Genuine Multipartite Negativity (GMN)

• Genuine Multipartite Negativity (GMN) is a negativity-based entanglement measure that certifies genuine multipartite entanglement by requiring nonzero negativity across every bipartition.
• GMN is constructed through multiple approaches—including the minimal-bipartition, mixed-convex-roof, and geometric-mean formulations—each highlighting different aspects of multipartite quantum correlations.
• GMN offers both exact analytical results for symmetric state families and experimentally friendly lower bounds, making it a practical tool in dynamical, many-body, and open-system settings.

Searching arXiv for the cited GMN papers to ground the article in the current literature. Genuine Multipartite Negativity (GMN) is a negativity-based measure of genuine multipartite entanglement (GME), namely entanglement that is present across every possible bipartition of a multipartite quantum system. In the arXiv literature, the term covers several closely related constructions. One widely used line defines GMN through the weakest bipartition negativity, so that a state is assigned a nonzero value only if it is entangled across all cuts (Hofmann et al., 2014, Ma et al., 2014). A more recent line places negativity inside a unified geometric-mean framework over all bipartitions, denoting the resulting quantity by $G_N$ and emphasizing its sensitivity to the global distribution of entanglement rather than only the least entangled cut (Wang et al., 14 Aug 2025). Across these formulations, GMN functions as a genuine multipartite entanglement monotone: it vanishes for biseparable states and is intended to quantify irreducibly multipartite quantum correlations.

1. Conceptual scope and basic definition

The defining operational idea behind GMN is that genuine multipartite entanglement should not be inferred from entanglement across a single cut. A genuinely multipartite entangled state must be entangled with respect to every bipartition, whereas a biseparable state fails this condition on at least one cut. In the tripartite pure-state formulation, this is expressed by

$$N_{\text{GME}}(|\psi\rangle) = \min \{ N_A(|\psi\rangle), N_B(|\psi\rangle), N_C(|\psi\rangle) \},$$

where each $N_\gamma$ is a bipartition negativity-like quantity built from operators $J^{\gamma|\bar\gamma}_{\alpha\beta}$ (Ma et al., 2014). This makes the weakest bipartition the decisive one.

A closely related characterization appears in the analytical study of the genuine multiparticle negativity, where for pure states

$N_g(|\psi\rangle) = \min_m N_m(|\psi\rangle),$

with $N_m$ the standard bipartite negativity for the partition $m|\bar m$ (Hofmann et al., 2014). In this form, GMN reduces to standard negativity in the bipartite case, while remaining tailored to multipartite GME detection.

An operational restatement appears in later many-body applications: for a chosen subregion, one computes the negativity for all possible bipartitions of the reduced density matrix, and the GMN is the minimum of these negativities; a nonzero value certifies genuine multipartite entanglement (Garcia-Gaitan et al., 2 Oct 2025). This criterion encodes the same logic as the earlier pure-state definitions, but directly at the level of mixed reduced states in open-system settings.

2. Core constructions: minimum, witness, and geometric mean

The literature contains at least three prominent negativity-based constructions that are closely related but not identical. The first is the minimum-over-bipartitions form just described. The second is a witness-based formulation connected to semidefinite programming. The third is the geometric-mean construction introduced in a unified framework for GME measures.

Before comparing them, it is useful to display their representative formulas.

Construction	Representative formula	Characteristic feature
Minimal-bipartition GMN	$$N_{\text{GME}}(|\psi\rangle)=\min_\gamma N_\gamma(|\psi\rangle)$$	Determined by the least entangled cut
Mixed-convex-roof GMN	$$N_g(\varrho)=\min_{\varrho=\sum_k p_k\varrho_k}\sum_k p_k \min_m N_m(\varrho_k)$$	Analytical characterization via mixed convex roof
Geometric-mean GMN	$$G_N(|\varphi\rangle)=\left(\prod_\gamma N_\gamma(|\varphi\rangle)\right)^{1/c(\gamma)}$$	Uses all bipartitions simultaneously

In the witness-based formulation, the original GMN is introduced as a computable measure of genuine multipartite entanglement based on the detection of non-PPT mixtures using fully decomposable witnesses (Hofmann et al., 2014). For a state $$N_{\text{GME}}(|\psi\rangle) = \min \{ N_A(|\psi\rangle), N_B(|\psi\rangle), N_C(|\psi\rangle) \},$$0,

$$N_{\text{GME}}(|\psi\rangle) = \min \{ N_A(|\psi\rangle), N_B(|\psi\rangle), N_C(|\psi\rangle) \},$$1

subject to

$$N_{\text{GME}}(|\psi\rangle) = \min \{ N_A(|\psi\rangle), N_B(|\psi\rangle), N_C(|\psi\rangle) \},$$2

