Tensor Rank and Stochastic Entanglement Catalysis for Multipartite Pure States

Lin Chen, Eric Chitambar, Runyao Duan, Zhengfeng Ji, Andreas WinterPhysical 2010 · DOI · arXiv

Abstract

The tensor rank (also known as generalized Schmidt rank) of multipartite pure states plays an important role in the study of entanglement classifications and transformations. We employ powerful tools from the theory of homogeneous polynomials to investigate the tensor rank of symmetric states such as the tripartite state $|W_3\rangle = 1/\sqrt{3}(|100\rangle + |010\rangle + |001\rangle)$ and its $N$-partite generalization $|W_N\rangle$. Previous tensor rank estimates are dramatically improved and we show that (i) three copies of $|W_3\rangle$ have a rank of either 15 or 16, (ii) two copies of $|W_N\rangle$ have a rank of $3N - 2$, and (iii) $n$ copies of $|W_N\rangle$ have a rank of $O(N)$. A remarkable consequence of these results is that certain multipartite transformations, impossible even probabilistically, can become possible when performed in multiple-copy bunches or when assisted by some catalyzing state. This effect is impossible for bipartite pure states.

Introduction

In the realm of quantum information, understanding how quantum states can be transformed into one another is a foundational challenge. For bipartite systems (two parties), this process is well-governed by the "Schmidt rank." If the rank of state A is greater than or equal to that of state B, the transformation from A to B is generally possible through Local Operations and Classical Communication (LOCC), at least stochastically. A defining characteristic of the Schmidt rank is its multiplicativity: the rank of two copies of a state is simply the square of the rank of a single copy.

However, as we move into multipartite systems involving three or more parties, the mathematical landscape shifts dramatically. The equivalent measure, known as "tensor rank," is notoriously difficult to compute and, crucially, is not multiplicative. This paper explores the profound physical consequences of this non-multiplicativity, specifically the phenomenon of entanglement catalysis.

Methodology: Bridging Physics and Algebraic Geometry

The researchers address the complexity of multipartite entanglement by establishing a correspondence between quantum physics and algebraic geometry. They focus on symmetric multipartite states—states that remain invariant regardless of how the parties are permuted.

By mapping these quantum states to homogeneous polynomials, the authors leverage established mathematical results regarding "polynomial rank" to estimate the tensor rank of specific quantum states. The study focuses on:

  • Dicke States: Specifically the W-state, where a single excitation is shared across $N$ qubits.
  • Transformation Classes: The study examines Stochastic LOCC (SLOCC), where transformations succeed with non-zero probability, alongside Multi-copy (SMLOCC) and Entanglement-assisted (SELOCC) variations.

The Non-Multiplicativity of Tensor Rank

The core insight of this work is the "Aha!" moment regarding non-multiplicativity. In bipartite systems, the relationship is straightforward: $Rank(A \otimes B) = Rank(A) \times Rank(B)$. In multipartite systems, the authors prove that the rank of multiple copies can be significantly lower than the product of the individual ranks.

This mathematical quirk has a startling physical implication: Catalysis. Because the rank of a bundled state does not grow as quickly as expected, a transformation $|\psi\rangle \to |\phi\rangle$ that is impossible for a single copy might become possible when multiple copies are processed together ($|\psi\rangle^{\otimes n} \to |\phi\rangle^{\otimes n}$) or when an auxiliary "catalyst" state is used to assist the process. This phenomenon is strictly impossible in the bipartite world.

Key Results and Scaling

The paper provides several concrete improvements over previous tensor rank estimates:

  1. Refined Bounds for W-states: While a single $|W_3\rangle$ state has a rank of 3, the authors demonstrate that three copies ($|W_3\rangle^{\otimes 3}$) possess a rank of only 15 or 16. This is a significant reduction from previous upper bounds, which were as high as 21.
  2. Linear Scaling of $|W_N\rangle$: For two copies of an $N$-partite W-state, the rank is shown to be exactly $3N - 2$. More importantly, for $n$ copies, the rank scales linearly with $N$ ($O(N)$), rather than the exponential scaling often feared in multipartite systems.
  3. Demonstration of Catalysis: The authors provide explicit examples of transformations that fail under standard SLOCC but succeed under SMLOCC or SELOCC. For instance, they demonstrate that transformations from GHZ-type states to W-states become possible for $N \geq 5$ parties when catalysis is employed.

Conclusion

By linking the tensor rank of symmetric states to the theory of homogeneous polynomials, this research provides a powerful new toolkit for quantum information theory. The discovery that multipartite entanglement transformations can be "unlocked" through multiple copies or catalysts suggests that multipartite entanglement is a far more flexible resource than previously understood.

While the exact rank of three copies of $|W_3\rangle$ and the precise asymptotic growth rates remain open questions, this work lays the groundwork for a more sophisticated classification of multipartite entanglement and the potential for new protocols in quantum communication and computation.

Summary updated

Abstract

The tensor rank (also known as generalized Schmidt rank) of multipartite pure states plays an important role in the study of entanglement classifications and transformations. We employ powerful tools from the theory of homogeneous polynomials to investigate the tensor rank of symmetric states such as the tripartite state |W3>=1/√3(|100> + |010> + |001>) and its N-partite generalization |W(N)>. Previous tensor rank estimates are dramatically improved and we show that (i) three copies of |W3> have a rank of either 15 or 16, (ii) two copies of |W(N)> have a rank of 3N - 2, and (iii) n copies of |W(N)> have a rank of O(N). A remarkable consequence of these results is that certain multipartite transformations, impossible even probabilistically, can become possible when performed in multiple-copy bunches or when assisted by some catalyzing state. This effect is impossible for bipartite pure states.
BibTeX
@inproceedings{chen2010tensor,
  author = {Lin Chen and Eric Chitambar and Runyao Duan and Zhengfeng Ji and Andreas Winter},
  title = {{Tensor Rank and Stochastic Entanglement Catalysis for Multipartite Pure States}},
  booktitle = {Physical},
  year = {2010},
  doi = {10.1103/physrevlett.105.200501},
}
Figure