arXiv:2610.28349
CO - COmpiling PHotonics into dIgital quantum circuits
Grzegorz Czelusta, Marcin Markiewicz, Jakub Mielczarek, Artur Miroszewski
Abstract
Photonic and qubit-based (digital) approaches to quantum computing are largely regarded as separate paradigms. This article aims to bridge these paradigms by developing a method for mapping quantum-optical systems onto qubit-based quantum circuits. The key ingredient enabling this mapping is the introduction of a truncation of the Fock space describing single-mode optical states. Within this framework, representations of basic photonic operations are developed, including the phase-rotation operator, the displacement operator, the squeezing operator, and the beam splitter. Explicit quantum circuit constructions are provided for Fock-space truncations of dimension $N=2$ and $N=4$. We discuss benchmarking of the implemented operations and identify the regime of applicability and the limitations of the proposed method. The presented results can be used to simulate quantum optical circuits on qubit-based quantum computers, with special relevance to Gaussian Boson Sampling in the low-photon-number regime. To demonstrate the practical applicability of this approach, we complement the theoretical framework with a Python-based quantum programming tool CO$Φ$ that compiles Blackbird quantum-optics programs into qubit-based OpenQASM circuits using a truncated Fock-space encoding. The compiler is available on GitHub at https://github.com/Quantum-Cosmos-Lab/cophi.
AI Summary
The paper addresses how to compile quantum-optical circuits into gate-based qubit circuits while retaining unitary dynamics in a finite-dimensional approximation. It truncates each mode to Fock states $|0\rangle,\ldots,|N-1\rangle$, encodes photon numbers in binary using $\log_2N$ qubits for power-of-two cutoffs, and exponentiates generators built from mutually adjoint truncated ladder operators. An algebraic connection to root-of-unity $q$-deformed oscillators is also developed, separately from the unitary optical implementation. Explicit circuit decompositions are given for phase shifts, displacements, squeezing and beam splitters at $N=2$ and $N=4$. For the four-qubit, $N=4$ beam splitter, photon-number block structure enables an optimized circuit with 22 CNOTs, six controlled-$Y$ gates, 36 single-qubit gates and depth 45. Reported comparisons require 115 CNOTs for the block-by-block construction and 93 for generic quantum Shannon decomposition. The accompanying Python compiler, CO$\Phi$, translates Blackbird programs containing these operations into OpenQASM using PennyLane. Phase shifts are exact throughout the retained space. Passive interferometric evolution is guaranteed exact when total photon number satisfies $n_{\mathrm{tot}}\leq N-1$; higher sectors generally differ from full optical evolution despite remaining unitary. Displaced and squeezed vacuum generation are benchmarked numerically through relative mean-photon-number errors for $N=4,8,16,32$, rather than state-fidelity or sampling-distribution bounds. At $N=2$, squeezing is identically trivial. Larger-cutoff circuit constructions remain future work, and applicability to Gaussian Boson Sampling is restricted by both state-generation errors and cutoff-modified higher-photon-number dynamics.
The optimisation process has been conducted with the use of Claude AI (Opus 5).
Error found: The N=4 displacement derivation prints an incorrect matrix M D M†, including incorrect entries used to extract its single-qubit factors, so the displacement circuit is not established correctly as written. The paper identifies its N=2 truncation with fermionic statistics even though independently encoded modes commute across modes; a free-fermion interpretation requires additional restrictions or parity strings.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 38 pages, 16 figures, GitHub repository: https://github.com/Quantum-Cosmos-Lab/cophi