Exact Haar Statistics of Planar $k$-Purity in Multipartite Quantum Systems

AI summary

Overview

• Research area: Quantum information theory — multipartite entanglement, random (Haar) state statistics, and the geometry of subsystem reductions.
• Technical level: Advanced. The paper relies on replica/swap representations of purity, the Haar projector onto symmetric subspaces, permutation-group cycle counts, and finite geometric sums over interval displacements.
• Scope (one sentence): The paper computes the exact Haar-random mean, two-interval covariance kernel, variance, and (numerically evaluated) skewness of the "planar k-purity" — the average purity of all k-site contiguous intervals on a ring of n parties with local dimension p — and compares it with the analogous absolute balanced purity.

What This Paper Is About

Planar k-uniform states are multipartite pure states whose reduced density matrix on every k-site contiguous block of a ring is exactly maximally mixed. The paper proposes "planar k-purity," the mean purity of all those intervals, as a single scalar cost function whose minimum (p⁻ᵏ) is attained exactly by planar k-uniform states. The goal is to characterize how this cost function fluctuates for a typical (Haar-random) state: what its average is, how the purities of overlapping intervals are correlated, and how large the variance and skewness are as functions of n, k, and p.

Key Contributions

1. A faithful scalar cost function with a geometric meaning. The paper defines π_P^(k)(ψ) = (1/n) Σ_s π_{A_s^(k)}(ψ) over all cyclic k-site intervals, proves the bound π_P^(k) ≥ p⁻ᵏ, and notes that equality holds exactly on the planar k-uniform set — the minimum is attainable for all admissible n, k, p by Wang's construction (with the k = 1 case from generalized GHZ states).
2. Exact Haar mean and full two-interval covariance kernel. The mean is derived as (p^k + p^{n−k})/(N+1) with N = pⁿ, independent of which subsystem is chosen; the joint purity of two intervals is shown to depend only on their overlap q, via an exhaustive four-replica enumeration of the 24 permutations of S₄ into eight contraction classes.
3. Closed variance for every 1 ≤ k ≤ ⌊n/2⌋ and every p ≥ 2, plus compact parity-dependent balanced formulas (k = ⌊n/2⌋) for qubits and for arbitrary local dimension.
4. Exact higher-moment machinery and validation. A site-factorized permutation representation gives every raw moment as a finite sum; the six-replica (720-permutation) sum is evaluated in exact integer arithmetic to obtain skewness, and fixed-seed Monte Carlo simulations test the general-k, qutrit, and balanced-qubit predictions.

