The last homework will be due on Apr 28th. It may be downloaded here.
Available presentation dates
May 3rd — TAKEN
Apr 28 — TAKEN
Apr 26 — TAKEN
Apr 21 — TAKEN
Homework 2 solutions
Solutions to homework 2 may be downloaded here.
Homework 4
Homework 4 is available here.
Final project
For your final project, please find another student from the class to work with and pick a paper from the list below. Email me to let me know what your group is and what paper you have chosen. Papers will be assigned to groups on a first come first serve basis.
Please prepare a presentation about the paper which will summarize the main idea in a way that is understandable all the students in the class. Your presentation should be 35 minutes, leaving 5 minutes for questions — though I will interrupt you and ask questions in the middle of the presentation.
Keep in mind that to understand the paper you chose, you may need to read other papers — usually you need to understand the previous work before you can begin to read recent progress.
In addition, please prepare a short report — up to five pages — summarizing the main idea of the paper. Your report should not copy and paste things from the paper. Rather, it should contain something additional relative to the paper — ideally, a clear and short exposition of the main idea in a special case which avoids most of the technicalities, though any explanation of what the main techniques of the paper are really about is acceptable.
The report will be due during finals week, and your presentations will be in the last 1-2 weeks of class.
List of papers:
- Efficiency loss in a network resource allocation game
- Convex relaxations of optimal power flow, part I — TAKEN
- Convex relaxations of optimal power flow, part II
- Robust distributed routing in flow networks, part I
- Robust distributed routing in flow networks, part II
- A geometric alternative to Nesterov’s accelerated gradient descent — TAKEN
- On the convergence of alternating minimization — TAKEN
- Tight bounds for Douglas-Rachford splitting
- Chance constrained optimal power flow
- Optimal demand response based on utility maximization – TAKEN
- Power systems dynamics as a primal dual algorithm
- Stability of stochastic gradient descent — TAKEN
- Gradient descent converges to the minimizers
- Robustness of marginal cost taxes
- Convexity and robustness of dynamic traffic assignment
- Stochastic gradient descent, weighted sampling, and randomized kaczmarz — TAKEN
- Proximal splitting methods in signal processing
Homework 3
Homework 3 may be downloaded here. It will be due on March 15th.
Homework 1 solutions
The solutions to homework 1 may be found here.
Homework 2
Homework 2 may be downloaded here. It will be due Tuesday Feb 23rd.
Corrected proof: stationary points of gradient descent with the Armijo rule
The corrected proof is spelled out in this document.
Office hours
I will have office hours on Thursdays 10-12 in 145 CSL.