TENTATIVE SYLLABUS AND COURSE OVERVIEW

  (UPDATED THROUGHOUT SEMESTER)

ECE 619

Spring 2018

Applications of Constrained Optimization

Instructor: Kevin Tomsovic

email: tomsovic at utk.edu


Assignments:

Lectures/Schedule:

Suggested References:

Overview

This course will cover application of optimization and some computational intelligent system methods to engineering problems. We will begin with mathematical programming techniques and the formulation practical application of these methods. We then begin loosening up the restrictions of the linear formulation to consider a broad class of problems both convex and non-convex. As time permits, computational intelligence approaches as represented by stochastic optimization, neural networks and other pattern matching approaches (sometimes referred to as soft computation methods) will then be contrasted with these more traditional approaches. The recommended background for this course is primarily a solid foundation in linear algebra. There will be some minor programming required in Matlab, so it will be desirable to be familiar with Matlab. Most of the examples will be chosen from applications to power systems but the course will be taught in a general way and no specific background in power systems is necessary.

The following list of topics is way too long to cover in a semester so I will pick and choose among the latter topics. Students with a preference for certain topics should feel to make requests.

I. Fundamentals of Constrained Optimization


Linear Formulation


Integer and Logical Constraints


Network Problems


Lagrangian Duality


Gradient and gradient derived methods


Formulating meaning and useful optimization problems



Summary of first part of course

Midterm Exam - March 22 Solutions



II. Optimal use of data - curve fitting, model identification and classification





III.  Optimization for Difficult Domains






Grading

Homework assignments (approximately weekly) - 50%
Midterm exam - 25%
Project  - 25%