Phythesis: Physics-Guided Evolutionary Scene Synthesis for Energy-Efficient Data Center Design via LLMs

2025-12-11

AI Summary

Traditional data center design methodologies struggle to cope with escalating system complexity, while existing generative AI approaches neglect physical constraints and fail to satisfy quantifiable operational objectives—such as Power Usage Effectiveness (PUE) and thermal uniformity—as well as stringent engineering requirements. To address this gap, we propose the first large language model (LLM)-enhanced, physics-constrained evolutionary optimization framework. It features a two-level iterative architecture: an LLM generates initial topologies, while a CFD surrogate model and a physics-consistency self-critique mechanism jointly validate and refine layouts in closed-loop. Crucially, we embed quantifiable thermal and energy constraints directly into the generative design pipeline for the first time, enabling automatic synthesis of simulation-ready 3D layouts. Evaluated across three scale scenarios, our method achieves a 57.3% improvement in layout generation success rate and reduces PUE by 11.5%, significantly outperforming pure-LLM baselines.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Machine Learning: Large Multimodal Models (LMMs)
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Data center (DC) infrastructure serves as the backbone to support the escalating demand for computing capacity. Traditional design methodologies that blend human expertise with specialized simulation tools scale poorly with the increasing system complexity. Recent studies adopt generative artificial intelligence to design plausible human-centric indoor layouts. However, they do not consider the underlying physics, making them unsuitable for the DC design that sets quantifiable operational objectives and strict physical constraints. To bridge the gap, we propose Phythesis, a novel framework that synergizes large language models (LLMs) and physics-guided evolutionary optimization to automate simulation-ready (SimReady) scene synthesis for energy-efficient DC design. Phythesis employs an iterative bi-level optimization architecture, where (i) the LLM-driven optimization level generates physically plausible three-dimensional layouts and self-criticizes them to refine the scene topology, and (ii) the physics-informed optimization level identifies the optimal asset parameters and selects the best asset combination. Experiments on three generation scales show that Phythesis achieves 57.3% generation success rate increase and 11.5% power usage effectiveness (PUE) improvement, compared with the vanilla LLM-based solution.

Problem

• Automates energy-efficient data center design with physics constraints
• Integrates large language models with evolutionary optimization for layouts
• Improves generation success and power efficiency over AI-only methods

Innovation

• LLMs generate 3D layouts and self-criticize topology
• Physics-guided evolutionary optimization selects optimal asset parameters
• Bi-level optimization automates simulation-ready scene synthesis