🔢 mathematics

Counterexamples to additivity of minimum output $p$ -Rényi entropy of quantum channels for $p>3/4$ and $0\leq p<1/4$

This paper establishes counterexamples to the additivity of minimum output $p$ -Rényi entropy for quantum channels in the ranges $p>3/4$ and $0\leq p<1/4$ , thereby significantly narrowing the previously open interval for additivity and improving dimension thresholds for von Neumann entropy violations.

Original authors: Debbie Leung, Benjamin Lovitz, Peixue Wu

Published 2026-07-17

📖 7 min read🧠 Deep dive

CC BY 4.0

Original authors: Debbie Leung, Benjamin Lovitz, Peixue Wu

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 stored in the delicate, ghostly states of particles. This is the world of quantum information, a realm where the rules of logic get a little wobbly and things can be in two places at once. In this library, scientists study "quantum channels," which are like special mail slots that take a piece of information, shuffle it around, and send it out. A central mystery in this field has been whether these mail slots work better when you use them one at a time or when you stack them together.

For a long time, scientists wondered if the "noise" or "messiness" (called entropy) created by two mail slots working together was simply the sum of the messiness each one made alone. It's like asking if two noisy fans together make exactly twice as much noise as one. For some types of messiness, the answer is a boring "yes." But for the most important kind of quantum messiness, the answer turned out to be a surprising "no." This discovery was huge because it meant that by using entangled inputs (where two pieces of information are linked in a spooky way), you could actually send more information through a quantum channel than anyone thought possible. However, there was a stubborn gap in our understanding: we knew this "noise reduction" trick worked for very high levels of messiness and for very low levels, but the middle ground remained a foggy mystery.

This paper steps into that foggy middle ground to clear things up. The authors, Debbie Leung, Benjamin Lovitz, and Peixue Wu, have proven that the "noisy fans" trick works for a much wider range of messiness levels than we previously knew. Specifically, they showed that for almost all levels of messiness except for a tiny, narrow strip in the middle (between 1/4 and 3/4), you can find quantum channels that break the "additivity" rule. They didn't just guess; they built a mathematical proof using random projections—think of them as magical, random filters—to demonstrate that these channels exist. Their work shrinks the mystery zone significantly, proving that the strange, non-additive behavior of quantum channels is far more common than we thought, leaving only a small, specific interval of the "messiness scale" still waiting to be solved.

The Story of the Noisy Mail Slots

To understand what these scientists did, let's imagine a game involving "quantum mail slots." In the quantum world, a "channel" is a device that takes an input (like a letter) and transforms it into an output. Sometimes, this process adds "noise," making the letter harder to read. Scientists measure this noise using something called Rényi entropy. Think of entropy as a score for how "mixed up" or "uncertain" the information is. A low score means the information is clear; a high score means it's a jumbled mess.

The big question in this field was: If you have two channels, Channel A and Channel B, is the minimum noise you get from using them together (A and B at the same time) just the sum of the minimum noise from A plus the minimum noise from B?

For a long time, scientists thought the answer was "yes" for all types of noise. But then, they discovered that for certain types of noise (specifically when the "messiness parameter" $p$ is greater than 1, or exactly 1, or very close to 0), the answer is actually "no." When you use two channels together with a special kind of entangled input, the total noise can be less than the sum of the individual noises. It's like if two noisy fans, when placed in a specific configuration, somehow cancel each other out and become quieter than either one alone. This is called a violation of additivity.

The problem was that for the middle range of messiness (between $p=0$ and $p=1$ ), we didn't know exactly where this "quieting" trick worked. We knew it worked near the edges ( $p=0$ and $p=1$ ), but the middle was a black box.

The New Discovery: Shattering the Middle

The authors of this paper decided to tackle that black box. They wanted to know: "For which specific values of $p$ between 0 and 1 can we prove that two channels together are better than the sum of their parts?"

They used a clever mathematical tool involving random projections. Imagine you have a giant, multi-dimensional block of cheese. You randomly slice it with a knife. The shape of the slice is random, but if you do this enough times, a pattern emerges. The authors used these random slices to create "projection-induced channels." These are special quantum channels built from random geometric shapes.

