Last Updated: February 20, 2024 - Added reading on feature importance to schedule.

Instructor

Prof. Kristofer Reyes

• [email protected]
• 134 Bell Hall
• Office Hours: Fridays, 2-3 pm

Course time and location

• 123 Clemens Hall
• Tuesdays and Thursdays, 9:30 - 10:50 am

Assessment

This course is graded on three homework sets and a final MDI interconnect project.

Textbook

There is no official textbook. Everything will be covered in lecture, homework, or additional reading materials. These materials are listed in the schedule.

Prerequisite material can be reviewed in the following textbooks, which are not required but highly recommended:

• “Introduction to Probability Models,” Sheldon Ross.
• “Introduction to Linear Algebra,” Gilbert Strang.
• “All the Math You Missed (But Need to Know for Graduate School,” Thomas Gerrity
• Python programming tutorials at “https://diveintopython.org”

Prerequisites

• MDI 504 - Multivariate Statistics and Materials Informatics
• Python Programming
• Introduction to probability theory and linear algebra.

Description

This course introduces students to the various interconnected techniques needed to model, learn, and control dynamic materials systems. We present these methods with the nominal motivation of experimental design: selecting experiments to run to build up a training data set or achieve a scientific objective. More broadly, however, we shall develop a toolbox of advanced algorithmic techniques to solve problems in dynamics, inference, sampling, control, and optimization. We will cover the topics in 6 modules:

• Module 1: Probabilistic modeling and uncertainty quantification
• Module 2: Sensitivity analysis
• Module 3: Experimental design
• Module 4: State estimation and dynamics
• Module 5: Control and Reinforcement Learning
• Module 6: Generative design

Grading and assessment

Homework sets (100 points each)

Homework will be provided as three problem sets that mix programming assignments and hand-worked problems. Homework sets are due 11:59 pm on their respective due dates, posted below. Homework submissions may be subject to recorded oral follow-up and reexamination via Zoom. In this case, assessments of student understanding derived from oral reexamination will supersede the scores assigned to written homework solutions.

The homework sets for a module will be progressively filled in as we progress through that module. It is up to you to ensure you have the most up-to-date copy of the homework.