arXiv:2609.18002
Algorithmic Design of Heralded Linear Optical Circuits for Multipartite Entanglement
Jaehee Kim, Hon Wai Lau, William J. Munro, Joonsuk Huh, Seungbeom Chin
Abstract
Heralded multipartite entanglement is a key resource for various quantum information tasks. However, designing linear optical circuits that generate specific target states is generally challenging due to the complexity of the required optical structures. Here we formulate the design of heralded photonic circuits as an algorithmic graph-search problem. Our framework enables the automated construction and optimization of heralded photonic circuits by substantially reducing the search space using the linear quantum graph (LQG) picture. Our strategy reconstructs circuit structures as graphs in the picture and identifies suitable graphs automatically. As a result, we design efficient schemes for a broad range of useful multipartite resource states, including hypergraph magic states, quantum error correcting codes, general three-qubit states and length-1 caterpillar graph states. Our work establishes an algorithmic framework for the systematic discovery of heralded resource states, laying the foundation for the automated design of increasingly complex multipartite entangled resources.
AI Summary
The paper addresses automated design of heralded linear-optical circuits that prepare multipartite entanglement while leaving the output photons available for use. It searches a restricted class of effective-perfect-matching (EPM) graphs in the existing linear quantum graph formalism, with $N$ output qubits, $M$ ancillas and $2N+M$ input photons. The workflow enumerates non-isomorphic graphs, filters by a necessary strong-connectivity condition for genuine multipartite entanglement, and computes states from perfect matchings. Target searches use reduced-state spectra, permutations, local-unitary optimization and adjustable edge weights. Graph-to-circuit rules produce optical networks, whose outputs and heralding probabilities are checked by creation-operator evolution, including outcome-dependent single-qubit phase corrections. Results include an eight-photon $CCZ|+++\rangle$ preparation with success probability $1/128$, versus approximately $3.4\times10^{-3}$ for a cited six-photon scheme, and an eleven-photon $CCCZ|++++\rangle$ preparation with probability approximately $3.51\times10^{-4}$. Parametrized circuits prepare arbitrary three-qubit pure states up to local unitaries, although generic prescribed phases require accepting only a subset of detector outcomes. Finite-size graph patterns yield general constructions for $[[2k,1,2]]$ encoded states, with success probability $1/(2\cdot4^{2k})$ for nonzero logical amplitudes, and length-1 caterpillar graph states. The stated limitations are combinatorial enumeration cost, restriction to EPM-representable circuits rather than all heralded linear-optical networks, and the absence of a fully constructive arbitrary-target design method. Enumeration reaches only small systems; the $(5,3)$ run takes approximately $1.48\times10^6$ seconds. Numerical local-unitary searches can confirm matches but cannot certify inequivalence when optimization fails.
Error found: The stated endpoint limits for the three-qubit amplitude-damping-code success probability exceed the limits of its displayed formula by a factor of three. The general three-qubit circuit appendix incorrectly equates the success probability for a prescribed generic phase with a formula that sums twelve distinct phase classes, overstating that prescribed-phase probability by a factor of twelve.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- REVTex 4.2, 19 pages including Appendix, comments welcome