They then tested two different strategies to see if these channels could break the additivity rule:

The High-P Strategy ( $p > 3/4$ ): They paired a channel with its "complex conjugate" (a mathematical mirror image). They sent a maximally entangled input (a super-linked letter) through this pair. They found that for any messiness level greater than $3/4$ , the noise produced by the pair was strictly less than the sum of the individual noises.
The Low-P Strategy ( $p < 1/4$ ): They paired a channel with its "transpose-complement" (a different kind of mirror image). They found that for messiness levels less than $1/4$ , the pair produced a state that was "rank-deficient," meaning it was missing some information dimensions. This lack of dimensions forced the noise score to be lower than expected, breaking the additivity rule.

The Result: A Much Clearer Picture

The paper proves a very specific and rigorous result: For every value of $p$ greater than $3/4$ and every value of $p$ less than $1/4$ , there exist finite-dimensional quantum channels that violate additivity.

This is a massive improvement over previous knowledge. Before this, we only knew the trick worked at the very edges. Now, we know it works for almost the entire range, except for a tiny, stubborn gap between $1/4$ and $3/4$ .

The authors didn't just say "it probably works." They provided a mathematical proof that these channels must exist. They also calculated the exact dimensions needed for these channels to work. For example, for the standard quantum noise ( $p=1$ ), they found that the trick works with channels that have an output dimension of just 182. This is an improvement over a previous result that required a dimension of 183, showing that their method is slightly more efficient.

What This Means for the Mystery

The paper leaves us with a much smaller mystery. The "unresolved part" is now reduced to the interval $[1/4, 3/4]$ . This means that for 75% of the possible messiness levels in the quantum world, we now have a definitive proof that quantum channels can be "super-additive" (better together than apart).

The authors also showed that the mechanism they used doesn't have a "singular point" at $p=1$ (the most important case for real-world communication). The math flows smoothly through this point, suggesting that the weird quantum behavior isn't a fluke of a specific number but a fundamental feature of how these random channels behave.

In short, this paper is like finding a map that fills in most of the blank spots on a treasure map. We still don't know what's in the tiny island between 1/4 and 3/4, but we now know for sure that the treasure (the violation of additivity) is hidden in the vast oceans surrounding it. The authors have proven that the "quantum magic" of beating the noise limit is far more widespread than we ever imagined.

Technical Summary: Counterexamples to Additivity of Minimum Output $p$ -Rényi Entropy

Problem Statement
The additivity of minimum output entropy is a central question in quantum information theory. For a quantum channel $\Phi$ , the minimum output entropy is defined as $H_{\min}(\Phi) := \min_{\rho} H(\Phi(\rho))$ , where $H$ is the von Neumann entropy. The additivity conjecture posits that for any two channels $\Phi$ and $\Psi$ , the minimum output entropy of their product channel satisfies $H_{\min}(\Phi \otimes \Psi) = H_{\min}(\Phi) + H_{\min}(\Psi)$ .

This problem generalizes to the $p$ -Rényi entropy $S_p(\rho) = \frac{1}{1-p}\log \text{Tr}(\rho^p)$ for $0 \le p \le \infty$ . While non-additivity has been established for $p > 1$ , at the von Neumann point $p=1$ (via Hastings' random construction), and near $p=0$ (via Cubitt et al.), the intermediate regime $0 < p < 1$ has largely remained unresolved. Specifically, rigorous explicit intervals for non-additivity adjacent to the endpoints $p=0$ and $p=1$ were previously unknown or only numerically suggested.

