Generative modelling powered by room-temperature polariton condensates

2026-06-13

AI Summary

This work proposes a novel approach to generative modeling by harnessing the intrinsic randomness and strong nonlinear dynamics of exciton–polariton condensates in organic dye microcavities at room temperature. For the first time, such polariton condensates are integrated into a generative adversarial network (GAN) as a physically implemented stochastic transformation layer, leveraging their light–matter coupling–induced many-body nonlinearity and inherent quantum fluctuations. The spatial correlations of the condensate output naturally regularize the generator, substantially enhancing training stability and sample diversity. Experimental results demonstrate that the proposed Polariton GAN outperforms conventional digital perturbation methods and standard laser-based systems, achieving significant improvements in Inception Score, numerical fidelity, and structural similarity metrics.

Abstract

Generative modelling requires efficient stochastic nonlinear transformations and physical platforms that can naturally realise them. We experimentally demonstrate that nonlinear optical systems operating in the strong light-matter coupling regime can serve as physical transformation layers for conditional generative modelling. Specifically, we develop a workflow in which room-temperature exciton-polariton condensates formed in organic dye microcavities act as a physical stochastic transform within a generative adversarial network and enable conditional digit-to-image translation. By using the nonlinear many-body dynamics and intrinsic stochasticity of polariton condensates, the workflow outperforms baseline approaches based on digitally injected perturbations. We find that polariton-enabled sampling via generative adversarial network (Polariton GAN) yields improved inception score, digit preservation accuracy and structural similarity compared with both digital sampling and laser-based systems. We further show that spatially correlated output variations can naturally regularise adversarial training and enhance output diversity. Our results establish polariton condensation as a new computational resource for generative modelling, opening a pathway towards physics-enhanced machine learning systems.

Problem

• generative modelling
• stochastic nonlinear transformations
• polariton condensates
• conditional generation
• physics-enhanced machine learning

Innovation

• polariton condensates
• generative adversarial network
• physical stochastic transform
• room-temperature nonlinear optics
• physics-enhanced machine learning