Simulation is a general technique of using computer-generated models for counter-factual analysis (i.e., answering what-if questions) of complex real-world systems. In other words, our focus will be on data-driven engineering - going beyond learning from data (‘data-science’) to using data to design and improve complex systems.
There are many different simulation techniques across different fields. Our focus will be on stochastic simulation; in particular, we will cover two topics:
Stochastic simulation deals with predicting certain aspects of the behavior of some complex system through approximate models. Manufacturers use simulation to model factory staffing, conveyors, automated guided vehicles, storage and retrieval systems. Airlines and transportation companies use simulation to model fleet logistics and traffic. Designers of communications networks and computer systems use simulation to model data transmission and processing. Health care providers use simulation to model medicine and staff levels and placement in hospitals and pharmacies. Epidemiologists use simulation to model spread of diseases. The defense community uses simulation to model aircraft readiness and combat strategy. In public services, simulation is used to model police, fire fighting, ambulance and judicial systems. Many aspects of financial, marketing and information systems can be studied using simulation. information systems can be studied using simulation.
Textbook: Simulation by Ross; This is the suggested textbook for the course. An e-copy is available online on the Cornell library website.
Other References: The following books (available online on the Cornell library website) are good references for the material we will cover.
The assignments in the first half of semester can be done in any high-level language; our recommendation is for using Python (in particular, we will use Python 3.6 for our solutions) and Jupyter notebooks. These are installed in the labs in 571 Rhodes Hall and 453 Rhodes Hall.
In the second half of semester, we will work with a commercial simulation package, Simio, which only operates on PCs, or on Macs running Windows. Simio is available in 571 Rhodes Hall and 453 Rhodes Hall. Students may consider obtaining the student version of Simio ($25, available here) - this imposes limits on the size of the models, but should be sufficient for the homework or project. The Simio software includes documentation.
Familiarity with the topics covered in ENGRD 2700 and ORIE 3500⁄5500 is required, but ORIE 3500⁄5500 may be taken concurrently. The initial part of the course includes a short review of probability and statistics; if this material is unfamiliar to you, then you should study Chapters 1 through 6 of Introduction to Probability and Statistics for Engineers and Scientists, 2nd ed., by Sheldon Ross, or Chapters 1-5 and Chapter 7 of Probability and Statistics for Engineering and the Sciences, 8th ed., by Devore. This material is standard, and can be found in other books at a similar level.
The course involves some coding, and some prior programming experience is useful. The programming in the first part of the course can be done using any high-level language of your choice (in particular, Python, MATLAB, R, Julia, C++ or Java); our preference is that students use Python, and submit iPython notebooks with annotated code and plots. There will be a recitation section introducing these for interested students, and students can use them in the lab computers in 571 Rhodes Hall and 453 Rhodes Hall. The second half of the course will be based on Simio, a commercial simulation package, which we will teach in class.