Local Unitary Equivalence of Tripartite Quantum States In Terms of Trace Identities

Isaac Dobes, Naihuan Jing · 2025

Read Local Unitary Equivalence of Tripartite Quantum States In Terms of Trace Identities by Isaac Dobes, Naihuan Jing (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-06240.txt.

Introduction

Isaac Dobes and Naihuan Jing settle in 2025 when two three-party quantum states are the same up to local twists. Local twists are the free moves in quantum labs, so telling true sameness from mere twist is basic bookkeeping. An earlier proof tied the two-party case to a matrix puzzle; their rewrite uses a richer algebra of multi-way arrays.

With that frame, the tie stretches to three parties, they show. A classic test for matrix systems then trims the check to a set of trace identities plus a few extra marks. The results reach any number of parties in principle, with practical limits noted. A finite checklist for sameness turns a wild search into arithmetic.

Catalog record

CREATORIsaac Dobes, Naihuan Jing
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE72 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-06240

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Local Unitary Equivalence of Tripartite Quantum States In Terms of Trace Identities Isaac Dobes, Naihuan Jing Abstract In this paper, we generalize the major results from Jing-Yang-Zhao’s paper ”Local Unitary Equivalence of Quantum States and Simultaneous Orthogonal Equivalence,” which established a correspondence between local unitary (LU) equivalence and simultaneous orthogonal (SO) equivalence of bipartite quantum states, and then used this correspondence to show that the problem of determining LU equivalence of bipartite states reduces to checking trace identities. In particular, we extend both results to tripartite quantum states. We are able to do this by utilizing a hypermatrix algebra framework and by applying a powerful generalization of Specht’s criterion proved in Futorny-Horn-Sergeichuk’s paper ”Specht’s Criterion for Systems of Linear Mappings.” With our established hypermatrix algebra framework and the aforementioned generalization of Specht’s criterion, it is apparent that our results can be extended to arbitrary multipartite quantum states, however there are some practical limitations which are explored and discussed towards the end. 1 Introduction Quantum entanglement is a peculiar phenomenon that has intrigued yet also eluded researchers for nearly a century [ 1 ] . In the last few decades, however, serious interest and efforts in understanding entanglement has grown rapidly due to its applications to quantum computing [ 2 , 3 , 4 ] . Since entanglement pertains to nonlocal properties of quantum states [ 5 ] , a choice of local basis should have no effect on the degree to which a quantum state is entangled [ 6 ] . Therefore, one important task in the study of entanglement is the classification of quantum states up to local unitary (LU) equivalence. Two density matrices \rho and \widehat{\rho} representing states in the same quantum system \mathbb{C}^{d_{1}}\otimes...\otimes\mathbb{C}^{d_{N}} are locally unitary (LU) equivalent if there exist U_{i}\in SU(d_{i}) ( 1\leq i\leq N ) such that [ 7 ] \widehat{\rho}=(U_{1}\otimes...\otimes U_{N})\rho(U_{1}\otimes...\otimes U_{N})^{\dagger}. In general, it is difficult to determine LU equivalence between two density matrices. One of the first major accomplishments in characterizing LU equivalence was due to Yuriy Makhlin in [ 8 ] , where he provided necessary and sufficient conditions for the LU equivalence of 2 -qubit density matrices in terms of their Fano form. The next major development came from Barbara Kraus in [ 9 ] , where she characterized LU equivalence of pure n -qubit states in terms of their ”standard form.” Deriving a state’s ”standard form” involved diagonalizing the corresponding reduced states, and provided an efficient way to check for LU equivalence of pure n -qubit states; however, there are degenerate cases where Kraus’s methods breaks down. Inspired by Kraus’s work, Jun-Li Li, Cong-Feng Qiao, and others represent arbitrary multipartite quantum states and characterize their LU equivalence in terms of their higher order singular value decomposition [ 10 ] . Li and Qiao later extend their results to mixed multipartite states of arbitrary dimension [ 11 ] , however as with Kraus’s method there is a degenerate cases in which their method is very computationally taxing. An alternative approach to characterizing LU equivalence in terms of trace identities is later given by Jing et al. in [ 12 ] . In their paper, they establish a correspondence between simultaneous orthogonal equivalence and quasi-LU equivalence (a property that follows from LU equivalence and in fact implies LU equivalence in certain cases for 2 -qubit density matrices), and then apply a generalization of Specht’s criterion [ 13 ] to show that quasi-LU equivalence of bipartite quantum states reduces to checking trace identities. This provides a nice and efficient way to check for LU equivalence, however there is a minor error in establishing