🔢 mathematics

Moment-based PPT criteria for random bipartite states

This paper establishes that for random bipartite mixed states on $\mathbb C^d\otimes\mathbb C^d$ , there exists a critical threshold environment dimension $s=\lambda_md^2$ for each level $m$ of moment-based PPT criteria, above which the states generically satisfy the criterion as the local dimension $d$ grows large.

Original authors: Cécilia Lancien, Kieran McShane

Published 2026-07-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Cécilia Lancien, Kieran McShane

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 holding a magical, two-sided coin that represents a quantum system. One side is the "coin" itself, and the other is a hidden "environment" that the coin is interacting with. In the world of quantum physics, these systems can be in a special, spooky state called entanglement, where the two sides are so deeply connected that what happens to one instantly affects the other, no matter how far apart they are.

The big problem? Figuring out if a coin is actually entangled is a nightmare. It's like trying to solve a Rubik's cube while wearing thick oven mitts; you need to know every single detail of the coin's internal structure (its "spectrum") to be sure, which is incredibly hard to measure in a real lab.

To make things easier, scientists have created a ladder of "checkpoints" called moment-based PPT criteria. Think of these as a series of increasingly strict security scanners.

The PPT criterion is the gold standard: it checks the entire spectrum. It's the most accurate but the hardest to use.
The moment-based criteria are the "quick scans." Instead of checking everything, they only look at a few specific numbers (called "moments") that are easy to measure.
The $p_3$ -PPT is the first quick scan (checking just the first few numbers).
The $p_5$ -PPT, $p_7$ -PPT, and so on are deeper scans that check more numbers, getting closer and closer to the gold standard.

The Great Experiment: Random Coins in a Crowd

Cécilia Lancien and Kieran McShane asked a fascinating question: If you take a random quantum coin (one that was created by a chaotic, uniform process) and put it in a room with a specific amount of "environment" noise, will these quick scans catch the entanglement?

They imagined a scenario where the size of the environment ( $s$ ) grows in relation to the size of the coin ( $d$ ). Specifically, they looked at what happens when the environment size is roughly $\lambda$ times the square of the coin's size ( $s = \lambda d^2$ ).

The Magic Thresholds

The authors discovered that there isn't just one "magic number" for when entanglement is detected. Instead, there is a different threshold for every level of the scanner.

If the environment is too small (specifically, if $\lambda$ is below a certain number), the random coin is almost certainly entangled, and the scanner will catch it. If the environment is too big (above that number), the coin usually behaves like a normal, non-entangled object, and the scanner will say "all clear."

Here are the specific "tipping points" they found, where the behavior flips from "caught" to "missed" as the coin gets huge:

For the first quick scan ( $p_3$ -PPT): The threshold is $\lambda = 1$ . If the environment is smaller than $1 \times d^2$ , the scanner catches the entanglement. If it's bigger, it usually misses it.
For the second scan ( $p_5$ -PPT): The threshold moves up to $\lambda = 4 \cos^2(\pi/4)$ , which equals 2.
For the third scan ( $p_7$ -PPT): The threshold is $\lambda = 4 \cos^2(\pi/5)$ , which is about 2.62.
As you go higher up the ladder: The threshold keeps rising, getting closer and closer to 4.

The authors proved mathematically that as the coin gets infinitely large, these thresholds are sharp. If you are just below the number, the scanner catches the entanglement with near 100% certainty. If you are just above it, the scanner misses it with near 100% certainty.

What They Ruled Out

The paper explicitly argues against the idea that these quick scans are useless or that they fail to detect entanglement in high-dimensional systems.

They are not "too weak": Even the very first, simplest scan ( $p_3$ -PPT) works surprisingly well. It detects entanglement up to an environment size of $d^2$ . This is actually better than some other famous methods (like the "realignment" criterion) which only work up to a smaller size.
They are not "perfect" yet: The quick scans are not the same as the gold standard (PPT) until you go infinitely high up the ladder. The gold standard only kicks in when the environment is larger than $4d^2$ . So, for any single, fixed scan, there is a "blind spot" between its specific threshold and 4 where it might miss entanglement that the full PPT test would catch.

