Course Evaluation / Report 2018
Teacher: Martin Sahlén, Astronomy - This email address is being protected from spambots. You need JavaScript enabled to view it.
Dates: Week 32 - 38 (6 August - 21 September 2018)
The course will be a mix of self-study, seminars, hands-on workshops, and lectures.
A fairly compressed, hands-on course with the explicit aim of allowing the participants to work on a current research problem of their own and gain directly relevant skills, tools and results alongside an overview of the topic. The content will be geared towards astronomy/astrophysics, but should be relevant and accessible for anyone using data to test parametric models using computational methods. Examination consists of two hand-ins, workshop participation and an oral project report. You are also welcome to attend parts of the course “not for credit”.
If you are interested in attending, please contact This email address is being protected from spambots. You need JavaScript enabled to view it.
Programming experience and knowledge of basic statistics required (e.g. PhD course “Statistical Methods in Physics”). Limited number of places.
- Hand-in 1: Build and test a model (compulsory, due 26 August)
- Hand-in 2: Exploration of model and data, future observations (NB: PCA section revised 4 September 2018, Fisher forecast instructions added 8 Sept! Compulsory, due 16 September)
- Lecture 1: Monte Carlo Markov Chain methods. See Slack #lectures for lecture notes.
- Lecture 2: Data-based modelling, forecasting and optimisation of experiments. Course handout preparatory reading. See Slack #lectures for lecture notes.
- Extra lecture: MCMC and inference mechanics in more detail. See Slack #lectures for lecture notes.
- Workshop 1: Monte Carlo Markov Chain parameter inference, visualisation of results (compulsory; preparatory instructions; full instructions; errata [important!]; example Python script)
- Workshop 2: Model building (compulsory; preparatory instructions, full instructions). See Slack #workshops for workshop notes.
- Workshop 3: Model-data exploration (compulsory; preparatory instructions, full instructions, Gaussian process Python script)
- Workshop 4: Hack Day (flexible attendance; instructions)
- Seminar 1: Theoretical modelling, Bayesian model inference. See Slack #seminars for lecture notes.
- Seminar 2: Handling large data sets. See Slack #seminars for lecture notes.
- Seminar 3: Presentations of projects (compulsory)
- emcee Monte Carlo Markov Chain engine: http://dfm.io/emcee/current/
- scipy toolkit: www.scipy.org
- scikit-learn: scikit-learn.org
- Foreman-Mackey, D. et al.: Emcee documentation, https://arxiv.org/abs/1202.3665
- MacKay, D. Information Theory, Inference, and Learning Algorithms.Freely available at http://www.inference.org.uk/mackay/itprnn/book.html
- von Toussaint, U.: ”Bayesian inference in physics”, Reviews of Modern Physics 83, 943 (2011), available at https://journals.aps.org/rmp/abstract/10.1103/RevModPhys.83.943
- Trotta, R: “Bayesian methods in cosmology”, available at https://arxiv.org/abs/1701.01467
- Sharma, S: "Markov Chain Monte Carlo Methods for Bayesian Data Analysis in Astronomy", Annual Review of Astronomy and Astrophysics 55:213-59 (2017), available at https://www.annualreviews.org/doi/abs/10.1146/annurev-astro-082214-122339
- Instruction and course handouts
- Additional extra reading will be communicated during the course