STAT 8054: Advanced Statistical Computing
Instructor. Aaron J. Molstad (amolstad@umn.edu)
Office hours. Monday 3:00 - 4:00PM and Wednesday 9:00 - 10:00AM in Ford Hall 384
Syllabus. [pdf]
Lecture. Monday, Wednesday, Friday at 12:20 - 1:10PM in Ford Hall 170
Note that you must be logged into your UMN Canvas account to access course materials.
| Lecture | Topics | |||
| 1.1 (1/22) | Course overview [slides] | |||
| 1.2 (1/24) | Fundamentals of numerical linear algebra [slides] | |||
| 1.3 (1/27) | Computational complexity, matrix decompositions [slides] |
Lecture 1 [notes]
| 2.1 (1/29) | Unconstrained optimization overview [slides] | |||
| 2.2 (1/31) | Optimality conditions, convexity [slides] | |||
| 2.3 (2/3) | Quasiconvexity, strong convexity, L-smoothness [slides |
Lecture 2 [notes]
| 3.1 (2/5) | Steepest descent, gradient descent [slides] | |||
| 3.2 (2/7) | Example, accelerated gradient descent [slides] | |||
| 3.3 (2/10) | Newton’s method [slides] |
Lecture 3 [notes]
| 4.1 (2/12) | Majorize-minimize principle [slides] | |||
| 4.2 (2/14) | Expectation-maximization algorithm [slides] | |||
| 4.3 (2/17) | Proximal gradient descent [slides] | |||
| 4.4 (2/19) | Proximal gradient descent [slides] | |||
| 4.5 (2/21) | Generalized gradient descent [slides] | |||
| 4.6 (2/24) | Stochastic gradient descent [slides] |
Lecture 4 [notes]
| 5.1 (2/26) | Constrained optimization, KKT conditions [slides] | |||
| 5.2 (2/28) | Duality [slides] | |||
| 5.3 (3/3) | Barrier method [slides] | |||
| 5.4 (3/7) | Quadratic penalty method [slides] | |||
| 5.5 (3/17) | ADMM [slides] | |||
| 5.6 (3/19) | ADMM continued [slides] | |||
| 5.7 (3/21) | Distance-to-set penalty method [slides] |
Lecture 5 [partial notes]
| 6.1 (3/24) | Inversion method, rejection sampling [slides] | |||
| 6.2 (3/26) | Rejection sampling continued [slides] | |||
| 6.3 (3/28) | Importance sampling [slides] |