Fermionic Genuine Multiparty Entanglement

AI summary

Overview

• Research area: Quantum information theory — entanglement in fermionic many-body systems, specifically genuine multiparty entanglement (GME) and entanglement monotones.
• Technical level: Advanced (assumes familiarity with entanglement theory, fermionic parity superselection, partial transpose/partial time reversal, and semidefinite programming).
• Scope: The paper introduces and validates fermion genuine multiparty negativity (fGMN), the first efficiently computable measure of genuine multiparty entanglement for mixed fermionic states, and benchmarks it on white-noise-mixed GHZ/W states and on the 1D Kitaev chain.

What This Paper Is About

Bipartite entanglement measures for fermions are well established (notably the fermionic negativity built on the fermionic partial transpose), but there was no efficient, quantitative measure of genuine multiparty entanglement for fermionic systems — a prior study of fermion-biseparable states had suggested fermionic GME suffers sudden death, but no measure existed to test or quantify it. The authors construct such a measure by adapting the renormalized genuine multiparty negativity to fermionic systems, and then use it to map where fermionic multiparty entanglement behaves differently from its non-fermionic counterpart and from bipartite fermionic entanglement.

Key Contributions

1. A new measure: Defines fermion genuine multiparty negativity (fGMN), an efficiently computable (via semidefinite programming) measure of fermionic GME for mixed states, obtained by replacing the partial transpose in the renormalized GMN with the fermionic partial transpose and relaxing the constraints on the witness components.
2. A proof package: Establishes six properties — vanishing on all fermion-biseparable states, monotonicity under fermionic LOCC (fLOCC), convexity, reduction to the bipartite fermionic negativity in the two-party case, an upper bound by the minimum bipartite fermionic negativity (with equality for pure states and for mixed fermionic stabilizer states), and an upper bound of ½(d_min − 1).
3. A large class of fermionically-GME states: Exhibits many states with nonzero fGMN but no non-fermionic GME, including large classes of fermionic stabilizer states, demonstrating that fermionic GME is far more permissive than non-fermionic GME.
4. Phenomenology: Documents qualitative departures from bipartite fermionic negativity — finite sudden-death points in white-noise mixing, temperature, and separation — plus a lower-bounding scheme for fGMN from measured observables.

Main Findings

• Fermionic-only multiparty entanglement exists in bulk: The measure detects fGMN in states with no non-fermionic GME, including mixed fermionic stabilizer states of the form ρ = 2^(−(N−S)) ∏ₛ ½(I + gₛ), which include ground states of frustration-free fermion Hamiltonians such as Majorana dimer states, even-degree graph states, and Majorana fermion code states.
• Explicit 3-party example: For ρ = 2⁻³(I − γ₁γ₂γ₃γ₆)(I − γ₂γ₃γ₄γ₅), every partition leaves at least one stabilizer with an odd fermion number on one side, so minₘ 𝒩ᵐ_F and hence 𝒩_F are nonzero — despite the state having only two stabilizers and therefore no non-fermionic GME.
• Two parties never suddenly die, three or more can: Under white-noise mixing ρ = (1−p)|Ψ⟩⟨Ψ| + p I/d, two-party fermionic entanglement survives for all p < 1. Sudden-death mixing values reported: GHZ₂,X at 1; GHZ₃,X at 0.799962 (explicit biseparable lower bound 0.8); GHZ₄,X at 0.799949 (0.8); GHZ₄,Z at 0.727227 (lower bound 8/11); W₃ at 0.727224 (8/11).
• Temperature sudden death in the Kitaev chain: For N = 1000 sites with t = Δ = 1 and μ = 2, sudden-death inverse temperatures β are: 2 parties — non-fermionic 0.9661, fermionic 0 (none); 3 parties — 1.3425 and 0.4204; 4 parties — 1.5698 and 0.6087; 5 parties — 1.7424 and 0.7041; 6 parties — 1.8683 and 0.7701. Only the bipartite fermionic negativity lacks a sudden-death temperature.
• Topological-phase anomaly: In the Kitaev chain ground state near μ = 0, all fermionic negativities converge to ½(√2 − 1) while all non-fermionic entanglement drops to zero. This follows from the reduced density matrix of k contiguous sites being ρ = 2^(−k) ∏_{j=1}^{k−1}(I − iγ₂ⱼγ₂ⱼ₊₁), a mixed stabilizer state to which property (v.b) applies, making fGMN equal to the negativity of a single Majorana bond.
• Distance sudden death in pure states: At the critical point μ = 2, two-party fermionic entanglement remains long ranged. For three parties with single-qubit subregions, entanglement suddenly dies once qubits pass next-nearest-neighbor separation, while two-qubit subregions retain finite entanglement up to separations of 6 sites.
• Operational meaning: Fermionic GME is the resource required to decode phases s₁, s₂ = ±1 in three-party stabilizer states built from quartic Majorana generators g₁ = −γ¹_{A₁}γ²_{A₁}γ¹_{A₂}γ²_{A₃} and g₂ = −γ²_{A₁}γ¹_{A₂}γ²_{A₂}γ¹_{A₃}; no fermionic biseparable resource suffices under fLOCC.