A renormalized version relaxes the bounds on $$N_{\text{GME}}(|\psi\rangle) = \min \{ N_A(|\psi\rangle), N_B(|\psi\rangle), N_C(|\psi\rangle) \},$$3,

$$N_{\text{GME}}(|\psi\rangle) = \min \{ N_A(|\psi\rangle), N_B(|\psi\rangle), N_C(|\psi\rangle) \},$$4

subject to

$$N_{\text{GME}}(|\psi\rangle) = \min \{ N_A(|\psi\rangle), N_B(|\psi\rangle), N_C(|\psi\rangle) \},$$5

This renormalized form admits the mixed-convex-roof representation

$$N_{\text{GME}}(|\psi\rangle) = \min \{ N_A(|\psi\rangle), N_B(|\psi\rangle), N_C(|\psi\rangle) \},$$6

The 2025 unified framework reformulates negativity-based GME quantification in terms of the geometric mean of bipartition convex-roof extended negativities (Wang et al., 14 Aug 2025). For a pure state $$N_{\text{GME}}(|\psi\rangle) = \min \{ N_A(|\psi\rangle), N_B(|\psi\rangle), N_C(|\psi\rangle) \},$$7,

$$N_{\text{GME}}(|\psi\rangle) = \min \{ N_A(|\psi\rangle), N_B(|\psi\rangle), N_C(|\psi\rangle) \},$$8

and

$$N_{\text{GME}}(|\psi\rangle) = \min \{ N_A(|\psi\rangle), N_B(|\psi\rangle), N_C(|\psi\rangle) \},$$9

where $N_\gamma$0 is a normalization factor counting the bipartitions without double counting. For mixed states,

$N_\gamma$1

This framework is presented as replacing the minimal-distance viewpoint by a geometric mean over all bipartitions, with the stated aim of capturing the global structure of multipartite entanglement (Wang et al., 14 Aug 2025).

3. Structural properties and analytical characterization

The witness-based GMN has a precise set of monotonicity and convexity properties. It vanishes for all biseparable states, is an entanglement monotone, is convex, and is invariant under local unitaries; for bipartite states, it reduces to the standard negativity (Hofmann et al., 2014). In the unified geometric-mean framework, the negativity-based construction is zero on all bi-separable states, strictly positive for GME states, convex by virtue of the convex roof, and monotonic under LOCC (Wang et al., 14 Aug 2025).

The mixed-convex-roof representation has particular conceptual importance because it connects GMN to convex-roof entanglement theory while remaining analytically tractable for certain highly symmetric families (Hofmann et al., 2014). For pure states, the formula collapses to the minimal bipartite negativity. For mixed states, every valid witness provides a lower bound and every decomposition provides an upper bound, making exact evaluation feasible for select state classes.

Two exact analytical results are especially prominent. For $N_\gamma$2-qubit GHZ-diagonal states,

$N_\gamma$3

where $N_\gamma$4 is the fidelity with the $N_\gamma$5th GHZ-basis state (Hofmann et al., 2014). For 4-qubit cluster-diagonal states,

$N_\gamma$6

with the minimum taken over the explicit witness families constructed in the paper (Hofmann et al., 2014). These results show that GMN is not merely numerically computable but can be resolved exactly for nontrivial state families.

Analytical control also extends to superpositions. For a normalized tripartite superposition

$N_\gamma$7

tight upper and lower bounds for $N_\gamma$8 are given in terms of quantities $N_\gamma$9 and $J^{\gamma|\bar\gamma}_{\alpha\beta}$0 built from bipartition operators $J^{\gamma|\bar\gamma}_{\alpha\beta}$1 (Ma et al., 2014): $J^{\gamma|\bar\gamma}_{\alpha\beta}$2

$J^{\gamma|\bar\gamma}_{\alpha\beta}$3

For the GHZ–W superposition,

$J^{\gamma|\bar\gamma}_{\alpha\beta}$4

the upper bound matches the actual value, and the genuine negativity is

$J^{\gamma|\bar\gamma}_{\alpha\beta}$5

(Ma et al., 2014). These formulas make explicit that interference terms can enhance or suppress GMN in superposed multipartite states.

4. Lower bounds and experimental accessibility

A central recent development is the derivation of fidelity-based analytical lower bounds for geometric-mean GME measures, including the negativity-based $J^{\gamma|\bar\gamma}_{\alpha\beta}$6 (Wang et al., 14 Aug 2025). For a mixed state $J^{\gamma|\bar\gamma}_{\alpha\beta}$7,

$J^{\gamma|\bar\gamma}_{\alpha\beta}$8

Here $J^{\gamma|\bar\gamma}_{\alpha\beta}$9, and the $N_g(|\psi\rangle) = \min_m N_m(|\psi\rangle),$0 are defined from the fidelity $N_g(|\psi\rangle) = \min_m N_m(|\psi\rangle),$1 together with the first and second largest Schmidt coefficients of $N_g(|\psi\rangle) = \min_m N_m(|\psi\rangle),$2 over all bipartitions (Wang et al., 14 Aug 2025).

