Beskjeder

Thanks for your efforts & stamina with the two-week Exam Project. The oral examination part will take around 30 minutes per student, and is being organised as follows. Please report immediately if there are problems related to the time slot you are allocated to here. (This particular message, with candidate names & dates etc., will be deleted after the exam.) The examinations are taking place in room B 81, 8th floor.

Fri 20.6:

14:30: Lukas Gudmundsson

Tue 24.6:

12:00 Christopher Nuth

12:45 Roar Br?nden

13:30 Bj?rnar Mortensen

14:30 Geir W?hler Gustavsen

15:15 Mads Opstad Reistadbekk

16:00 Dina Abdin Ahmed Salama

Wed 25.6:

9:00 Kjell Andreas Solberg

9:45 J?rgen Hande

10:30 Linn Saxrud Johansen

11:15 Ida Solhjell

12:45 Linn Cecilie Bergersen

Good luck, everyone: the Exam Project set is now available. I may perhaps direct your attention to the following sentence (page 1): "The full exam set is (admittedly) laborious, and candidates are allowed not to despair if they do not manage to answer all questions well."

For the Exam Project, starting Mon 2 June, note that there will be two special pages that each student need to submit with her or his report:

Page A is the erkl?ring (self-declaration form), properly signed etc.

Page B is the student's one-page summary of the exam project report, which should also contain a brief self-assessment of its quality.

We have agreed on a slightly revised time schedule for the exam project (in order to make it easier for some of the students to travel to the Vilnius conference):

(1) An exam project will be made downloadable from the course site on Mon June 2nd. It will have both practical and theoretical aspects. The students are required to work independently for the exam project.

(2) Individual written exam project reports need to be handed in (in duplicate: one for Nils, one for the external examiner) by Mon June 16 (14:45 at the latest), to the Math Dept Reception office (7th floor). The reports are then scrutinised & marked, by Nils and external examiner. Students who travel to Vilnius may give their reports to Nils directly, in Vilnius, or to the Math Dept Reception office by Fri June 13.

(3) Dates for the oral examination part are unchanged: These sessions will take place Tue and Wed, June 24 and 25. Specific details will be provided later.

I have uploaded "Exercises, Version H"; please download and print out.

For Victory Day May 8, we attempt to go through most of Exercises 23, 24, 25, 26, 27, 28, 29; please prepare.

Note com15d, just uploaded, an R script for one-dimensional image restoration using Markov modelling, cf. Exercise 22.

I am workshopping in Warwick & Cambridge next week, so there is no teaching Apr 17th. Apr 24th we start with the third & final main part of the course, namely the statistics of Extreme Values.

For Apr 24th, attempt to carry out all of the points of Exercise 22, including the parts pertaining to estimation of the theta parameter of the Markov model, (a) using PL and ML for the (in this case) known x-true image; (b) using only the y image data.

The Exam: we have apparently converged to the following scheme of things:

(1) An exam project will be made downloadable from the course site on Wed June 4th. It will have both practical and theoretical aspects. The students are required to work independently for the exam project.

(2) Individual written exam project reports need to be handed in (in duplicate: one for Nils, one for the external examiner) by Wed June 18 (14:55 at the latest). The reports are then scrutinised & marked, by Nils and external examiner.

(3) Finally, as rules require, there will be an oral examination, perhaps ca. 30 minutes per candidate. These sessions will take place Tue and Wed, June 24 and 25. Specific details will be provided later.

(4) The written exam report will be considered more important than the oral examination part. For the final evaluation mark, the written report will count about 3/4, and the oral part about 1/4.

For Apr 10th, prepare by working through Exercises 20, 21, 22. You need data from the file "markov-with-noise", and the last version (G) of the Exercises & Notes, now uploaded to the site.

We shall work one more week with Markov Random Fields etc., after which we turn to the last part of the course, pertaining to extreme value statistics.

Corrected versions of R scripts com10c and com11b are now uploaded, and are to replace earlier versions com10b and com11a.

For Apr 3rd, prepare by working through Exercises 18, 19, 20. These are included in Version E of the Exercises file, now available at this site.

I have uploaded R scripts "com10b" (contour plots for Kriging) and "com11a" (simulation of Gaussian processes, conditional on data).

For Feb 27th, we start Chapter 4, on Markov random fields, etc.

For Feb 13th, prepare Exercises 15(c), 16, 17, now available in the updated "Exercises, Version D" (now comprising 17 exercises and 18 pages).

Note: I have taken out "com3a" and replaced it with the updated version "com3b", which also separately includes the useful "squareroot" operation for matrices.

For Feb 6th I plan to summarise the remaining material of Ch 2. I shall also go through and discuss further aspects of Exercises 12, 13, 14, 15; cf. the updated "Version C" of the Exercises, now comprising 15 exercises and 15 pages.

I have uploaded R script files com3a (spatial interpolation), com4b (batmobile models), com5a (heights of Hjort brothers). Enjoy.

For Feb 21th, prepare solutions for the following practicals: (a) For the data of Exercise 10, estimate all three parameters of the multinormal data, and produce and display the associated spatial interpolator. (b) For the same data, phrase to interpolation problem in Kriging terms, i.e. minimisation of prediction squared error; again, compute and display the result, using the mean, standard deviation and correlation function as given in Exercise 10.

We shall otherwise progress further in Ch 2.

Note: an updated "version B" is now in place for the Exercises.

Feb 7th: We went through the first parts of Ch 2, and discussed Exercise 3.

For Feb 14th, two hours will be devoted to Exercises and one to progress further in Ch 2. Please prepare by working through Exercises 4, 5(e) (how tall is Professor Hjort?), 9, 10.

The lectures Thu Jan 31st are unfortunately cancelled.

An updated version of the Exercises will be uploaded to the course website soon.

As mentioned last week, the main exercise for Jan 31rd is: Go through Exercises 1 and 2 again, and supplement previous analyses with AIC scores, where AIC = -2 logLmax(model) + 2 dim(model). For example, simulate data from the AR(2) version of the model, and compute AIC scores for AR(m) candidates for m = 0,1,2,3,4, and see what model is selected. Include also the heteroscedastic model that uses a non-constant noise level, as in Exercise 1(c).

Exercises for Jan 24th: 1 and 2, from the "Exercises, Version A" set that is now downloadable from the website.

Please go to the "h03" (autumn 2003) version of this course website for information about and access to course literature. We shall start with Richard L. Smith's "Environmental Statistics", Chapters 1 and 2 (with further information becoming available later).