Optimized Three-Dimensional Photovoltaic Structures with LLM guided Tree Search

2026-05-15

AI Summary

This study addresses the limited energy harvesting efficiency of conventional planar photovoltaic panels in mid-latitude regions, which arises from suboptimal solar incidence angles throughout the day. To overcome this challenge, the authors propose an AI-driven approach that integrates encoded agents with a large language model (LLM)-guided ERA tree search algorithm to autonomously generate high-performance three-dimensional photovoltaic structures. By introducing a novel paradigm of iterative refinement under physics-engine constraints, the method effectively mitigates reward hacking while leveraging optical solvers and physical constraint optimization. This enables the design of diverse self-shading-free 3D configurations capable of zenith-tracking. Experimental results demonstrate that the generated structures achieve significantly higher energy density compared to traditional flat panels under full-day varying solar angles.

Abstract

We present a case study for how AI coding systems can be used to generate novel scientific hypotheses. We combine a generic coding agent (Google's AntiGravity) with an LLM-driven tree search algorithm (Empirical Research Assistance / ERA) to autonomously generate high-efficiency three-dimensional photovoltaic (3DPV) structures that overcome losses limiting flat solar panels at mid-latitudes. These structures operate by presenting favorable angles to the sun throughout the day, and for illustrative purposes we focus on optimizing performance for a single solar day. Our workflow begins by using AntiGravity to reproduce calculations \cite{bernardi2012solar} showing that 3DPV can have energy densities much higher than stationary flat PV panels. We use these initial designs as the starting point for large scale tree search, where we seek improved solutions and score them for their diurnal yield. The initial tree search leads to nominally more efficient solutions, yet they are caused by algorithmic reward hacking, arising from non-physical design features such as structurally levitating disconnected tiers and exploitations of the discretizations in the optics solver. To counteract this, we develop a workflow where the coding agent iteratively patches the physics engine with constraints to eliminate reward hacking. With reward-hacking eliminated, ERA discovers a series of designs with various constraints and improved performance, including optimal designs with different fixed collector areas, optimizing zenith tracking and avoiding self shadowing. Combining coding agents with tree search (ERA) provides a powerful platform for scientific discovery, for problems whose solutions can be empirically evaluated with a score function.

Problem

• three-dimensional photovoltaic
• solar energy efficiency
• diurnal yield
• self-shadowing
• zenith tracking

Innovation

• LLM-guided tree search
• 3D photovoltaics
• reward hacking mitigation
• AI-driven scientific discovery
• physics-constrained optimization