This lower bound is reported to be tight for structured states such as generalized GHZ states, and its experimental relevance lies in the fact that it avoids full state tomography and can be estimated from fidelity measurements with a suitably chosen observable pure state (Wang et al., 14 Aug 2025). The same source states that any observable pure state can be used to estimate the lower bound and that optimization over observable states can tighten it. It further claims improvement over earlier bounds whenever the maximal Schmidt coefficients across different bipartitions are unequal.

These developments sharpen a long-standing distinction in GMN practice. Exact convex-roof or mixed-convex-roof evaluation is generally difficult, but witness methods, fidelity-based bounds, and symmetry-restricted formulas make nontrivial cases accessible. This suggests a division of labor within the subject: exact formulas for highly structured families, and experimentally friendly lower bounds for generic mixed states.

5. GMN within broader frameworks of multipartite entanglement measures

Negativity-based genuine multipartite measures have also been analyzed within broader axiomatic frameworks for genuine multipartite entanglement measures (GMEMs). One such framework specifies the conditions $N_g(|\psi\rangle) = \min_m N_m(|\psi\rangle),$3 for any biseparable state, $N_g(|\psi\rangle) = \min_m N_m(|\psi\rangle),$4 for any genuinely entangled state, convexity, and monotonicity under LOCC (Guo, 2021). Within that setting, a negativity-based measure $N_g(|\psi\rangle) = \min_m N_m(|\psi\rangle),$5 is treated as a unified GMEM, but not as a complete one; correspondingly, it is completely monogamous but not tightly completely monogamous (Guo, 2021).

A related classification based on reduced functions reaches a similar conclusion for negativity-based constructions. The negativity reduced function is described as strictly concave but not subadditive, and the associated negativity-based GMEM can be unified and completely monogamous but is not necessarily a complete GMEM (Guo, 2023). The same work states that, for $N_g(|\psi\rangle) = \min_m N_m(|\psi\rangle),$6, negativity-based GMEMs fail completeness in contrast to certain entropy-based or concurrence-based constructions.

These results do not invalidate GMN as a GME monotone. Rather, they place it within a hierarchy of multipartite entanglement quantifiers. A recurring point in this literature is that different negativity-based GMNs are not interchangeable: the minimum-over-cuts form, the witness/mixed-convex-roof form, and the geometric-mean form all satisfy the core requirement of vanishing on biseparable states, but they differ in how they encode inter-cut structure and in how they behave under completeness and monogamy criteria.

6. Dynamical applications, many-body settings, and terminological cautions

GMN has been used as a dynamical probe in open-system and relativistic settings. In the geometric-mean framework, negativity-based GME measures are applied to the sudden death of genuine quadripartite entanglement for a four-qubit generalized GHZ state in individual non-Markovian environments, where the initial weight $N_g(|\psi\rangle) = \min_m N_m(|\psi\rangle),$7 affects the timing of entanglement death and analytic conditions are derived for its occurrence (Wang et al., 14 Aug 2025). The same framework is applied to Dirac particles with Hawking radiation in the background of a Schwarzschild black hole; because exact GME concurrence or negativity is not analytically tractable for the evolved mixed state, fidelity-based lower bounds are used to estimate the multipartite entanglement under relativistic and gravitational effects (Wang et al., 14 Aug 2025).

A later many-body application studies the Kitaev quantum spin liquid coupled to a dissipative bosonic bath and tracks GMN in reduced subregions of the time-evolved density matrix (Garcia-Gaitan et al., 2 Oct 2025). In the Markovian regime, GMN remains non-zero in larger loopy subregions up to temperatures $N_g(|\psi\rangle) = \min_m N_m(|\psi\rangle),$8, while in the non-Markovian regime it remains non-zero up to $N_g(|\psi\rangle) = \min_m N_m(|\psi\rangle),$9 and can also become non-zero in smaller non-loopy subregions (Garcia-Gaitan et al., 2 Oct 2025). This use of GMN emphasizes its sensitivity to multipartite structure beyond conventional correlators and Wilson-loop expectation values.

A persistent source of confusion is terminological. In several Bell-theoretic papers, the acronym “GMN” denotes genuine multipartite nonlocality rather than genuine multipartite negativity (Gebhart et al., 2021, 2207.13579). The two notions are distinct: one is an entanglement measure, the other a form of multipartite Bell nonlocality. For negativity-based GME work, the context is therefore decisive.

Taken together, the arXiv literature presents GMN as a family of closely related negativity-based monotones for GME, with three major themes. First, GMN can be defined through the least entangled bipartition, through witness-based mixed-convex-roof constructions, or through the geometric mean over all bipartitions. Second, despite the general hardness of convex-roof problems, exact results are available for highly symmetric state families and tight bounds exist for superpositions and fidelity-accessible experiments. Third, GMN has become a practical diagnostic in dynamical multipartite settings ranging from non-Markovian decoherence to curved-spacetime quantum fields, while remaining conceptually embedded in larger discussions of monogamy, completeness, and the geometry of multipartite entanglement.