Agent-Based Modeling and Digital Twins


Units: 6


No pre-requisites

In the rapidly evolving landscape of technology and data science, the ability to model and simulate complex systems is paramount. The growth of Large Language Models is also resulting in generative agents that combine the capabilities of LLMs with reasoning, planning, and other traditional AI areas to further advance the discipline of single-agent and multi-agent systems.

This course on Agent-based Modeling and Digital Twins offers a comprehensive deep dive into the world of system thinking, enabling students to discern when and how to deploy agent-based simulations and digital twins as effective solutions. Starting from traditional single-agent and multi-agent models we will explore digital twins and the emerging area of generative agents. Drawing from real-world scientific, social, and business challenges, participants will master the art of designing, constructing, and simulating intricate agent-based models. They will also gain proficiency in bridging the gap between the physical and digital worlds, creating digital twins that can be calibrated, monitored, and utilized to evaluate policy decisions, model behavioral interventions, and understand human behaviors.

Beyond foundational knowledge, this course extends its reach to practical applications, guiding students in harnessing the power of agent-based models across diverse sectors. From manufacturing to healthcare, automotive to aviation, and utilities to agriculture, participants will discover the transformative potential of digital twins. Students will develop specific use cases of digital twins in these industry sectors. As we delve deeper, the course will also explore the cutting-edge realm of generative agents, focusing on those employing LLMs. These agents, capable of logical reasoning, planning, optimization, and executing goal-driven tasks, represent the future of agent-based modeling.

To ensure a holistic learning experience, the course provides hands-on exposure to a suite of tools, both open-source and proprietary. Students will familiarize themselves with some of the platforms like Netlogo, Meson, AnyLogic and Microsoft’s AutoGen. Industry scenarios and guest lectures will bring the real-world application of these technologies to the students. By the end of this journey, students will be equipped with the knowledge and skills to lead innovations in agent-based modeling and digital twin technologies, ready to make impactful contributions to their respective fields.

Learning Outcomes

Upon completion of this course, students will be able to:

  • System Thinking & Problem Solving: Apply system thinking principles to discern when agent-based simulation and digital twins are the appropriate solutions.
  • Complex Problem Analysis: Tackle intricate scientific, social, and business challenges by designing, constructing, and simulating advanced agent-based models and digital twins.
  • Design, Development, and Evaluation: Seamlessly integrate the tangible and digital realms by designing, building, calibrating, and monitoring digital twins.  Such calibrated and embedded models can be used evaluate policy decisions, model behavioral interventions, and learn human behaviors.
  • Applications: Utilize agent-based models and digital twins across a diverse range of scientific, engineering, and business sectors including manufacturing, healthcare, and life sciences, automotive and aviation, utilities, mining, and agriculture.
  • Emerging Trends: Delve into the forefront of generative agents, particularly those employing LLMs, to develop agents that can perform logical reasoning, planning, optimization, and other goal-driven tasks.
  • Frameworks and tools: Gain hands-on exposure to open-source tools like Netlogo, Mesa, Meson, AnyLogic, etc.

Prerequisites Description

Basic understanding of statistics, machine learning, and programming is useful. The course will be offered with both a programming option and a non-programming option.