📄 中文摘要
自动化化学配方设计是材料科学的基石，但在高维组合空间中进行设计时面临许多挑战，包括离散组成选择和连续几何约束。现有的大型语言模型（LLM）代理在长时间推理和路径依赖探索中存在显著局限，可能导致模式崩溃。为了解决这些问题，提出了AI4S-SDS，一个闭环神经符号框架，集成了多智能体协作与定制的蒙特卡罗树搜索（MCTS）引擎。该系统引入了一种稀疏状态存储机制和动态路径重构方法，能够将推理历史与上下文解耦，从而提高了设计过程的效率和准确性。
AI4S-SDS：一种通过稀疏蒙特卡罗树搜索和可微物理对齐的神经符号溶剂设计系统
📄 English Summary
AI4S-SDS: A Neuro-Symbolic Solvent Design System via Sparse MCTS and Differentiable Physics Alignment
The automated design of chemical formulations is a cornerstone of materials science, yet it faces significant challenges in navigating a high-dimensional combinatorial space involving discrete compositional choices and continuous geometric constraints. Existing Large Language Model (LLM) agents struggle with limitations in context windows during long-horizon reasoning and path-dependent exploration, which can lead to mode collapse. To address these issues, AI4S-SDS is introduced as a closed-loop neuro-symbolic framework that integrates multi-agent collaboration with a tailored Monte Carlo Tree Search (MCTS) engine. A Sparse State Storage mechanism with Dynamic Path Reconstruction is proposed, decoupling reasoning history from context, thereby enhancing the efficiency and accuracy of the design process.