ReCon: Reconfiguring Analog Rydberg Atom Quantum Computers for Quantum Generative Adversarial Networks

TL;DR

ReCon is proposed, the first work to implement quantum GANs on analog Rydberg atom quantum computers, and shows 33% better quality in generated images than the state-of-the-art technique implemented on superconducting-qubit technology.

Abstract

Quantum computing has shown theoretical promise of speedup in several machine learning tasks, including generative tasks using generative adversarial networks (GANs). While quantum computers have been implemented with different types of technologies, recently, analog Rydberg atom quantum computers have been demonstrated to have desirable properties such as reconfigurable qubit (quantum bit) positions and multi-qubit operations. To leverage the properties of this technology, we propose ReCon, the first work to implement quantum GANs on analog Rydberg atom quantum computers. Our evaluation using simulations and real-computer executions shows 33% better quality (measured using Frechet Inception Distance (FID)) in generated images than the state-of-the-art technique implemented on superconducting-qubit technology.

Figures (20)

• Figure 1: (a) The analog quantum model involves the evolution of quantum register under some (time-dependent) Hamiltonian $(H)$ followed by qubit measurement at the end (circular meters). (b) Atoms can be reconfigured into different positions that affect their interaction strength.
• Figure 2: Rydberg atom quantum computers can apply global operational pulses that affect all qubits simultaneously.
• Figure 3: Training of a classical GAN. Once trained, the generator is used to run inference tasks to generate images.
• Figure 4: ReCon uses PCA and inverse PCA in its design.
• Figure 5: ReCon parameterizes the qubit positions (spatial), as well as the amplitudes of the local detuning, constant global detuning (not shown), and global frequency of the Hamiltonian pulses (temporal).