|Prof. Dr. Sven Seuken, Prof. Dr. Peter Widmayer Dr. Paul Dütting
|Time and Location
|3 (including a mark)
|Send email for appointments.
In this seminar, we will discuss advanced topics in economics and computation (list of topics will be made available in the kick-off meeting). Students review a paper, independently acquire the necessary background knowledge, and write a ca. 10 pages manuscript. They give a presentation (20 min.) on the topic of their paper, and lead a short discussion (10 min.) following their presentation. Students support each other as "buddies".
The seminar will be held as a "Block-Seminar". The kick-off meeting will be on Wednesday, 24.02.2016.
After the kick-off meeting, students can report which topics they prefer and in which order (cf. the kick-off slides for the format), and we will then assign topics to students accordingly via the RSD mechanism. After that, students will receive their own topic and date of their talk as well as the list of all assigned topics and they have to confirm their participation in the seminar. They may also submit a preference order for the topic for which to act as the "buddy". We will then assign buddies using RSD again.
To guarantee an acceptable student/staff ratio, we have to restrict the number of participants to 15. In case more than 15 students apply, the participants will be randomly selected.
Together with their topic, students are assigned an advisor (Prof. Seuken or Prof. Widmayer or Dr. Dütting or a PhD student or PostDoc with expertise in the relevant area). Students read and understand their paper and write their manuscript, which has to be sent to the advisor and to the buddy at least four weeks before the talk (22.04.2016). About three weeks before the talk, a meeting with the advisor and the buddy is held in which the manuscript is discussed. The aim of this meeting is to test the student's understanding of the topic, clear any remaining gaps in understanding and to ensure a high quality of the talk. The advisor may suggest improvements to the manuscript, which should be implemented until the day of the talk: the final version of the manuscript is due the day of the talk.
The version of the manuscript sent to the advisor will be subject to grading, so this version should be in an essentially-final state.
The manuscript should read like a written version of your talk. It should contain everything you want to present, e.g. motivation, formal model, related literature, core theorems and interesting proofs or proof ideas. It does not have to contain every aspect of your topic in full detail (but references where to find the details are appreciated).
The value of 10 pages is by no means mandatory as the length of your manuscript. However, experience has shown that good manuscripts tend to be about this long, so if your manuscript is much shorter, you might want to re-think whether you have laid out your topic in sufficient detail. On the other hand, if your manuscript is much longer, you should check if some aspect can be described in a more compact or simplified way. Furthermore, you should think about the manuscript as a "speaker's manuscript.'' Thus, if it is much longer than 10 pages, you will not be able to cover everything in 20 minutes!
Every student supports another student as a "buddy" in understanding the material and preparing the talk. In particular, it is expected that a buddy ...
The "buddy" relation will not in general be symmetric.
You can find some useful guides (how to read a paper, how to give a talk) at this website (at "Links and resources") of a previous course taught by Tim Roughgarden at Stanford.
Fri, 20.05.2016, ETH, room CAB H52
|General Revision Protocols in Best Response Algorithms for Potential Games
|Convergence to approximate Nash equilibria in congestion games
|Learning Equilibria of Games via Payoff Queries
|Mechanism design for fair division: allocating divisible items without payments
|FEEDBACK / BUFFER
|Unbalanced Random Matching Markets: The Stark Effect of Competition
|An Analysis of One-Dimensional Schelling Segregation
|Algorithms, games, and evolution
|Sensitivity and Computational Complexity in Financial Networks
|FEEDBACK / BUFFER
Successful completion of one of the courses "Economics and Computation" or "Algorithmic Game Theory" or explicit consent from an instructor. Students who have not taken any of the two lectures, but have enough background in relevant areas (e.g., microeconomics, game theory, multi-agent systems, auction theory, or mechanism design) may also be eligible to participate, but must contact an instructor ahead of time.
Suitable for all BSc and MSc students who have successfully completed one of the courses "Economics and Computation" or "Algorithmic Game Theory", or who have obtained similar background knowledge elsewhere. Specifically recommended for students thinking about writing their BSc or MSc thesis on a topic related to Economics and Computation.
1. Final manuscript (~10 pages).
2. Oral presentation of the paper (at the blackboard or using slides) and leading a discussion.
3. Acting as a buddy/shepherd for another student
4. Active participation during the seminar.