantum state, which is forbidden.
Graph-Theoretic Mapping: The authors map the problem of summing recovery probabilities to the Lovász number ( $\vartheta$ ) of a graph. They define the "disjointness graph" $G_F$ for a family of carrier sets $F$ , where vertices are sets and edges connect disjoint sets. By constructing an orthonormal representation of the intersection graph (the complement of the disjointness graph) using the recovery operators restricted to the code subspace, they bound the sum of probabilities.
Computational Verification: For small $n$ ( $n \le 4$ ), the authors use the Sage mathematical software to compute the extreme points of the set $T_n$ (the set of tuples satisfying basic no-cloning constraints) and verify that $Q_n = T_n$ .

Key Results

Characterization for Small $n$ : For $n \le 4$ , the set of realizable tuples $Q_n$ is exactly the set of non-negative tuples satisfying the partition constraints (the sum of probabilities over any partition of carriers is $\le 1$ ).
Strict Inclusion for $n > 4$ : For $n \ge 5$ , $Q_n$ is a strict subset of $T_n$ . The authors provide a counterexample for $n=5$ : a tuple where every 2-element set has a recovery probability of $1/2$ and all other sets have 0. While this tuple satisfies the partition constraints, it is not realizable because it leads to a logical contradiction (implying the cloning of a state) when applying the recovery operators.
New Lovász Number Constraint: The paper establishes a general upper bound on the sum of recovery probabilities for any family of sets $F$ :
$\sum_{S \in F} q_S \le \vartheta(G_F)$
where $\vartheta(G_F)$ is the Lovász number of the disjointness graph of $F$ . For the specific case of $k$ -sets in $n$ elements, this bound is given by the binomial coefficient $\binom{n-1}{k-1}$ .
Proof of the Erasure-Simulation Conjecture: Using the derived Lovász number constraints, the authors prove Hastings' conjecture. They show that if a channel $E_t$ (with $t \ge 1/2$ ) could simulate $E_r$ , the asymptotic behavior of the recovery probabilities would violate the established bounds unless $r \ge t$ . Thus, for $t \ge 1/2$ , $E_t$ cannot simulate any erasure channel with a lower erasure probability.

Significance
The paper provides a rigorous proof of the erasure-simulation conjecture, resolving a significant open problem in the theory of quantum channel simulation. By moving beyond the worst-case metric of minimum distance, the authors establish a "fine-grained" characterization of quantum code performance. The introduction of the Lovász number as a bound on recovery probabilities connects quantum error correction to deep results in graph theory and combinatorial optimization. The work clarifies that for systems with five or more carriers, the simple no-cloning partition constraints are insufficient to describe the limits of quantum information recovery, necessitating more complex graph-theoretic constraints.

Limitations and Open Questions
The authors note that for $n \ge 5$ , a complete understanding of the set $Q_n$ is lacking. While they prove $Q_n \subsetneq T_n$ and provide the Lovász bound, they do not claim to have found a complete set of constraints for all $n$ . They explicitly leave open the question of whether constraints beyond no-cloning exist and whether specific inequalities (such as the pentagon bound for $n=5$ ) can be saturated by a quantum code.

More like this