LLM agents benefit substantially from domain context when analyzing time-series observations from dynamical systems

Our results show that agents benefit substantially from domain context

http://arxiv.org/abs/2608.27182v1