Methodology in Plain English

The authors start from an existing multiparty entanglement measure (the renormalized genuine multiparty negativity) that works by searching over "witnesses" — operators that are guaranteed to have non-negative expectation value on any state that is merely a mixture of states separable across some cut. Finding the best such witness is a semidefinite program, which computers solve efficiently.

To make this fermionic, they swap the ordinary partial transpose inside the witness for the fermionic partial transpose (a partial time reversal that gives odd Majorana operators a phase of −i and even ones +i on the transposed subsystem). That swap alone does not reproduce the known two-party fermionic negativity, because the fermionic partial transpose can produce complex eigenvalues. Their fix is to let the witness component that does the detecting live in a disk of radius ½ in the complex plane (‖2Qₘ − I‖ ≤ 1) rather than on the real interval [0, 1], and to allow the positive "buffer" component to carry an anti-Hermitian part while keeping its Hermitian part positive semidefinite. The relaxed condition is still expressible as a semidefinite constraint, so the problem stays efficiently solvable, and the measure optimizes only the real part of tr(ρW).

They then prove the measure's properties analytically, and numerically evaluate it with the MOSEK and YALMIP optimization packages. For the Kitaev chain, they exploit the free-fermion structure: covariance matrices of ground or thermal states are computed for the large chain, restricted to the subregions of interest, and converted to the occupation-number basis via a Jordan–Wigner transformation so the SDP can be run on the reduced density matrices.

Why This Matters

• Fills a conceptual gap: Fermionic entanglement had a well-developed bipartite theory but no quantitative multiparty counterpart; fGMN makes fermionic GME measurable, comparable across states, and computable in practice.
• Corrects an intuition: The bipartite fermionic negativity famously avoids sudden death, which encouraged the view that fermionic entanglement is uniformly robust. The new measure shows that robustness is a bipartite phenomenon — multiparty fermionic entanglement does die abruptly in temperature, noise, and distance.
• Reveals a genuinely fermionic resource: Stabilizer states with fGMN but no non-fermionic GME mean fermionic multiparty entanglement is a resource invisible to standard (non-fermionic) entanglement theory, relevant to any protocol constrained by parity superselection.
• Practical detection: Because optimizing over a subset of witnesses gives a lower bound, fGMN can be bounded from measured observable expectation values without full tomography — useful in experiments.

Real-world applications (as framed by the paper's motivation for GME generally):

• Measurement-based quantum computing.
• Quantum error correction.
• Quantum metrology.
• Quantum key distribution.

Industry relevance: Directly relevant to quantum-hardware and quantum-software groups working with fermionic platforms (superconducting circuits, quantum dots, cold atoms) and to developers of quantum algorithms and error-correcting codes built on Majorana modes, where parity superselection is a physical constraint rather than a technicality. The SDP-based measure is implementable with standard convex-optimization tooling.

Future Directions

1. Gaussian-state formulation: Adapt fGMN so it operates only on the correlation matrix, as is possible for the bipartite fermionic negativity. The authors note an obstacle: the sum of two Gaussian states is not necessarily Gaussian, so demanding that W and all its components Pₘ, Qₘ be simultaneously Gaussian is ill-posed.
2. Area-law and log-law saturation: Since property (v) bounds fGMN by the minimum bipartite fermionic negativity, the area-law/log-law scaling of bipartite entanglement bounds multiparty fGMN too; the argument of Ref. 52 suggests this should be generically saturated for large party size, which remains to be tested, especially where bipartite fermionic entanglement has log-law scaling.
3. Field-theoretic interpretation: Search for a replica-style expression of fGMN in terms of moments of the reduced density matrix, analogous to the field-theory understanding of bipartite negativity — the authors note no such answer exists yet for any GME measure.
4. Broader state surveys: Compare fermionic and non-fermionic entanglement across many more states and models, following the observation that fermionic multiparty entanglement was generally more abundant and more noise-robust in the systems studied, with the Kitaev-chain topological phase as a notable exception.

Target Audience

Quantum information theorists and condensed matter physicists working on entanglement in fermionic systems; researchers using entanglement negativity and its multiparty generalizations; practitioners of semidefinite programming in quantum physics; and graduate students or advanced readers seeking a rigorous, computable handle on multiparty entanglement under parity superselection. Readers without background in fermionic entanglement, the fermionic partial transpose, or convex-optimization-based entanglement witnesses will find the technical core demanding.

Authors’ abstract

Entanglement can show fundamentally different behavior in fermionic systems. However, while bipartite measures of fermionic entanglement have been established, genuine multiparty entanglement (GME) in fermionic systems is much less understood. We introduce an efficiently computable measure (via semi-definite programming) of fermionic GME, fermion genuine multiparty negativity (fGMN), and show that it is an entanglement monotone and a natural multipartite extension of the fermionic negativity. Using this measure, we find many states which have fGMN but no non-fermionic GME, including large classes of fermionic stabilizer states. The fGMN also displays some major phenomenological differences to the bipartite fermionic negativity, such as finite sudden death points over separation and temperature.