## MAE 598/494 Design Optimization | Fall 2017

**Faculty**: Max Yi Ren

**Email**: yiren@asu.edu

**Office**: GWC464

**Location & Time**: ECAA219, TU/TH 4:30pm-5:45pm

**Office hours**: TU/TH 2:00pm-4:00pm in GWC464, or by appointment

**Prerequisites**: No previous knowledge about optimization is expected or required.
However, solid background in calculus and linear algebra is assumed ([Math review][mathreview]).
Students are also expected to have basic programming skills (Matlab and Python preferred).

### Course Overview

The purpose of this course is to introduce the student to mathematical modeling, optimization theory, and computational methods for analytical and simulation-based optimal system design. The student will learn to (i) develop proper mathematical models to formulate design optimization problems and (ii) develop and apply appropriate optimization algorithms to solve them.

### Workload and grading

There are 5 homework assignments, 2 mid-term exams, and an individual project.

- Homework (8%*5)
- Project report (30%)
- Exams (15%*2)

### Course schedule

Date | Topic | Assignment | Note |
---|---|---|---|

Aug 17 (TH) | Introduction to optimization | HW 1 in | |

Aug 22 (TU) | Math review 1: Linear algebra and matrix calculus | ||

Aug 24 (TH) | Unconstrained optimization 1: Convexity and optimality conditions | HW 1 due, HW 2 in | |

Aug 29 (TU) | Unconstrained optimization 2: Gradient descent and convergence | ||

Aug 31 (TH) | Unconstrained optimization 3: Modern line search techniques | ||

Sep 5 (TU) | Unconstrained optimization 4: Newton’s method | ||

Sep 7 (TH) | Unconstrained optimization 5: Application - Origami design | HW 2 due | |

Sep 12 (TU) | Unconstrained optimization 6: Trust Region | ||

Sep 14 (TH) | Mid-term 1 |
||

Sep 19 (TU) | Math review 2: Random variables and probability distributions | HW 3 in | |

Sep 21 (TH) | Metamodeling 1: Ordinary least square, Regularization, Information criteria | Max out of town | |

Sep 26 (TU) | Metamodeling 2: Design of experiments | ||

Sep 28 (TH) | Metamodeling 3: Neural networks | ||

Oct 3 (TU) | Metamodeling 4: Gaussian process | ||

Oct 5 (TH) | Metamodeling 5: Application - Efficient global optimization | HW 3 due | |

Oct 10 (TU) | Full break | ||

Oct 12 (TH) | Constrained optimization 1: Reduced gradient | HW 4 in | |

Oct 17 (TU) | Constrained optimization 2: KKT conditions | ||

Oct 19 (TH) | Constrained optimization 3: KKT geometry, sensitivity analysis | ||

Oct 24 (TU) | Constrained optimization 4: Generalized reduced gradient | ||

Oct 26 (TH) | Constrained optimization 5: Application - Topology optimization | ||

Oct 31 (TU) | Constrained optimization 7: Quasi-Newton methods | HW 4 due, HW 5 in | |

Nov 2 (TH) | Constrained optimization 8: Active set strategy | ||

Nov 7 (TU) | Constrained optimization 9: Sequential Quadratic Programming | ||

Nov 9 (TH) | Mid-term 2 |
||

Nov 14 (TU) | Constrained optimization 10: Duality | HW 5 due | |

Nov 16 (TH) | Math review 3: Markov decision process | ||

Nov 21 (TU) | Optimal control 1: Model-free methods | ||

Nov 23 (TH) | Thanksgiving | ||

Nov 28 (TU) | Optimal control 2: Model-based methods | ||

Nov 30 (TH) | Optimal control 3: Evolutionary strategies |

### Project

The course project for this year will be on robot topology, sensor, and controller design through evolution and policy search. Details TBD.

### Academic Integrity

Each student has an obligation to act with honesty and integrity, and to respect the rights of others in carrying out all academic assignments. MAE 540 will follow the process defined by the Office of the Dean of Students, which states that any student who is found to have violated the Student Code of Conduct will, at a minimum, receive an E in the course. The College Policy defines the process to be used if the student wishes to appeal this action.