Structural Physical Approximations

• Structural Physical Approximations are convex constructions that transform non-completely positive maps into implementable quantum channels by mixing them with maximally mixed states.
• They facilitate practical entanglement detection by converting abstract entanglement witnesses into experimentally realizable forms through optimal parameter selection.
• The disproval of the SPA conjecture for decomposable witnesses reveals that not all optimal entanglement criteria yield separable, entanglement-breaking outputs.

A structural physical approximation (SPA) is a convex construction designed to render a positive but not completely positive (non-CP) map or entanglement witness (EW) physically implementable as a quantum channel or as a legitimate state. The SPA protocol has been central in quantum information theory for formulating experimentally realizable entanglement detection and for probing the subtle mathematical structure of positive maps, entanglement witnesses, and the associated separability and entanglement-breaking (EB) channels. Its study has led to important progress including the disproof of the long-standing SPA conjecture for decomposable entanglement witnesses.

1. Formal Construction of Structural Physical Approximations

The SPA is defined for Hermitian operators and positive maps that are not physical in themselves due to non-complete positivity. For a Hermitian operator $W$ acting on a finite-dimensional bipartite Hilbert space $\mathcal{H}_A \otimes \mathcal{H}_B$ of total dimension $D = d_A d_B$, normalized such that $\operatorname{Tr} W = 1$, $W$ is an entanglement witness if:

• $$\langle \psi \otimes \phi| W |\psi \otimes \phi \rangle \geq 0$$ for all product vectors $|\psi \otimes \phi\rangle$,
• $W$ has at least one negative eigenvalue.

To "physicalize" $W$, one forms a convex mixture with the maximally mixed state:

$$W(p) = p W + (1-p)\frac{I}{D}, \qquad 0 \leq p \leq 1.$$

There exists a maximal $\mathcal{H}_A \otimes \mathcal{H}_B$0 such that $\mathcal{H}_A \otimes \mathcal{H}_B$1 for all $\mathcal{H}_A \otimes \mathcal{H}_B$2. The SPA of $\mathcal{H}_A \otimes \mathcal{H}_B$3 is defined as:

$\mathcal{H}_A \otimes \mathcal{H}_B$4

This construction equivalently applies to positive but not CP maps $\mathcal{H}_A \otimes \mathcal{H}_B$5, for which the SPA channel is

$\mathcal{H}_A \otimes \mathcal{H}_B$6

with $\mathcal{H}_A \otimes \mathcal{H}_B$7 as large as possible so $\mathcal{H}_A \otimes \mathcal{H}_B$8 is CP (Chruściński et al., 2013).

2. The SPA Conjecture and Its Scope

Korbicz–Almeida–Lewenstein–Acín et al. conjectured (commonly called the SPA conjecture) that the SPA to any optimal positive map is entanglement-breaking, i.e., the SPA outputs a separable state for any input. In the witness language:

SPA Conjecture: Let $\mathcal{H}_A \otimes \mathcal{H}_B$9 be an optimal EW (normalized, $D = d_A d_B$0), define $D = d_A d_B$1, and let $D = d_A d_B$2 be maximal such that $D = d_A d_B$3. Then $D = d_A d_B$4 is separable.

An EW is optimal if no strictly finer EW exists (i.e., none detects a strictly larger set of entangled states), or equivalently, no positive operator can be subtracted from $D = d_A d_B$5 while retaining the entanglement witness property (Chruściński et al., 2013).

The conjecture is motivated by the physical intuition that minimal structural noise should suffice to destroy all quantum correlations for optimal entanglement detection, which would significantly facilitate experimental implementation.

3. Counterexample: Decomposable Witnesses with Entangled SPAs

Chruściński and Sarbicki (Chruściński et al., 2013) disproved the SPA conjecture for decomposable witnesses by constructing a one-parameter family of optimal decomposable entanglement witnesses $D = d_A d_B$6 ($D = d_A d_B$7) in $D = d_A d_B$8 dimensions:

• They define three Bell-type projectors $D = d_A d_B$9 with $\operatorname{Tr} W = 1$0 constructed via Weyl operators acting on the maximally entangled state $\operatorname{Tr} W = 1$1.
• Then, for an explicit $\operatorname{Tr} W = 1$2, they form $\operatorname{Tr} W = 1$3, where $\operatorname{Tr} W = 1$4 denotes partial transposition.
• $\operatorname{Tr} W = 1$5 is decomposable, has a threefold degenerate negative eigenvalue, and is optimal by the spanning-kernel criterion.
• Its SPA is $\operatorname{Tr} W = 1$6 with $\operatorname{Tr} W = 1$7 at positivity threshold.

Since $\operatorname{Tr} W = 1$8 is decomposable, $\operatorname{Tr} W = 1$9 (i.e., $W$0 is PPT). However, applying the realignment criterion reveals that for certain $W$1 (including $W$2 explicitly), the realigned matrix of $W$3 has $W$4, certifying that $W$5 is PPT entangled (bound entangled) and not separable. This directly disproves the SPA conjecture in the decomposable case (Chruściński et al., 2013).

4. Special Case: Extremal Rank-One Decomposable Witnesses

Despite the general failure of the SPA conjecture, it does hold for extremal decomposable EWs of the form $W$6, with $W$7 a pure entangled state. For these, the SPA is of "isotropic" form:

$W$8

with $W$9 (where $$\langle \psi \otimes \phi| W |\psi \otimes \phi \rangle \geq 0$$0 are Schmidt coefficients of $$\langle \psi \otimes \phi| W |\psi \otimes \phi \rangle \geq 0$$1). The Vidal–Tarrach criterion implies $$\langle \psi \otimes \phi| W |\psi \otimes \phi \rangle \geq 0$$2 is separable if and only if it's PPT, and this bound is sharp. Therefore, the SPA is always separable (i.e., entanglement-breaking) in the extremal rank-one decomposable case (Chruściński et al., 2013).

5. Consequences for Entanglement Detection and Quantum Channels

• The existence of decomposable EWs with entangled (PPT) SPAs shows that the SPA construction does not always yield entanglement-breaking (or separable) states, even for optimal decomposable EWs.
• For practical entanglement detection and the physical implementation of positive but non-CP maps (e.g., in quantum optics), SPA no longer provides an automatic guarantee that the underlying operation is entanglement-breaking. Thus, further analysis (e.g., via the realignment or PPT criterion) is necessary for each case.
• This result prompts a refinement of the SPA conjecture, focusing on the characterization of exactly which EWs or positive maps admit separable (or entanglement-breaking) SPAs and which do not. The clear distinction between Bell-isotropic (extremal rank-one) and Bell-diagonal (non-extremal) families is an important step in that direction (Chruściński et al., 2013).
• Methodologically, combining the SPA construction with realignment and PPT checks provides a powerful approach for investigating separability, PPT entanglement, and the boundary of the set of CP maps.

6. Broader Context and Theoretical Outlook

• The disproof of the SPA conjecture highlights the subtlety of entanglement geometry even among decomposable EWs, which were previously regarded as relatively benign.
• Further developments have established that while the SPA conjecture fails in general, it continues to hold for certain classes, such as extremal rank-one decomposable EWs. This nuanced understanding guides current efforts at identifying structural, spectral, or symmetry-based criteria to delineate when SPAs yield entanglement-breaking channels.
• The SPA machinery, involving convex mixing, eigenvalue characterization, and connection to operationally meaningful quantities (e.g., using realignment norm or Schmidt decompositions), remains an integral tool for analytical and experimental advances in the study of quantum entanglement and the structure of positive maps (Chruściński et al., 2013).