"data for MLLM-IA"

"data for manuscript \"MLLM-IA: Harnessing Multimodal Large Language Model-based Inductive Agent for Governing Equation Discovery of Dynamical Systems\"This dataset provides a comprehensive, multi-tiered benchmark designed for evaluating Symbolic Regression (SR) algorithms and AI models in the discovery of scientific equations. It comprises four key components that span varying levels of mathematical and observational complexity. First, it includes data trajectories and initial conditions for 63 standard dynamical systems derived from ODEBench. Second, it features 100 simulated chemical reaction kinetics equations based on organic catalytic principles, which challenge models with higher-dimensional ODEs (4 to 6 variables) and strong interdependencies. Third, the dataset contains trajectories from complex nonlinear oscillator systems common in physics and engineering, governed by nonlinear force equations. Finally, to test capabilities in real-world, partially observed scenarios, the dataset includes experimental temporal concentration profiles from two actual catalytic organic reactions ([2+2]-cycloaddition and C\u2013H amination). Bridging the gap between simulated physics environments and empirical chemical experiments, this dataset serves as a robust testbed for advancing data-driven scientific discovery."