Efficient lattice field theory simulation using adaptive normalizing flow on a resistive memory-based neural differential equation solver

2025-09-16

AI Summary

To address computational bottlenecks—high cost, poor parallelism, and low hardware energy efficiency—in high-dimensional lattice field theory simulations with conventional sampling algorithms (e.g., Hamiltonian Monte Carlo, HMC), this work proposes a software–hardware co-design framework. Software-wise, it employs adaptive normalizing flows (ANFs) augmented with low-rank adaptation (LoRA) to enable rapid cross-parameter fine-tuning. Hardware-wise, it introduces an in-memory computing architecture based on resistive memory devices, integrated with a hybrid analog–digital neural ordinary differential equation (Neural ODE) solver for efficient generation of statistically independent configurations. Evaluated on scalar ϕ⁴ theory and an effective field theory of graphene nanoribbons, the framework reduces integrated autocorrelation time by 8.2–13.9× versus HMC, incurs <8% parameter update overhead, achieves 16.1–17.0× speedup over state-of-the-art GPU implementations, and improves energy efficiency by 73.7–138.0×.

Technology Category

• Machine Learning: Hardware-aware ML
• Search and Optimization: Sampling/Simulation-based Search
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Lattice field theory (LFT) simulations underpin advances in classical statistical mechanics and quantum field theory, providing a unified computational framework across particle, nuclear, and condensed matter physics. However, the application of these methods to high-dimensional systems remains severely constrained by several challenges, including the prohibitive computational cost and limited parallelizability of conventional sampling algorithms such as hybrid Monte Carlo (HMC), the substantial training expense associated with traditional normalizing flow models, and the inherent energy inefficiency of digital hardware architectures. Here, we introduce a software-hardware co-design that integrates an adaptive normalizing flow (ANF) model with a resistive memory-based neural differential equation solver, enabling efficient generation of LFT configurations. Software-wise, ANF enables efficient parallel generation of statistically independent configurations, thereby reducing computational costs, while low-rank adaptation (LoRA) allows cost-effective fine-tuning across diverse simulation parameters. Hardware-wise, in-memory computing with resistive memory substantially enhances both parallelism and energy efficiency. We validate our approach on the scalar phi4 theory and the effective field theory of graphene wires, using a hybrid analog-digital neural differential equation solver equipped with a 180 nm resistive memory in-memory computing macro. Our co-design enables low-cost computation, achieving approximately 8.2-fold and 13.9-fold reductions in integrated autocorrelation time over HMC, while requiring fine-tuning of less than 8% of the weights via LoRA. Compared to state-of-the-art GPUs, our co-design achieves up to approximately 16.1- and 17.0-fold speedups for the two tasks, as well as 73.7- and 138.0-fold improvements in energy efficiency.

Problem

• Reducing computational costs in lattice field theory simulations
• Overcoming limited parallelizability of conventional sampling algorithms
• Addressing energy inefficiency in digital hardware architectures

Innovation

• Adaptive normalizing flow for parallel configuration generation
• Resistive memory-based neural differential equation solver
• Low-rank adaptation for efficient parameter fine-tuning