How Sure Are They?

The authors didn't just guess or run a few computer simulations. They used heavy-duty mathematics involving random matrix theory, combinatorics (counting permutations), and concentration of measure (a fancy way of saying "random things tend to stick to their average when numbers get huge").

They proved that for any fixed level of the scanner, as the system size grows, the probability of the result flipping from "detected" to "undetected" becomes 100% at these exact thresholds. They didn't just suggest it; they derived the exact formula for the threshold:
$\lambda_m = 4 \cos^2 \left( \frac{\pi}{m + 2} \right)$
where $m$ is the level of the scanner.

The Takeaway

Imagine you are trying to find a needle in a haystack. The "gold standard" is to turn the haystack inside out and look at every single piece of straw. The "moment-based criteria" are like using a metal detector.

The first metal detector ( $p_3$ ) finds needles if the haystack is small (size $d^2$ ).
A slightly better detector ( $p_5$ ) finds them if the haystack is up to size $2d^2$ .
A super detector ( $p_7$ ) works up to $2.62d^2$ .
And so on, until you reach the limit of $4d^2$ , where the metal detector becomes as good as turning the haystack inside out.

The paper shows that even the simplest, easiest-to-use detectors are surprisingly powerful, catching entanglement in a huge range of random states, and that we can predict exactly where they stop working based on a simple trigonometric formula.

Technical Summary: Moment-Based PPT Criteria for Random Bipartite States

1. Problem Statement

Characterizing quantum entanglement in high-dimensional bipartite systems is a computationally hard problem (NP-hard). The most widely used necessary condition for separability is the Positive Partial Transpose (PPT) criterion. However, verifying the PPT criterion requires full state tomography to access the entire spectrum of the partially transposed state $\rho^\Gamma$ , which is experimentally demanding.

To address this, recent works have introduced moment-based relaxations of the PPT criterion. These criteria rely only on the low-order moments $p_k = \text{Tr}((\rho^\Gamma)^k)$ , which are experimentally accessible via local randomized measurements. These relaxations form a hierarchy indexed by an integer $m$ , known as the $p_{2m+1}$ -PPT criterion.

The central problem addressed in this work is to determine the typical detection performance of these moment-based criteria on random high-dimensional bipartite states. Specifically, the authors investigate whether a random mixed state on $\mathbb{C}^d \otimes \mathbb{C}^d$ , induced by an environment of dimension $s$ , typically satisfies or violates the $m$ -th level of this hierarchy as the local dimension $d \to \infty$ .

2. Methodology

The authors employ tools from random matrix theory and asymptotic geometric analysis to analyze the statistical properties of random induced states.

2.1. Model of Random States

The study considers a random pure state uniformly distributed on $\mathbb{C}^d \otimes \mathbb{C}^d \otimes \mathbb{C}^s$ . The bipartite mixed state $\rho$ is obtained by tracing out the environment $\mathbb{C}^s$ . This state is equivalent to a normalized Wishart matrix $W \sim \mathcal{W}_{d^2, s}$ , where $s = \lambda d^2$ for a fixed parameter $\lambda > 0$ .

2.2. Combinatorial and Analytical Techniques

To determine the behavior of the moments $p_k = \text{Tr}((\rho^\Gamma)^k)$ , the authors:

Compute Expectations and Variances: Using Wick's formula for complex Gaussian variables and the combinatorics of permutations and non-crossing partitions, they derive exact formulas for the expectation and variance of the moments of partially transposed Wishart matrices.
Asymptotic Analysis: They identify the dominant terms in the large $d$ limit. The leading order behavior is governed by permutations lying on the geodesic between specific cycle permutations, which correspond to non-crossing partitions with cycles of length 1 or 2.
Concentration of Measure: They apply deviation inequalities for polynomials in Gaussian variables (specifically Theorem 5.1) to prove that the random moments concentrate sharply around their asymptotic mean values as $d \to \infty$ .
Hankel Determinant Evaluation: The $p_{2m+1}$ -PPT criterion is satisfied if and only if a specific Hankel matrix $B_m$ constructed from the moments is positive semi-definite (PSD). By Sylvester's criterion, this is equivalent to the non-negativity of its leading principal minors (determinants). The authors use orthogonal polynomials (specifically Chebyshev polynomials of the second kind) to derive a closed-form expression for the asymptotic Hankel determinants associated with the average moments.

