# TMA 4300 Computer Intensive Statistical Methods

## Messages

• 11.06.2014: A solution to the exam is available here.
• 22.05.2014: You are also allowed to bring a dictionary in any language to the exam.
• 15.05.2014: As noted here, you are allowed to bring one yellow paper (A5 with stamp) with your own handwritten formulas and notes. You can thereby write on both sides of the sheet.
• 13.05.2014: Exercise 3 is graded. You can go to Geir-Arne (room 1238, Sentralbygg II) to get your evaluation. As stated in the course description the exam will be only in English. You are free to answer in Norwegian or English.
• 24.04.2014: Exercise 2 is graded. You can go to Geir-Arne (room 1238, Sentralbygg II) to get your evaluation. Do not forget the deadline on Monday, the 28th of April, for exercise 3.
• 08.04.2014: Exercise 3, problem D: With 1e-5 we mean 0.00001.
• 08.04.2014: A summary of the reference group meeting today is here
• 08.04.2014: Today after the exercise class will be the last reference group meeting. If you have comments please tell them the members of the reference group.
• 03.04.2014: Exercise 3, Problem C 2. Use lm(log(meas) ~ pers ,data=bilirubin) to fit the regression model
• 27.03.2014: Exercise 3 is out. You find the problem description, as well as the schedule for the oral presentation here. Also data files and R-helper functions are povided in a zipped-archive on the exercises webpage. Recall that the deadline for submitting the solutions for exercise 2 is April 1st.
• 21.03.2014: Excercise 2: part 3c). There is work around regarding the problem with the geweke.plot() (thanks to MMR). The object transmitted to the function as.mcmc() should be a matrix, then it works. Hence, transforming a sample vector first to a matrix there seems to be no problem, for example

mysamples ← rnorm(100000)

plotMCMC ← as.matrix(mysamples)

library(coda)

geweke.plot(as.mcmc(plotMCMC))

• 20.03.2014: Excercise 2: In part 3c) the geweke.plot() function should be used. It seems that the sample size for this function is limited to 10^4 - 1 samples. Thus, you will get an error if using all 10000 samples. To solve this just use one sample less for each parameter you look at, namely 9999 samples.
• 18.03.2014: A summary of the second reference group meeting can be found here.
• 17.03.2014: Exercise 1 is graded. You can go to Geir-Arne (room 1238, Sentralbygg II) to pick up your reports.
• 13.03.2014: Exercise 2: Please find here a function implementing the density function of a multivariate normal distribution in canonical representation, i.e. using b and the precision matrix Q. This should help you in calculating the proposal ratio in the acceptance probability.
• 13.03.2014: Exercise 2: Please find here some R-code that illustrates how to compose the block matrix used in the block sampler.
• 12.03.2014: On Tuesday 18.03 and Tuesday 08.04 will be the second and third reference group meeting. Please send your comments and feedback to the reference group
• 10.03.2014: Please note, the additional files mentioned in Exercise 2 are also available from the exercise page. They are prepared for downloading in a zip-compressed archive. If you extract the archive you should be able to load/source the files.
• 03.03.2014: Exercise 2 is out. You find the problem description, as well as the schedule for the oral presentation here. Recall that the deadline for submitting the solutions for exercise 1 is tomorrow, March 4th.
• 20.02.2014: Note the time for the oral presentations is during the official exercise hours that means 16:15-17:00 in K5, see Time plan. Only one presentation will be during the lecture today due to certain circumstances.
• 18.02.2014: The oral presentation schedule is updated, see . In particular, there will be already one presentation TMA 4300 Computer Intensive Statistical Methods at the beginning of the lecture coming Thursday. Remember, that the lecture coming Thursday will be in R21.
• 04.02.2014: A summary of the reference group meeting today is here
• 30.01.2014: The first reference group meeting will be on 04.02.2014. Please send your comments and feedback to the members of the reference group.
• 30.01.2014: Exercise 1 is out. You find the problem description, as well as the schedule for the oral presentation and some guidelines here.
• 23.01.2014: ==== IMPORTANT ==== : Please send me (andrea [dot] riebler [at] math [dot] ntnu [dot] no) until tomorrow (24.01) whether you take part in the exercises and whether you do it individually or in a team of two (provide also the name of your teampartner). Please send only ONE email per team.
• 17.01.2014: The names and contact information for the reference group members is available here.
• 16.01.2014: IMPORTANT: If you do not already have access to the computer lab (Nullrommet 380A) please send me an email with your name, student number and study programme.
• 14.01.2014: We need minimum three students for the reference group of TMA4300 - that should be two from Industrial Mathematics and one from others (Erasmus, other study programmes). Please email Andrea (see course info tab for email) if you will join the reference group. There will be 3 meetings, and a (short) closing report must be filed.
• 10.01.2014: Some information regarding the statistical software R can be found here.
• 08.01.2014: Welcome to the home page of TMA300 Computer Intensive Statistical Methods for the spring semester 2014. The first lecture will be on Thursday January 09 at 12:15-14:00 in room R21. The lectures in this course will be in English.

## Lecturer

Andrea Riebler, room 1242, Sentralbygg II, andrea [dot] riebler [at] math [dot] ntnu [dot] no
NB!! Please use TMA4300 in the subject header of all emails concerning the course!

## Exercises

Geir-Arne Fuglstad, room 1238, Sentralbygg II, geir [dash] arne [dot] fuglstad [at] math [dot] ntnu [dot] no
NB!! Please use TMA300 in the subject header of all emails concerning the course!

## Lectures

• Tuesday 12:15–14, F2
• Thursday 12:15–14, R21