Main Findings

• The mean is universal across subsystem choices. E[π_P^(k)] = (p^k + p^{n−k})/(N+1), and at balance E[π_P^(r)] = E[π_A] = (p^r + p^{n−r})/(N+1) with r = ⌊n/2⌋ — planar and absolute balanced purity have the same Haar mean but different fluctuations.
• Correlations are purely geometric. For displacement d = t − s (mod n), the overlap is q_d^(k) = max(0, k−d) + max(0, k−n+d); the covariance kernel C_{n,k}^{(p)}(d) = T_{n,k}^{(p)}(q_d^(k))/(N)₄ − ((a+b)/(N+1))², with a = p^k, b = p^{n−k}, and the trace polynomial T built from x = p^q, y = p^{k−q}, w = p^{n−2k+q}. The variance is the average of this kernel over displacements: Var = (1/n) Σ_d C(d).
• Balanced qubit variances (exact). For even n, Var = [ (20/(3n))(N²−1) − 8N ] / [(N+1)²(N+2)(N+3)]; for odd n, Var = [ (6/n)(N²−1) − 9N ] / [(N+1)²(N+2)(N+3)].
• Balanced asymptotic decay. For k = ⌊n/2⌋ and N = 2ⁿ the variance is asymptotic to 20/(3nN²) for even n and 6/(nN²) for odd n. Since n = log₂N, the decay is Θ(N⁻²/log N) rather than a constant multiple of N⁻².
• General-p balanced formulas. With D_N = (N+1)²(N+2)(N+3): even n gives [4(p²+1)/(n(p²−1))·(N²−1) − 8N]/D_N; odd n gives [2(p+1)/(n(p−1))·(N²−1) − (2(p+1)²/p)N]/D_N. The qubit results are recovered at p = 2.
• Three asymptotic regimes away from balance. Fixed k: Var ~ 2c_{k,p}/(n p^{2k} N²) with c_{k,p} = p^{2k} + 2(p^{2k}−p²)/(p²−1) − 2k + 1. Extensive scaling k/n → α with 0 < α < 1/2: Var ~ 2(p²+1)/((p²−1) n N²). At balance the coefficients are 4(p²+1)/(p²−1) (even n) and 2(p+1)/(p−1) (odd n) instead.
• Planar purity fluctuates more than absolute balanced purity. The absolute variance (previously known, Eq. (34) with f₂(N) from Eq. (35)) decays approximately as N^(−2.415), faster than the planar variance; the ratio Var(π_P^(r))/Var(π_A) grows with system size, reaching approximately 2.83 at n = 10. The paper notes that averaging the same kernel with binomial (\binom{r}{d}²) multiplicities exactly recovers the known absolute result, serving as an independent analytic check.
• Typicality versus extremality. With Δ_{k,p} = E[π_P^(k)] − p^(−k), the ratio Δ/√Var ~ (p^{2k}−1)√(n/(2c_{k,p})) at fixed k, and grows as p^k√n for extensive k — so exact planar uniformity is already atypical at fixed k and exponentially more so at linear k.
• Monte Carlo agreement. For 4 ≤ n ≤ 9 with M = 40,000 states and n = 10 with M = 120,000 states, the paper reports mean, variance, and skewness residuals within 2.21, 0.98, and 1.42 standard errors respectively across 21 displayed moment comparisons. The general-parameter table (six-qubit blocks k = 1,2,3 and four-qutrit blocks k = 1,2, each on 40,000 states) has all ten mean and variance residuals within 1.74 estimated standard errors (the tabulated |z| values top out at 1.73).
• One legacy discrepancy, resolved by replication. The legacy 40,000-state n = 10 dataset gave sample variance 6.3774(455)×10⁻⁷, 3.00 standard errors above the exact value; a new independent 120,000-state run (seed 4910) gave 6.259(26)×10⁻⁷, or 0.73 standard errors above theory, matching the exact 6.24056×10⁻⁷.
• Sign of the interpretation. Because a single interval has the same Haar marginal distribution as any other subsystem of the same size, subsystem incidence is invisible to every one-cut marginal — yet it controls the collective fluctuations of geometrically related cuts.

Methodology in Plain English

The author avoids computing state integrals coordinate by coordinate. Instead, purity is written as a "swap" expectation: Tr[(|ψ⟩⟨ψ|)^⊗2 F_A], where F_A exchanges subsystem A between two copies of the state. Averaging over Haar-random states is then done with the standard formula that replaces (|ψ⟩⟨ψ|)^⊗m by a normalized sum over all m-replica permutations. For the mean, m = 2 suffices; for two purities at once, m = 4, which means enumerating 24 permutations. Because the two intervals split the ring into four disjoint regions (their overlap, the two one-sided remainders, and the rest), the trace factorizes region by region, and each permutation contributes a monomial in the region dimensions x, y, y, w. The 24 permutations collapse into eight contraction classes, giving a single trace polynomial T(q).

The remaining work is geometric bookkeeping: translating one interval around the ring changes only its overlap q with the other, and the overlap sequence takes value k once, each 1 ≤ q < k twice, and 0 exactly n−2k+1 times. Summing the resulting powers of p over displacements becomes two elementary finite geometric sums, S₊ and S₋, which produce the closed variance. Averaging the same kernel with binomial multiplicities instead of one-interval-per-displacement reproduces the known absolute-purity variance.

For higher moments, the same projector formula is applied at order m, giving an exact finite sum over m-tuples of interval start positions and over S_{2m} permutations, weighted by p raised to the total cycle count. At m = 3 this is a 720-permutation sum evaluated exactly in integer arithmetic. Numerically, states are drawn as normalized complex Gaussian vectors (exactly Haar on the unit sphere, no full unitary needed), reduced density matrices are formed from reshaped amplitude matrices C_A C_A†, and purities Tr(ρ_A²) are computed in batches. Standard errors use s/√M for the mean, a fourth-moment expression for the variance, and an empirical influence function for the skewness.

