Origins Data Science Statistics Block Course

22 Sept 2025, 14:00 → 2 Oct 2025, 12:00 Europe/Berlin

Alps Seminar Room (Max Planck Institute for Physics (MPP))

Jonas Würzinger (ODSL/MPP/TUM), Nicole Hartman (TUM)

The Origins Data Science Lab (ODSL) is organizing a block course on "Introduction to Statistics" targeted towards PhD students in physics. Interested MSc students from LMU and TUM are welcome as well.

Please Register for the Event using the Registration Link.

Lecturers:

• Nicole Hartman, nicole.hartman@tum.de
• Jonas Würzinger, jonas.wuerzinger@tum.de

Tutors:

• Nadine Bourriche

Block 1: Probabilistic reasoning

Block 2: Statistical inference

Forms of credit: We will try to make ECTS points are available (details forthcoming).

TutorialExercisesWeek2.pdf

Monday 22 September

14:00 → 15:20 Lecture 1: When the likelihood is tractable

Speaker: Nicole Hartman (TUM)

Lecture1-tractable-likelihoods-blank.pdf

Lecture1-tractable-likelihoods.pdf

prob-distributions-blank.pdf

Session-1.pdf

15:30 → 16:50 Tutorial 1

Speaker: Nadine Bourriche

Tutorial-1.pdf

Tutorial-1-SOLN.pdf

Tuesday 23 September

09:00 → 10:20 Lecture 2: Conjugate Priors

Speaker: Nicole Hartman (TUM)

Lecture2-likelihood-relationships-blank.pdf

Lecture2-likelihood-relationships.pdf

10:30 → 11:50 Tutorial 2

Tutorial-2.pdf

Tutorial-2-SOLN.pdf

14:00 → 15:20 Lecture 3: Gaussian Distribution

Gratefully drawing from Jakob Kollmüller's lecture from last ODSL stats course (pp. 23 - 35):

https://indico.ph.tum.de/event/7418/attachments/5500/7135/Lecturenotes_ProbabilisticReasoning.pdf

Speaker: Nicole Hartman (TUM)

session3-black-white.pdf

15:30 → 16:50 Tutorial 3

New room ORIGINS seminar room (basement)
https://share.google/BPlsmDYRzcfsI4xhQ

github: https://github.com/odsl-team/block-course-sept25-stats/tree/main/Tutorials

Tutorial-3.pdf

Wednesday 24 September

09:00 → 10:20 Lecture 4: Hierarchical Models and Monte Carlo sampling

New room ORIGINS seminar room (basement)
https://share.google/BPlsmDYRzcfsI4xhQ

Speaker: Nicole Hartman (TUM)

Lecture4-hierarchical-models-blank.pdf

10:30 → 11:50 Tutorial 4

14:00 → 15:20 Lecture 5: Markov Chain Monte Carlo

Speaker: Nicole Hartman (TUM)

Lecture5-MCMC-blank.pdf

15:20 → 16:40 Tutorial 5

Thursday 25 September

09:00 → 10:20 Lecture 6: Variational Inference

Speaker: Nicole Hartman (TUM)

Lecture6-VI-and-SBI-blank.pdf

10:30 → 11:50 Tutorial 6

Optional: Since we discussed SBI at the end of the lecture today, I give you an example showing how it works on a (simplified) LIGO simulation.

Since we didn't cover normalizing flows in detail today, it's not expected that you'll be able to work through the blank exercise on your own, but the idea to look at the solution (input data x, the 2d posterior theta being modelled).

Challenge: Is there a dataset from your research you could plug into this notebook and train a small proof-of-concept NPE posterior in this hour?

Speaker: Nadine Bourriche

Neural posterior estimation: LIGO

Monday 29 September

14:00 → 17:00 Lecture 1

New room ORIGINS seminar room (basement)
https://share.google/BPlsmDYRzcfsI4xhQ

Speaker: Jonas Würzinger (ODSL/MPP/TUM)

Stats_Blockcourse_Day1.pdf

Tuesday 30 September

09:00 → 12:00 Tutorial 1

New room ORIGINS seminar room (basement)
https://share.google/BPlsmDYRzcfsI4xhQ

Speaker: Nadine Bourriche

14:00 → 17:00 Lecture 2

New room ORIGINS seminar room (basement)
https://share.google/BPlsmDYRzcfsI4xhQ

Speaker: Jonas Würzinger (ODSL/MPP/TUM)

Stats_Blockcourse_Day2.pdf

Wednesday 1 October

09:00 → 12:00 Tutorial 2

New room ORIGINS seminar room (basement)
https://share.google/BPlsmDYRzcfsI4xhQ

Speaker: Nadine Bourriche

14:00 → 17:00 Lecture 3

Speaker: Jonas Würzinger (ODSL/MPP/TUM)

Stats_Blockcourse_Day3.pdf

Thursday 2 October

09:00 → 12:00 Tutorial 3

Speaker: Nadine Bourriche