TU Wien:Fixed-Parameter Algorithms and Complexity VU (Ganian)

Aus VoWi
Zur Navigation springen Zur Suche springen

Daten[Bearbeiten | Quelltext bearbeiten]

Vortragende Robert Ganian
ECTS 4,5
Letzte Abhaltung 2023W
Sprache English
Mattermost fixed-parameter-algorithms-and-complexityRegisterMattermost-Infos
Links tiss:192135, tiss:186855
Masterstudium Data Science Modul BDHPC/EX - Big Data and High Performance Computing - Extension
Masterstudium Logic and Computation Modul Algorithmics and Complexity (Gebundenes Wahlfach)
Masterstudium Software Engineering & Internet Computing Modul Algorithmik (Gebundenes Wahlfach)
Masterstudium Technische Informatik Modul Algorithms and Programming (Gebundenes Wahlfach)

Inhalt[Bearbeiten | Quelltext bearbeiten]

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.

Ablauf[Bearbeiten | Quelltext bearbeiten]

WS17: Blocked on 6 days in January. 4.5 blocks are lectures, the last block is for student presentations.

Benötigte/Empfehlenswerte Vorkenntnisse[Bearbeiten | Quelltext bearbeiten]

The FPT part from Algorithmics VU is helpful but not required.

Vortrag[Bearbeiten | Quelltext bearbeiten]

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.

Übungen[Bearbeiten | Quelltext bearbeiten]

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.

Prüfung, Benotung[Bearbeiten | Quelltext bearbeiten]

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[Bearbeiten | Quelltext bearbeiten]

noch offen

Zeitaufwand[Bearbeiten | Quelltext bearbeiten]

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.

Unterlagen[Bearbeiten | Quelltext bearbeiten]

noch offen

Tipps[Bearbeiten | Quelltext bearbeiten]

  • If you enjoyed Robert Ganians part of the Algorithmics VU, then this course is definitely worth doing.

Verbesserungsvorschläge / Kritik[Bearbeiten | Quelltext bearbeiten]

noch offen


Diese Seite hat noch keine Anhänge, du kannst aber neue hinzufügen.