Generative thermodynamic computing

2025-06-18

AI Summary

Conventional generative models rely on neural-network-based denoising and artificial noise injection, entailing high computational and thermodynamic costs. Method: This paper proposes a thermodynamically grounded generative computing framework based on Langevin dynamics, directly embedding generative modeling into the natural time evolution of physical systems. By maximizing the probability of reverse-noise trajectories under thermodynamic constraints, it enables low-energy data synthesis without parametric networks or active external control. Contribution/Results: The key innovation is the first formalization of generative modeling as a thermodynamically reversible evolution problem—bypassing iterative denoising and complex network architectures inherent to diffusion models. Numerical simulations demonstrate substantial reduction in thermodynamic dissipation. The framework establishes both theoretical foundations and a practical pathway toward physics-native generative models implemented on analog hardware, operating without artificial noise injection or external control.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation
• Computer Vision: Generative Adversarial Networks (GANs) for Vision

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

We introduce a generative modeling framework for thermodynamic computing, in which structured data is synthesized from noise by the natural time evolution of a physical system governed by Langevin dynamics. While conventional diffusion models use neural networks to perform denoising, here the information needed to generate structure from noise is encoded by the dynamics of a thermodynamic system. Training proceeds by maximizing the probability with which the computer generates the reverse of a noising trajectory, which ensures that the computer generates data with minimal heat emission. We demonstrate this framework within a digital simulation of a thermodynamic computer. If realized in analog hardware, such a system would function as a generative model that produces structured samples without the need for artificially-injected noise or active control of denoising.

Problem

• Generating structured data from noise using thermodynamic dynamics
• Training system to reverse noising with minimal heat emission
• Demonstrating framework in digital simulation for analog hardware

Innovation

• Generative modeling via thermodynamic system dynamics
• Training by reversing noising trajectory probability
• Analog hardware for noise-free structured sampling