LLM Relational Event Models

Fit Cox proportional hazards relational event models (REMs), including a separable formulation that partitions events into initiation and continuation sub-models. Optionally augments REM simulations with large language model (LLM) agents that select targets conditioned on event history, supporting multiple providers ('OpenAI', 'Anthropic', 'xAI'/'Grok', 'Google Gemini', 'Ollama', 'AWS Bedrock') through a common interface. See Butts (2008) for description of relational event modeling.