3. Key Contributions and Results

3.1. Sharp Thresholds for the Hierarchy

The primary result is the identification of a sharp threshold for the environment scaling parameter $\lambda$ for each level $m$ of the hierarchy.

Let $\rho$ be a random state on $\mathbb{C}^d \otimes \mathbb{C}^d$ induced by an environment of dimension $s = \lambda d^2$ . For a fixed level $m \in \mathbb{N}$ , define the threshold value:
$\lambda_m = 4 \cos^2 \left( \frac{\pi}{m + 2} \right)$

The paper establishes that as $d \to \infty$ :

If $\lambda < \lambda_m$ , the random state violates the $p_{2m+1}$ -PPT criterion with probability approaching 1.
If $\lambda > \lambda_m$ , the random state satisfies the $p_{2m+1}$ -PPT criterion with probability approaching 1.

3.2. Specific Cases and Limits

First Non-Trivial Level ( $m=1$ ): The threshold is $\lambda_1 = 4 \cos^2(\pi/3) = 1$ . This corresponds to the $p_3$ -PPT criterion (involving moments up to order 3).
Limit as $m \to \infty$ : The sequence of thresholds converges to $\lim_{m \to \infty} \lambda_m = 4$ . This recovers the known threshold for the full PPT criterion (established in [Aub12]), confirming that the hierarchy of moment-based criteria converges to the full PPT criterion.
Unbalanced Regime: The authors extend these results to the case where the local dimensions are $d_1$ and $d_2$ with $d_2 \gg d_1$ . They show that the threshold $\lambda_m(d_1)$ approaches $4 \cos^2(\pi/(m+2))$ as $d_1 \to \infty$ , with a correction term of order $O(1/d_1^2)$ .

3.3. Technical Derivations

Moment Asymptotics: The authors prove that the asymptotic average moment coefficients $M_k(\lambda)$ satisfy a recurrence relation that allows the Hankel determinants to be expressed in terms of Chebyshev polynomials $U_{m+1}$ .
Sign Analysis: The sign of the Hankel determinant $\det(H_m(\lambda))$ is determined by the sign of $\sin((m+2)\theta)$ where $\theta = \arccos(\sqrt{\lambda}/2)$ . The roots of this function yield the critical values $\lambda_m$ .

4. Significance and Claims

The paper claims to provide a quantitative comparison between the performance of implementable moment-based entanglement criteria and the theoretical PPT criterion on typical instances.

Performance Gap: The results demonstrate that even the first level of the hierarchy ( $p_3$ -PPT) performs significantly better than other known criteria (like the realignment criterion, which has a threshold of $\approx 0.72 d^2$ ) and is only slightly less powerful than the full PPT criterion (threshold $4d^2$ ) in terms of the scaling of the environment dimension required for detection.
Systematic Approach: This work offers a systematic exploration of the hierarchy, moving beyond specific cases to characterize the entire sequence of thresholds.
Limitations: The authors explicitly note that their results are rigorous only in the regime where $m$ is fixed and $d \to \infty$ . They caution against mathematically rigorous extrapolation to the regime where $m$ grows with $d$ (e.g., $m \sim d^2$ ), although the non-rigorous extrapolation correctly predicts the limit $\lambda = 4$ .

5. Conclusion

The study concludes that for random high-dimensional bipartite states, the moment-based PPT criteria exhibit a sharp phase transition. The threshold for detection moves from $\lambda = 1$ for the simplest non-trivial moment criterion to $\lambda = 4$ for the full PPT criterion. This suggests that low-order moment measurements are highly effective for detecting entanglement in random states, capturing a significant portion of the entangled state space that is missed by other efficient criteria, while remaining experimentally feasible.

The paper also outlines open questions, including the design of moment-based relaxations for other criteria (like realignment) and the geometric characterization (volume/size) of the sets of states satisfying these criteria, noting that the lack of a simple geometric description for $p_{2m+1}$ -PPT states makes such analysis challenging.