Methodology
The authors construct counterexamples using finite-dimensional projection-induced channels. The core methodology involves:

Random Projections: Utilizing Haar-distributed projections $P_n$ of rank $d_n$ on a bipartite space $\mathbb{C}^n \otimes \mathbb{C}^k$ , where the rank ratio $d_n/(nk) \to t \in (0,1)$ as $n \to \infty$ .
Local Normalization: Converting these projections into trace-preserving quantum channels via local normalization of the Choi matrix: $J_n = (P_{A,n}^{-1/2} \otimes I_k) P_n (P_{A,n}^{-1/2} \otimes I_k)$ , where $P_{A,n} = \text{Tr}_B P_n$ .
Free Probability Tools: Employing strong asymptotic freeness and free-probabilistic techniques to determine the deterministic large- $n$ limit of the output state sets. The authors prove that the output sets converge in Hausdorff distance to a specific compact convex set $K_{k,t}$ .
Correlated Constructions:
- High- $p$ Regime ( $p > 3/4$ ): The channel $\Phi_n$ is paired with its complex-conjugate channel $\bar{\Phi}_n$ . The product channel is evaluated on a maximally entangled input (Bell state). The output converges to an isotropic state $Z_{k,t}$ .
- Low- $p$ Regime ( $p < 1/4$ ): The channel $\Phi_n$ is paired with a "transpose-complement" channel $\Psi_n$ , derived from the projection $Q_n = I - P_n^T$ . This construction leverages the exact orthogonality $P_n Q_n^T = 0$ to create a joint output with a rank deficit.

Key Results
The paper establishes the existence of finite-dimensional projection-induced channels that violate additivity for specific ranges of $p$ :

High- $p$ Range ( $p > 3/4$ ):
- The authors show that for any $p > 3/4$ , there exists a parameter $t$ (sufficiently small) and a sufficiently large output dimension $k$ such that the Bell-state test yields a violation: $S_p^{\min}(\Phi \otimes \bar{\Phi}) < S_p^{\min}(\Phi) + S_p^{\min}(\bar{\Phi})$ .
- This range includes the von Neumann point $p=1$ . The analysis shows that the von Neumann point is not a singular point for this mechanism; the same asymptotic arguments apply continuously across $p=1$ .
- Numerical Improvement: For $p=1$ , this method detects a violation at output dimension $k=182$ , improving upon the previous best bound of $k=183$ established by Belinschi, Collins, and Nechita using the random-Stinespring ensemble.
Low- $p$ Range ( $0 \le p < 1/4$ ):
- By setting $t=1/2$ and using the transpose-complement construction, the authors prove that for any $0 \le p < 1/4$ , the joint output of $\Phi \otimes \Psi$ has a rank at most $k^2-1$ , while the individual channels have full rank outputs.
- This rank deficit leads to a strict inequality $S_p^{\min}(\Phi \otimes \Psi) < S_p^{\min}(\Phi) + S_p^{\min}(\Psi)$ for sufficiently large $k$ and input dimension $n$ .
- This result replaces the previously unspecified "sufficiently small positive $p$ " neighborhood with the explicit interval $0 \le p < 1/4$ .
Unresolved Regime:
- The additivity question remains open for the interval $1/4 \le p \le 3/4$ .

Significance and Claims
The paper claims to provide the first rigorous results establishing explicit, uniform numerical endpoints for non-additivity intervals adjacent to both ends of the open interval $(0, 1)$ .

It reduces the unresolved portion of the $p$ -Rényi entropy additivity problem to the interval $[1/4, 3/4]$ .
It demonstrates that the mechanism of projection-induced channels is robust across the von Neumann point ( $p=1$ ), requiring no separate "Hastings-type" argument for $p=1$ but rather a continuous extension of the high- $p$ asymptotics.
It improves the known lower bound on the output dimension required to observe additivity violations for the von Neumann entropy from 183 to 182 within the context of random projection ensembles.

The work relies entirely on probabilistic constructions (Haar-random projections) and asymptotic analysis; it does not provide deterministic finite-dimensional realizations of the mechanism, nor does it propose experimental implementations. The significance lies in the quantitative tightening of the known bounds for non-additivity in quantum information theory.

The Story of the Noisy Mail Slots

The New Discovery: Shattering the Middle

The Result: A Much Clearer Picture

What This Means for the Mystery

Technical Summary: Counterexamples to Additivity of Minimum Output ppp-Rényi Entropy

More like this

Technical Summary: Counterexamples to Additivity of Minimum Output $p$ -Rényi Entropy