the LU equivalence in the case of 2 -qubit density matrices, and furthermore their methods only apply to bipartite quantum states. In this paper, we present a modified version of the proof in [ 12 ] , utilizing a hypermatrix algebra framework to establish a correspondence between between simultaneous orthogonal equivalence and quasi-LU equivalence of bipartite quantum states, and also LU equivalence in the case of 2 -qubit density matrices. Since our proof is written in the language of hypermatrices, this allows us to generalize the correspondence between simultaneous orthogonal equivalence and quasi-LU equivalence of tripartite quantum states, and again LU equivalence in the case of 3 -qubit density matrices. We then apply a more powerful generalization of Specht’s criterion [ 14 ] to reduce the problem of determining quasi-LU equivalence of tripartite quantum states (or LU equivalence in the case of 3 -qubit density matrices) to trace identities. Indeed, it is apparent that our proofs can, with relative ease, be further generalized to arbitrary dimensional multipartite states, however, there are computational limits to the practicality of our results, hence we stop at the tripartite case. 2 Preliminaries 2.1 Hypermatrix Algebra Most of the concepts and terminology in this subsection can be found [ 15 ] . Let F be a field and suppose V and W are finite-dimensional vector spaces of dimensions m and n , respectively. Just as a linear transformation T:V\rightarrow W can be represented as a matrix A\in F^{m\times n} , similarly a tensor T\in V_{1}\otimes...\otimes V_{d} , where \dim(V_{i})=n_{i} for each 1\leq i\leq d , can be represented as a hypermatrix A\in F^{n_{1}\times...\times n_{d}} . The positive integer d is called the order of a hypermatrix. Suppose A\in F^{n_{1}\times...\times n_{d}} and B\in F^{k_{1}\times...\times k_{e}} are hypermatrices of order d and e . The outer product of A and B , denoted A\circ B , is the hypermatrix of order d+e whose (i_{1},...,i_{d},j_{1},...,j_{e}) coordinate is given by a_{i_{1},...,i_{d}}b_{j_{1},...,j_{e}} . For a hypermatrix A\in F^{n_{1}\times n_{2}\times...\times n_{d}} and matrices X_{1}\in F^{\bullet\times n_{1}},...,X_{d}\in F^{\bullet\times n_{d}} , the multilinear matrix multiplication of (X_{1},...,X_{d}) with A is defined to be the order d hypermatrix A^{\prime}:=(X_{1},...,X_{d})*A\in F^{n_{1}\times...\times n_{d}} , such that A^{\prime}_{i_{1}i_{2}...i_{d}}=\sum\limits_{j_{1},j_{2},...,j_{d}=1}^{n_{1},n_{2},...,n_{d}}(X_{1})_{i_{1}j_{1}}...(X_{d})_{i_{d}j_{d}}A_{j_{1}j_{2}...j_{d}}. Multilinear matrix multiplication is linear in terms of the matrices in both parts; that is, if \alpha,\beta\in F , X_{1},Y_{1}\in F^{m_{1}\times n_{1}} ;…; X_{d},Y_{d}\in F^{m_{d}\times n_{d}} ; and A,B\in F^{n_{1}\times n_{2}\times...\times n_{d}} ; then (X_{1},...,X_{d})*(\alpha A+\beta B)=\alpha(X_{1},...,X_{d})*A+\beta(Y_{1},...,Y_{d})*B and [\alpha(X_{1},...,X_{d})+\beta(Y_{1}...,Y_{d})]*A=\alpha(X_{1},...,X_{d})*A+\beta(Y_{1},...,Y_{d})*B. Next, k:[d]\rightarrow\mathbb{N} . The outer product interacts with multilinear matrix multiplication in the following way \displaystyle\begin{split}&(X_{1_{1}},...,X_{1_{k(1)}},...,X_{d_{1}},...,X_{d_{k(d)}})*(A_{1}\circ...\circ A_{d})\\ &\qquad=(X_{1_{1}},...,X_{1_{k(1)}})*A_{1}\circ...\circ(X_{d_{1}},...,X_{d_{k(d)}})*A_{d}\end{split} (1) where X_{i_{j}}\in F^{m^{(i)}_{j}\times n^{(i)}_{j}} and A_{i}\in F^{n^{(i)}_{1}\times....\times n^{(i)}_{k(i)}} . Consequently, for an arbitrary hypermatrix A\in F^{n_{1}\times...\times n_{d}} , which can be expressed as A=\sum\limits_{k=1}^{r}\alpha_{k}(v_{1}^{(k)}\circ...\circ v_{d}^{(k)}) for some r\in\mathbb{N} and collection of vectors v_{i}^{(k)}\in F^{n_{i}} for each i and k , then (X_{1},...,X_{d})*A=\sum\limits_{k=1}^{r}\alpha_{k}(X_{1}v_{1}^{(k)})\circ...\circ(X_{d}v_{d}^{(k)}) (2) for any X_{1}\in F^{\bullet\times n_{1}} ,…, X_{d}\in F^{\bullet\times n_{d}} . In some cases, it is more convenient to view hypermatrices as matrices. The k-mode unfolding [ 16 ] of a hypermatrix A\in F^{n_{1}\times n_{2}\times...\times n_{d}} is the n_{k}\times(n_{k+1}...n_{d}n_{1}...n_{k-1}) matrix, denoted A_{(k)} , whose (i_{k},j) entry is given by (i_{1},...,i_{d}) -entry of A , with j=1+\sum\limits_{\begin{subarray}{c}l=1\\ l\neq k\end{subarray}}^{d}\left[(i_{l}-1)\prod\limits_{\begin{subarray}{c}m=1\\ m\neq k\end{subarray}}^{l-1}n_{m}\right], or in the case where the index starts at 0 , j=\sum\limits_{\begin{subarray}{c}l=1\\ l\neq k\end{subarray}}^{d}\left[i_{l}\prod\limits_{\begin{subarray}{c}m=1\\ m\neq k\end{subarray}}^{l-1}n_{m}\right]. For example, if A=[A_{i_{1}i_{2}i_{3}}]\in F^{2\times 2\times 2} , then it can be verified that A_{(1)} is the 2\times 4 matrix given by A_{(1)}=\left[\begin{array}[]{cccc}A_{111}&A_{121}&A_{112}&A_{122}\\ A_{211}&A_{221}&A_{212}&A_{222}\end{array}\right]. We call the process of unfolding a hypermatrix A to a matrix A_{(k)} the matricization of A . 2.2 Cayley’s Second Hyperdeterminant Lastly, we review the Cayley’s second hyperdeterminant , also known as the geometric hyperdeterminant [

Read the complete text