The paper is explicit about reproducibility: states at n = 4,…,8 come from legacy datasets whose pseudorandom seeds were not preserved in the notebook and therefore cannot be regenerated bit for bit, while later datasets use recorded seeds 4909 (n = 9), 4910 (n = 10 replication), 4606 and 4704 (nonbalanced qubit and qutrit), and 4104 and 4108 (paired planar/absolute comparisons at n = 4 and n = 8).

Why This Matters

Impact on research. The work supplies a statistical baseline for an entanglement constraint that is strictly weaker than absolute maximal entanglement but exists for every admissible (n, k, p) — unlike AME states, which are frustrated (the paper cites that no seven-qubit AME state exists). It shows that the number of constraints matters: planar uniformity imposes only n local constraints instead of all \binom{n}{k} size-k constraints, and at balance the number of interval constraints is linear in n. The result gives a geometry-aware benchmark that remains meaningful beyond AME existence regimes, and it isolates a concrete quantity — the variance ratio, approximately 2.83 at n = 10 — that distinguishes two functionals with identical means.

Real-world applications (as framed by the paper):

• Quantum error-correcting codes built from highly entangled states, and multiunitary/combinatorial-design constructions.
• Teleportation and secret-sharing protocols that rely on maximal entanglement across designated cuts.
• Perfect tensors and block-perfect tensors in tensor-network models of holographic error correction.
• Ordered quantum architectures and tensor-isometry/communication protocols where participating groups respect a cyclic ordering of sites.

Industry relevance. For hardware with nearest-neighbour or ring-like connectivity, planar uniformity matches the native geometry, so the purity cost function and its typical fluctuation scale give a practical figure of merit for state-preparation targets on such devices. The paper separates these statistical conclusions (in its Section V) from the distinct engineering questions of state preparation and rare-event sampling.

Future Directions

• Extending the incidence method to other subsystem families. The author notes that because the calculation depends on subsystem incidence, it extends naturally to other specified families of cuts beyond cyclic intervals.
• Bridging statistics and preparation. Section V is described as separating the statistical conclusions from state preparation and rare-event sampling — the question of how to actually reach or sample the low-purity tail remains open in the presented text.
• Higher-moment and distributional characterization. Only exact skewness (via the m = 3, 720-permutation sum) is reported; full finite-dimensional distributions of planar k-purity are not established here, unlike the one-cut case where exact purity moments and finite-dimensional distributions exist.
• Regenerating the legacy datasets. The n = 4,…,8 archived arrays can be reanalyzed exactly but not regenerated bit for bit because their seeds were not preserved, leaving fully reproducible replications at those sizes as an explicit open item.

Target Audience

Quantum information researchers working on multipartite entanglement measures, random-state typicality, and AME/planar-uniform state constructions; mathematical physicists comfortable with replica methods, permutation-group traces, and Haar integration; and architects of ordered or ring-connected quantum hardware who need quantitative benchmarks for simultaneous entanglement constraints. The paper is written at an advanced level and assumes familiarity with reduced-state purity, the swap trick, and Rényi-2 entropy.

Authors’ abstract

Planar $k$-uniform states have maximally mixed reductions on every $k$-site interval of a ring. We introduce planar $k$-purity---the mean purity of those intervals---as a faithful cost function for this simultaneous constraint. For a Haar-random pure state of $n$ parties with local dimension $p$, the joint purity of two intervals depends only on their overlap. An exhaustive four-replica calculation therefore yields the complete cyclic covariance kernel and a closed variance for every $1\le k\le\lfloor n/2\rfloor$. In the balanced qubit case, $k=\lfloor n/2\rfloor$ and $N=2^n$, the variance is asymptotic to $20/(3nN^2)$ for even $n$ and $6/(nN^2)$ for odd $n$. A site-factorized permutation representation gives all raw moments and permits exact finite-sum evaluation of the skewness. Fixed-seed simulations validate nonbalanced qubit and qutrit cases, while balanced-qubit simulations through $n=10$ validate the parity formulas and finite-size skewness. Balanced planar and absolute balanced purity have the same Haar mean but different fluctuations; the planar variance is approximately $2.83$ times larger at $n=10$. Thus subsystem incidence, although invisible to every one-cut marginal distribution, controls the collective fluctuations of geometrically related cuts. Together with universal attainability and a linear number of interval constraints at balance, this provides a geometry-aware benchmark for multipartite entanglement beyond AME existence regimes.