MA8702 Advanced computer-intensive statistical methods - Spring 2026
This is a phd course in statistics - and requires self-study and active participation.
10.04.2026: Some addition to our discussion today (generated using ChatGPT):
10.04.2026: Reference group meeting (Participants: Andrea Riebler, Gard Gravdahl, Øyvind Singsaas, Elling Svee)
- Students would be interested in investigating more modern methods. They liked a lot the part about variational Bayes and the accompanying guest lecture, and would wish more like this.
- The projects could give more room for own creativity and exploration, and go more away from having a recipe, implementing it and discussing the results.
- One suggestion was to reduce the number of projects and possibly have projects that link different parts of the course in one realistic project. Another suggestion, as has been seen in another PhD course, was to write a small "paper" (about five pages) using methods seen in the course and relate it to the own PhD work. Peer students should then read the paper and write a review.
- The students generally like the structure of the course, going through research papers and discuss them. An alternative could be that students present basics of a topic and then an expert (a member from the department doing related research) presents some of his/her work in a conference style presentation, that is then discussed .
23.03.2026: There is no meeting today. This week starts with part 3 on sequential Monte Carlo. Please work through:
- Särkkä, S. (2013). Bayesian Filtering and Smoothing, Cambridge University Press. Chapter 1, 4, 5.1, 5.2.PDF
- Kalman Filter Intro Video, Part 1, by Jonathan Kelly: Youtube Video
- Kalman Filter Intro Video, Part 2, by Jonathan Kelly: Youtube Video
22.01.2026: Thanks for a nice session today. Here you find further references for Hamiltonian Monte Carlo, the NUTS and Stan. See you on Monday:
Our meeting with Øyvind presenting the Hamiltonian paper by Michael Betancourt will not be tomorrow but moved to Thursday 22.01 from 10:15-12:00 in room 634. Please come well prepared 
- 12.01.2026: For the next meeting on January 19th please go through the material from today and work through chapter 12 of Gelman et al. (2014). We will discuss the following paper: Betancourt, M. (2017) "A conceptual introduction to Hamiltonian Monte Carlo." , so please be well prepared.
- 05.01.2026: For the next meeting on January 12th please go through the material from today and work through chapter 11 of Gelman et al. (2014). We will discuss the following paper: Dunson, D.B., Johndrow, J.E., (2020). The Hastings algorithm at fifty, Biometrika,107(1) 1–23, so please be well prepared. PS: Project 1 is already out.
- 11.12.2025: The course starts on January 5th, 10:15-12:00, 656 Simastuen, Sentralbygg 2.
Course coordinator/lecturer: Andrea Riebler
Course description: https://www.ntnu.edu/studies/courses/MA8702#tab=omEmnet
Course parts:
- Part 1: Markov chain Monte Carlo techniques with a link to the software Stan
- Part 2: Approximate Bayesian inference and Gaussian process regression
- Part 3: Sequential Monte Carlo Methods
Course evaluation:
The grade for this course is pass/fail. There will be three projects, one for each topic, which need to be passed in order to be admitted to the exam.
There will also be required active participation in the meeting hours within discussion or presenation of research papers.
Recommend previous knowledge:
- TMA4300 Computer-intensive statistical methods
- TMA4295 Statistical inference
- TMA4267 Linear statistical models
- TMA4315 Generalized Linear Models
Programming/IT-knowledge
Experience and good programming skills in R, or another high-level programming language.
Reference group The whole course will be used as reference group.