TU Wien:Fixed-Parameter Algorithms and Complexity VU (Ganian)
|Alias||Fixed-Parameter Algorithms and Complexity (en)|
|Department||Forschungsbereich Algorithms and Complexity|
|Mattermost||Register • Mattermost-Infos•|
Algorithms, techniques and proofs from the area of fixed-parameter algorithms. Some of the proofs are rather hard, but the lecturer does a good job at explaining them and focuses on the big picture, instead of getting lost in details. Additionally, you do not need to understand the proofs from the lecture to pass the course (although it will probably help when preparing the presentation).
Topics in WS20: The complexity classes FPT, XP and the W-hierarchy. The exponential time hypothesis and techniques to prove lower bounds for kernels (cross composition and polynomial parameter transformations). Techniques to find FPT algorithms or kernels: bounded search trees, color coding, integer linear programing, iterative compression and crown decomposition. Also talked about model checking.
WS17: Blocked on 6 days in January. 4.5 blocks are lectures, the last block is for student presentations.
The FPT part from Algorithmics VU is helpful but not required.
In English. Mr Ganian is a great lecturer. He not only presents the results, but usually leads the students to them. The atmosphere is very relaxed.
One 20-25 minute presentation of a selected paper from International Symposium on Parameterized and Exact Computation (IPEC). You should present the main results and at least one non-trivial proof from the paper.
WS21: We had the choice to either do a 23-25 minutes paper presentation or to do a 10 minutes presentation and an additional writeup of one proof of the paper.
Doing the presentation gets you a "Befriedigend" (unless you do very badly, which didn't happen this semester). In order to get a better grade, you need to do an oral exam.
WS21: The atmosphere during the exam was very good. He started by asking some basic questions (example of a parameterized problem, definitions of FPT, XP, (polynomial) kernel, parameterized reductions, examples of XP and para-NP-hard problems). For the more advance topics I had the choice between Interative Compression or both ILP and Crown Reductions. I would recommend to prepare one problem for each method / complexity class. The grading was very fair. Little mistakes or "knowledge gaps" were not a big problem and he leads you to a solution if you are stuck somewhere.
Dauer der Zeugnisausstellung
Definitely within 3 ECTS. Attending the 6 lectures takes approx. 6*3=18 hours. Add to that the time required to understand a paper and prepare a presentation (15 hours?), and you have certainly passed the course. To get a better grade than 3, you will have to study for the exam, but if you pay attention during the lectures this shouldn't be too much work.
- If you enjoyed Robert Ganians part of the Algorithmics VU, then this course is definitely worth doing.
Verbesserungsvorschläge / Kritik