We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Quantum Physics

Title: Automatically Differentiable Quantum Circuit for Many-qubit State Preparation

Abstract: Constructing quantum circuits for efficient state preparation belongs to the central topics in the field of quantum information and computation. As the number of qubits grows fast, methods to derive large-scale quantum circuits are strongly desired. In this work, we propose the automatically differentiable quantum circuit (ADQC) approach to efficiently prepare arbitrary quantum many-qubit states. A key ingredient is to introduce the latent gates whose decompositions give the unitary gates that form the quantum circuit. The circuit is optimized by updating the latent gates using back propagation to minimize the distance between the evolved and target states. Taking the ground states of quantum lattice models and random matrix product states as examples, with the number of qubits where processing the full coefficients is unlikely, ADQC obtains high fidelities with small numbers of layers $N_L \sim O(1)$. Superior accuracy is reached compared with the existing state-preparation approach based on the matrix product disentangler. The parameter complexity of MPS can be significantly reduced by ADQC with the compression ratio $r \sim O(10^{-3})$. Our work sheds light on the "intelligent construction" of quantum circuits for many-qubit systems by combining with the machine learning methods.
Comments: 5 pages, 5 figures
Subjects: Quantum Physics (quant-ph); Strongly Correlated Electrons (cond-mat.str-el); Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
Journal reference: Phys. Rev. A 104, 042601 (2021)
DOI: 10.1103/PhysRevA.104.042601
Cite as: arXiv:2104.14949 [quant-ph]
  (or arXiv:2104.14949v1 [quant-ph] for this version)

Submission history

From: PengFei Zhou [view email]
[v1] Fri, 30 Apr 2021 12:22:26 GMT (1755kb,D)

Link back to: arXiv, form interface, contact.