Messages
You have done a good job.
Three of you were just above the line,
the rest was well above the line,
many works were excellent.
Your answers will be evaluated according to standard procedures for STK 9xxx courses.
NB for STK 4051 only:
All subtasks are of equal weight.
You need 40% to pass, if you solve 40% of the problems you have to have these 100% correct (difficult to achieve).
If you solve 80% of the problems you need to have an average of 50% correct to pass the exam (this is easier).
If you solve 100% of the problems and have these mostly correct, you have done excellent work. :-)
I have gotten question from students about whether STAN is relevant for the exam. The answer to this is that it is a part of syllabus, and is equally relevant as the other topics.
It is important to be operational (i.e. to be able to program) on all topics in the syllabus.
In the course we assume that you are using R and material are provided only in R. In general you are free to choose other programming languages (and many have done so), but you will then need to translate scripts that are given.
If you haven't gotten a feed back, send me a mail.
If you have any follow up questions, mail me.
When testing Rstan on Mac, you might experience runtime errors.
Per August Jarval Moen, found that it could be caused by communication errors in the paralelliztion.
try:
parallel:::setDefaultClusterOptions(setup_strategy = "sequential")
as a work around.
If you have any questions in the period until exam,
send questions on mail.
I will publish Q and A on the web page.
First question is allready in:-) . See here for Q & A
because I put the dates right here...
Just because the old expired ...
Start 13.15 as usual.
Odd Kolbj?rnsen is inviting you to a scheduled Zoom meeting.
Topic: Odd Kolbj?rnsen's Zoom Meeting
Time: May 14, 2020 01:00 PM Amsterdam, Berlin, Rome, Stockholm, Vienna
Join Zoom Meeting
https://uio.zoom.us/j/65101475206
Meeting ID: 651 0147 5206
Join by SIP
65101475206@109.105.112.236
65101475206@109.105.112.235
Join by H.323
109.105.112.236
109.105.112.235
Meeting ID: 651 0147 5206
I will give you all feedback, but I prioritize to give feedback to those that have an incomplete exercise in the first round.
An erroneous expression for the effective sample size was used.
The solution on web is now corrected
I have been through the syllabus. By next lecture you have finished your compulsory exercise. So it is time to summarize the course, and for you to start preparing for the exam.
Next lecture will be the summary. Since your exam is on the same form as the compulsory exercises, I will use more time to go through these than normal. I have suggested on the web page that I will use one week for each of the parts. I will also go through previous/trail exam in exercises.
In addition to the compulsory exercise I would also appreciate if you have questions for me about thing that you currently think is unclear. I will have time in lecture to answer these questions. I have listed these slots as selected topics from optimization/simulation in the Schedule.
First time for me, send me a message if it does not work.
For the PhD version, the exam is organized the same way as for STK4051. Ordinary rules for pass/fail will be applied. There will not be an additional oral presentation but instead an extra exercise on the exam.
Fore those attending the course STK4051. The exam is run as a 7-day individual home exam with all aids allowed. The tasks will be similar to a mandatory task. Collaboration will not be considered cheating, but you must have formulated and written the answer that is submitted, and it should reflect your understanding of the syllabus. The grade will be pass/fail, and the limit for passing will be 40% (like E at the regular exam). The exam is published in Inspera at the original examination date and must be submitted as one PDF file in Inspera within the same time 7 days later. The faculty works on solutions for students who may not have access to a computer/network.
McMC Bayes is a short note covering, many of the important aspects for the "McMC- part" of the course.
Neal report covers MCMC using Hamiltonian dynamics
On the page Reparameterization version 2,
there was a typo. the corrected version is on the webpage. (slides)
Next Thursday (26.march) we will have the first of our digital exercises and lectures.
I have set up a two recurrent meetings in Zoom.
Please find the two different links on the web page.
I will attempt to record and the lecture online, not the exercise
If you do not want to be identified in recording,
* Sign in with a dummy name
* Switch off your video
At entry to meeting all of your microphones are muted to avoid bacground noise. To ask questions during lecture. Use the chat or un-mute your microphone and ask the question. I will try to answer questions as they come.
I put a link to this "Cheat sheet" for matrix identities on the web page. Really useful.
I will try to arrange a video meeting march 19. Where I will go through the exercise. The current solution I have require a windows machine for those that will participate. If you have a windows machine and is interested to attend. Please send me your mail and I will summon the meeting online.
Enjoy...
If you have questions send them to me and I will collect questions and find a way to give answer.
Lecture next week (March 19th) is canceled as the university is closed, and there is currently no recording options. I will post some extra exercises for you to do on the exercise page. So that you have something meaningful to do while the country is “corona stranded”.
I hope we will be back on track with digital recordings the week after (March 26th)
There was a problem with the image in the video I uploaded. I checked the equipment prior to recording, but the technique failed me in the final take. The best solution for you is to download the slides and listen to the video as a soundtrack.