This page is the home of the 2014–2015 installment of the Laboratory
Class Scientific Computing.
Please register through Osiris Online. Course code: WISM454.
Instructor | : | Gerard Sleijpen, Room FG k504, E-mail: G.L.G.Sleijpen@uu.nl |
Tristan van Leeuwen, Room FG k609, E-mail: T.vanLeeuwen@uu.nl | ||
Schedule | : | Weekly computer sessions, in periods 1 and 2 (wk. 37 [2014] – 51 [2014]). |
Mondays 9:00–13:00. First session Monday, September 8, 2014 | ||
Note:The discussion on Monday, September 8, 2014 (9:00-9:15) did not lead to another timeslot: the lectures will be on Mondays from 9:00-12:45 (as announced in OSIRIS). Due to the large number of participants, we might move to another Lecture Room (to be announced at this site). | ||
Location | : | Freudenthal Gebouw 514 (old name Wiskunde Gebouw, Math. Building). |
Credits | : | 7.5 ECTS |
Grading | : | Grade = ((average assignment grade) + 4 x (average report grade))/5 |
We use lecture notes, available from the Beta student desk on the first floor of the Buys-Ballot building.
The aim of the course is to learn about a number of important scientific computing subjects, and to meet with the various aspects of scientific computing. You will (partially) write your own computer code and run simulations, handle output data and visualise it. The theory and results of the simulations are to be presented by means of written reports.
This is a hands on course which is intended to expose you to computer techniques used in scientific computation. It covers the full process from the theory to programming (C++), experimenting (MATLAB), and finally to writing a report. In this year's course we will focus our attention on two subjects:
We will cover topics such as how to choose a good random number generator and how to apply techniques such as biased sampling and discrepancy sampling. Since Monte Carlo does not suffer from the 'curse of dimensionality' we can fairly efficiently approximate the volume of, say, the d-dimensional unit sphere, for large d.
In genetic algorithms we use a population that represents elements in the search space of an optimisation problem. Techniques such as elitism and Gray codes improve the optimising effects of mutation and crossover. Combined with local search, the resulting hybrid method yields a powerful optimisation program, typically for hard (NP-hard) problems.
For each of the two main subjects in this course, you will have to present your results in a report written in English and formatted using LaTeX. Look here for deadlines and additional demands. Read the Introduction and Appendix A in the lecture notes for further instructions.