Achtung! Das Lehrangebot ist noch nicht vollständig und wird bis Semesterbeginn laufend ergänzt.
052112 VU Numerical High Performance Algorithms (2020W)
Prüfungsimmanente Lehrveranstaltung
Labels
An/Abmeldung
Hinweis: Ihr Anmeldezeitpunkt innerhalb der Frist hat keine Auswirkungen auf die Platzvergabe (kein "first come, first served").
- Anmeldung von Mo 14.09.2020 09:00 bis Mo 21.09.2020 09:00
- Abmeldung bis Mi 14.10.2020 23:59
Details
max. 25 Teilnehmer*innen
Sprache: Englisch
Lehrende
Termine (iCal) - nächster Termin ist mit N markiert
PLEASE NOTE: Due to the current COVID situation the course is online until further notice!
Donnerstag
01.10.
13:15 - 14:45
Digital
Dienstag
06.10.
13:15 - 14:45
Digital
Donnerstag
08.10.
13:15 - 14:45
Digital
Dienstag
13.10.
13:15 - 14:45
Digital
Donnerstag
15.10.
13:15 - 14:45
Digital
Dienstag
20.10.
13:15 - 14:45
Digital
Donnerstag
22.10.
13:15 - 14:45
Digital
Dienstag
27.10.
13:15 - 14:45
Digital
Donnerstag
29.10.
13:15 - 14:45
Digital
Dienstag
03.11.
13:15 - 14:45
Digital
Donnerstag
05.11.
13:15 - 14:45
Digital
Dienstag
10.11.
13:15 - 14:45
Digital
Donnerstag
12.11.
13:15 - 14:45
Digital
Dienstag
17.11.
13:15 - 14:45
Digital
Donnerstag
19.11.
13:15 - 14:45
Digital
Dienstag
24.11.
13:15 - 14:45
Digital
Donnerstag
26.11.
13:15 - 14:45
Digital
Dienstag
01.12.
13:15 - 14:45
Digital
Donnerstag
03.12.
13:15 - 14:45
Digital
Donnerstag
10.12.
13:15 - 14:45
Digital
Dienstag
15.12.
13:15 - 14:45
Digital
Donnerstag
17.12.
13:15 - 14:45
Digital
Donnerstag
07.01.
13:15 - 14:45
Digital
Dienstag
12.01.
13:15 - 14:45
Digital
Donnerstag
14.01.
13:15 - 14:45
Digital
Dienstag
19.01.
13:15 - 14:45
Digital
Donnerstag
21.01.
13:15 - 14:45
Digital
Dienstag
26.01.
13:15 - 14:45
Digital
Donnerstag
28.01.
13:15 - 14:45
Digital
Information
Ziele, Inhalte und Methode der Lehrveranstaltung
PLEASE NOTE: Due to the current COVID situation the course is online until further notice!Know and understand selected advanced numerical high performance algorithms (including divide-and-conquer eigensolver, GMRES, least squares solver, QR algorithm, communication-avoiding linear solver, etc.) for large and very large problems. Understand the interdependencies between problem data, algorithm, implementation of the algorithm, hardware, performance and accuracy. Understand basic techniques for analysis, implementation and optimization of numerical high performance algorithms. Implement and evaluate your own implementations.
Art der Leistungskontrolle und erlaubte Hilfsmittel
Two homework exercises (with theoretical and practical components - implementation, experimentation, analysis), presentation of assigned papers from the literature, and an individual semester project (involving literature research, implementation, experimentation, analysis), whose results have to be presented in class and documented in written form (project report, presentation slides) during the semester.
Mindestanforderungen und Beurteilungsmaßstab
The maximum possible score is 100 points (20 for the homework exercises, 30 for the paper presentations, 25 for the presentation of the semester project, 25 for the report of the semester project). At least 50 points are required for passing the course. For passing the course, in each component (homeworks, paper presentations, semester project) at least half of the available points have to be achieved.
Prüfungsstoff
There is no separate exam, grading takes into account discussions and questions for each component (homeworks, paper presentations, semester project).
Literatur
Slides presented in class, literature references given on the slides.J. Demmel: Applied Numerical Linear Algebra
L. N. Trefethen and D, Bau, III: Numerical Linear Algebra
Golub & van Loan: Matrix Computations
L. N. Trefethen and D, Bau, III: Numerical Linear Algebra
Golub & van Loan: Matrix Computations
Zuordnung im Vorlesungsverzeichnis
Module: HPA APS
Letzte Änderung: Do 08.10.2020 11:28