Variational M-Partite Geometric Entanglement Algorithm

Vahid Azimi-Mousolou, Prashant Singh · 2025

Read Variational M-Partite Geometric Entanglement Algorithm by Vahid Azimi-Mousolou, Prashant Singh (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-2503-20056.txt.

Introduction

Vahid Azimi-Mousolou and Prashant Singh present a 2025 quantum algorithm that measures how entangled a many part quantum system is. Entanglement links parts of a system so they cannot be described apart, and measuring it in large systems is a known hard problem. Near term quantum machines need methods that scale without huge cost.

Their method builds custom trial circuits for a system split into any number of parties, then tunes them to match the target state as closely as a fully separate state can. The match gap gives the entanglement measure, called the geometric measure. Checks against known exact answers confirmed the method's accuracy. The team also showed it at work on both standard and unusual splits of one and two dimensional systems.

Catalog record

CREATORVahid Azimi-Mousolou, Prashant Singh
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE19 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-20056

Browse the full library · About the Tower of Babel

Full text

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

Variational M-Partite Geometric Entanglement Algorithm Vahid Azimi-Mousolou, Prashant Singh Abstract We propose a variational quantum algorithm to evaluate the M M -partite geometric entanglement across arbitrary partitions of an N N -qubit system into M M parties. By constructing tailored variational ansatz circuits for both single- and multi-qubit parties, we optimize the overlap between a target quantum state and an M M -partite variational separable state. This method provides a flexible and scalable approach for characterizing arbitrary M M -partite entanglement in complex quantum systems of a given dimension. We assess the accuracy of the proposed method by reproducing known analytical results and demonstrate its capability to evaluate entanglement among M M parties for any given conventional or unconventional partitions of one- and two-dimensional spin systems, both near and at a quantum critical point. Our results establish the versatility of the variational approach in capturing different types of entanglement in various quantum systems, surpassing the capabilities of existing methods. Our approach offers a powerful methodology for advancing research in quantum information science, condensed matter physics, and quantum field theory. Additionally, we discuss its advantages, highlighting its adaptability to diverse system architectures in the context of near-term quantum devices. I Introduction Entanglement is a cornerstone of quantum mechanics and plays a critical role in diverse areas ranging from quantum information processing to condensed matter physics and quantum field theory Horodecki2009 . While bipartite entanglement measures, such as concurrence and negativity, are well-established, quantifying entanglement, especially in multipartite systems, remains a challenging task due to the exponential growth of the Hilbert space and the complexity of separability criteria Horodecki2009 ; Plenio2007 . Existing methods, such as the density matrix renormalization group (DMRG) Eisert2010 ; Schollwock2011 and tensor network techniques, including matrix product states (MPS), projected entangled pair states (PEPS) Verstraete2008 , and tree tensor networks (TTN) Pizorn2013 , have provided significant insights into ground state properties and entanglement structures, particularly in one-dimensional systems. While DMRG, which is closely related to MPS, is highly efficient for 1D gapped systems, it struggles with two- and higher-dimensional models due to an exponential growth in computational cost. MPS itself faces similar limitations, particularly in capturing long-range entanglement and critical systems. PEPS generalizes MPS to higher dimensions but suffers from computationally expensive tensor contractions and optimization challenges. TTN, while effective for hierarchical structures, lacks translation invariance and is less suited for capturing entanglement in complex geometries. Stochastic approaches, like quantum Monte Carlo methods, including variational Monte Carlo (VMC) Becca2017 , on the other hand, offer greater flexibility in wavefunction design and can handle higher-dimensional systems. While VMC has primarily been focused on energy minimization, recent studies have demonstrated its potential for evaluating bipartite and multipartite entanglement Ferris2012 ; Deng2017 ; Harney2020 . Despite these advances, the available methods still face significant challenges in quantifying multipartite entanglement and characterizing entanglement structures in high-dimensional, complex geometries. Therefore, there is a strong need for efficient and accurate approaches to evaluate various types of entanglement, particularly beyond bipartite entanglement. Here, we present a variational M -partite geometric entanglement (VMGE) algorithm, which is generic and dresses the current challenges to a great extent. Our method considers N -qubit system partitioned into an arbitrary M parties, defining M -partite geometric entanglement (GE) through the minimal distance between the target state and the set of