Quantum power iterations via tensor networks

Abstract:In optimization, one of the well-known classical algorithms is power iterations, which recovers the dominant eigenvector of a diagonalizable matrix. Since numerous optimization problems can be formulated as an eigenvalue/eigenvector search, this algorithm features wide applicability. Operationally, power iterations perform repeated matrix-to-vector multiplications (\texttt{MatVec}) followed by a renormalization step to converge to the dominant eigenvalue/eigenvector. However, classical tensor network realizations can exhibit worst-case exponential growth of tensor ranks with the number of iterations. We propose a quantum realization offering a possible route around this bottleneck. Our method involves casting low-rank representations---MPO for matrices and MPS for vectors---into quantum circuits. Specifically, we recover a unitary approximation by variationally minimizing the Frobenius distance between a target MPO and an MPO ansatz wherein the tensor cores are constrained to unitaries. Such unitary MPOs can be implemented as quantum circuits with ancillary qubits. With appropriate initialization and post-selection on the ancillary space, we realize a single classical iteration. With our proposed methodology, power iterations can be realized entirely on a quantum computer via repeated, static circuit blocks. By exploiting Riemannian optimization and cross-approximation techniques, our methodology becomes instance-agnostic, allowing one to address black-box optimization of functions admitting low-rank tensor network representations within quantum computing. We emphasize that this work establishes the construction and analyzes its reliable operating conditions; it does not demonstrate an end-to-end computational advantage over classical tensor network optimization, which we discuss explicitly.

arXiv:2404.15048 [quant-ph]

(or arXiv:2404.15048v2 [quant-ph] for this version)

https://doi.org/10.48550/arXiv.2404.15048 Focus to learn more arXiv-issued DOI via DataCite

Phys. Rev. Research 2026

https://doi.org/10.1103/sr9b-wg4l Focus to learn more DOI(s) linking to related resources

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.