Semester.ly

Johns Hopkins University | EN.550.662

Optimization Algorithms

3.0

credits

Average Course Rating

(4.5)

This course considers algorithms for solving various nonlinear constrained optimization problems and, in parallel, develops the supporting theory. Topics include: necessary and sufficient optimality conditions for constrained optimization; projected-gradient and two-phase accelerated subspace methods for bound-constrained optimization; simplex, interior-point, Bender's decomposition, and the Dantzig-Wolfe decomposition methods for linear programming; duality theory; penalty, augmented Lagrangian, sequential quadratic programming, and interior-point methods for general nonlinear programming. In addition, we will consider the Alternating Direction Method of Multipliers (ADMM), which is applicable to a huge range of problems including sparse inverse covariance estimation, consensus, and compressed sensing.

Spring 2013

(4.64)

Spring 2014

(4.41)

Spring 2015

(4.45)

Spring 2013

Professor: Daniel Robinson

(4.64)

The best aspects of the course included the professor and supplemental learning materials. The professor’s lectures were engaging and the notes/slides made available were very helpful to students. The worst aspect of the course was the fast-paced lectures during the second half of the semester. The course would improve if there was a better supporting textbook for the class and more practice examples of some of the problems found in the homework assignments. Prospective students should ensure they attend al the lectures and spend time practicing the material.

Spring 2014

Professor: Daniel Robinson

(4.41)

The instructor’s detailed notes, clear lectures, and methods of introducing complicated concepts were given near unanimous thumbs up by the students. Many students commented that Dr. Robinson’s approach to teaching difficult concepts was systematic and intuitive. However, some students thought class moved too quickly, and would have liked more practical applications for the algorithms. Some suggestions for improving the course included adding a discussion section where example implementations are carried out, more coding homework, and having take-home exams. Prospective students should have some background in optimization, linear algebra, and be familiar with MATLAB prior to starting this course.

Spring 2015

Professor: Daniel Robinson

(4.45)

The best aspects of this course included the wel -prepared lectures, the good balance of theory and application, and the comprehensive lecture notes circulated prior to lecture meetings. Students found that the professor was highly knowledgeable and well organized, but that he occasionally moved too quickly through slides, particularly with more difficult material. A few suggested that he had the tendency to come off as condescending to students. Suggestions for improvement included more frequent and detailed feedback on assignments and tests. Prospective students should have good working knowledge of algebra and matrix, and a background with optimization will be beneficial.