Table of Contents
Fetching ...

A Hierarchy of Deviation from Complete Positivity and Optimal Entanglement Witnesses

Mohsen Kian

TL;DR

The paper develops a spectral framework to quantify how far a Hermitian map is from complete positivity by introducing the CP-distance $d_{\mathrm{CP}}(\Phi)$ and its directional variant $d_\Gamma(\Phi)$, with explicit formulas in terms of the Choi matrix. It then generalizes this idea to a hierarchy $d_k(\Phi)$ based on the cones of $k$-positive maps, deriving a sharp spectral characterization via Schmidt-rank-$k$ vectors and the Choi matrix, which yields a rigorous threshold for certifying Schmidt numbers. The results provide a universal construction of dimension-sensitive entanglement witnesses: from any Hermitian map, one obtains optimal witnesses for Schmidt-number detection by shifting the Choi matrix by $d_k(\Phi)I_{mn}$. The framework links cone geometry, robustification against depolarizing noise, and entanglement certification, enabling scalable witness construction and stability analyses across dimensions.

Abstract

We introduce the \emph{CP-distance} to quantify the deviation of Hermitian linear maps from complete positivity, defined as the minimal depolarizing noise required to render a map completely positive. We derive a closed spectral formula for this distance and extend the framework to \emph{directional robustness} against arbitrary completely positive maps, establishing stability and tensor-product properties. Expanding this to the intermediate cones of $k$-positive maps, we introduce a \emph{hierarchy of deviation}, $d_k(Φ)$. We derive a spectral formula for $d_k$ based on entanglement depth and demonstrate that it serves as an optimal threshold for certifying Schmidt numbers, allowing for the universal construction of dimension-sensitive entanglement witnesses.

A Hierarchy of Deviation from Complete Positivity and Optimal Entanglement Witnesses

TL;DR

The paper develops a spectral framework to quantify how far a Hermitian map is from complete positivity by introducing the CP-distance and its directional variant , with explicit formulas in terms of the Choi matrix. It then generalizes this idea to a hierarchy based on the cones of -positive maps, deriving a sharp spectral characterization via Schmidt-rank- vectors and the Choi matrix, which yields a rigorous threshold for certifying Schmidt numbers. The results provide a universal construction of dimension-sensitive entanglement witnesses: from any Hermitian map, one obtains optimal witnesses for Schmidt-number detection by shifting the Choi matrix by . The framework links cone geometry, robustification against depolarizing noise, and entanglement certification, enabling scalable witness construction and stability analyses across dimensions.

Abstract

We introduce the \emph{CP-distance} to quantify the deviation of Hermitian linear maps from complete positivity, defined as the minimal depolarizing noise required to render a map completely positive. We derive a closed spectral formula for this distance and extend the framework to \emph{directional robustness} against arbitrary completely positive maps, establishing stability and tensor-product properties. Expanding this to the intermediate cones of -positive maps, we introduce a \emph{hierarchy of deviation}, . We derive a spectral formula for based on entanglement depth and demonstrate that it serves as an optimal threshold for certifying Schmidt numbers, allowing for the universal construction of dimension-sensitive entanglement witnesses.

Paper Structure

This paper contains 6 sections, 12 theorems, 98 equations.

Key Result

Theorem 2.1

Let $\Phi: \mathbb{M}_m \to \mathbb{M}_n$ be a Hermitian linear map. Then, there exists a smallest non-negative scalar $k_\Phi$ such that the map $\Psi: \mathbb{M}_m \to \mathbb{M}_n$ defined by: is completely positive and satisfies $\Phi \leq \Psi$.

Theorems & Definitions (32)

  • Theorem 2.1
  • proof
  • Example 2.2
  • Remark 2.3
  • Proposition 2.4
  • proof
  • Definition 3.1
  • Remark 3.2
  • Theorem 3.3
  • proof
  • ...and 22 more