parametrized variational M-partite separable states. We construct variational circuits tailored to each party, optimizing over universal parametrized single-qubit rotations and multi-qubit entangling gates with variational entanglement power to maximize overlap with the target state. The optimization is driven by a synergy between the non-gradient variational Monte Carlo method and gradient-based Broyden–Fletcher–Goldfarb–Shanno (BFGS) method Nocedal1999 . The VMGE algorithm provides a hybrid quantum-classical framework suitable for near-term quantum devices. We demonstrate the efficacy of our approach through detailed mathematical formulation and circuit design. Our results highlight the algorithm’s scalability, flexibility, making it a promising universal tool for multipartite entanglement quantification in complex quantum systems. This work opens new avenues for exploring multipartite entanglement in both theoretical and experimental settings, with potential applications in quantum computing, quantum simulation, and beyond. II Method II.1 Variational quantum algorithm For an N -qubit system partitioned into M parties, let us assume \mathcal{S}_{M} is the set of all M-partite separable states, i.e., states that can be written as \displaystyle|\phi\rangle=|\phi_{1}\rangle\otimes|\phi_{2}\rangle\otimes\cdots\otimes|\phi_{M}\rangle, (1) where each |\phi_{i}\rangle is a state involving m_{i} qubits, with \sum_{i=1}^{M}m_{i}=N . For a given N -qubit target state, |\psi\rangle , the M-partite GE is defined as Wei2003 \displaystyle E^{(M)}(|\psi\rangle)=1-\max_{|\phi\rangle\in\mathcal{S}_{M}}|\langle\phi|\psi\rangle|^{2}. (2) Fig. 1 illustrates a variational quantum algorithm to evaluate the overlap of between the N -qubit target state and a variational ansatz for M -Partite Separable States. Figure 1: (Color online) Variational M-partite geometric entanglement (VMGE) algorithm. For an N -qubit system, the algorithm involves M disjoint variational circuits corresponding to a given partition of the system into M parties. These circuits generate a general M -partite separable state, which is then compared with a given N -qubit target state |\psi\rangle . The geometric entanglement is evaluated through an optimization over the separable state space. For each party involving m qubits, the variational ansatz is constructed through a variational circuit, U(q_{i_{1}},q_{i_{2}},\dots,q_{i_{m}}) , shown in Fig 2 . Figure 2: (Color online) Variational single-qubit ( a ) and multi-qubit ( b ) circuits used in the VMGE algorithm, as illustrated in Fig. 1 . The algorithm employs universal single-qubit circuits combined with a two-qubit entangling circuits ( c ) with variational entangling power. The circuit is described as follows i) Single-qubit party, m=1 : As shown in Fig 2 ( a ) the variational circuit simplifies to \displaystyle U(q_{j})=R_{z}(\varphi_{j})R_{x}(\theta_{j}) (3) ii) Multi-qubit party, m>1 : In this case \displaystyle U(q_{i_{1}},q_{i_{2}},\dots,q_{i_{m}})=\bigotimes_{l=1}^{m}R_{z}(\varphi_{l})\bigotimes_{l=1}^{m-1}U_{l(l+1)}\bigotimes_{l=1}^{m}R_{x}(\theta_{l}) specifies the variational circuit in Fig 2 ( b ), where U_{l(l+1)} is a two-qubit entangling gate on nearest-neighbor qubit pairs. We assume a general nonlocal two-qubit operation \displaystyle U_{l(l+1)}=\exp\left[-\frac{i}{2}\sum_{i=1}^{3}c_{li}\sigma_{i}^{l}\sigma_{i}^{l+1}\right] (5) where c_{l1},c_{l2} , and c_{l3} are variational parameters controlling the entanglement power of the circuit up to arbitrary value. The real vector (c1,c2,c3) is a geometric coordinate on a 3-Torus, which represents local invariants restricted to the tetrahedral Weyl chamber Zhang2003 . Fig 2 ( c ) is the circuit representation of U_{l(l+1)} . By initializing all N qubits in the |0\rangle state, the overall M-Partite variational separable state is \displaystyle|\phi(\vec{\theta},\vec{\varphi},\vec{c})\rangle=\bigotimes_{j=1}^{M}U(q_{i_{1}},q_{i_{2}},\dots,q_{i_{m_{j}}})|0\rangle^{\otimes N}. (6) This lead to the objective function for measureing the overlap between the target state |\psi\rangle and the variational state |\phi\rangle as \displaystyle\mathcal{L}(\vec{\theta},\vec{\varphi},\vec{c})=|\langle\phi(\vec{\theta},\vec{\varphi},\vec{c})|\psi\rangle|^{2}. (7) Using a classical optimizer the proposed variational quantum algorithm allows to compute the M-partite GE through the optimization \displaystyle E^{(M)}(|\psi\rangle)=1-\max\mathcal{L}(\vec{\theta},\vec{\varphi},\vec{c})=\min\left(1-\mathcal{L}(\vec{\theta},\vec{\varphi},\vec{c})\right), (8) where maximize and minimum are taken over the variational parameters \vec{\theta},\vec{\varphi},\vec{c} . II.2 Optimization procedure We solve the optimization problem formulated above using a two-stage hybrid optimization process. In the first stage, a Sobol sequence joe2008 of n_{init}=2000 samples is used to perform a ’global’ Monte-Carlo search within the interval [0,2\pi] for each parameter to identify the parameter combination \vec{\theta}^{*},\vec{\varphi}^{*},\vec{c}^{*} offering the best overlap. This combination is then used as the starting point in the second stage composed of optimization using the quasi-Newton Broyden–Fletcher–Goldfarb–Shanno (BFGS) method Nocedal1999 . III Results To validate our scheme, we evaluate different types of GE for quantum states with known analytical solutions

Read the complete text