Winter school 2017

Recent Advances in Extremes

The Chair of Statistics at EPFL is organizing a winter school on February 13-14 2017 on multivariate and spatial extremes, just before a three-day workshop on Risk Quantification and Extreme Values in Applications. The school consists of lectures (2×4 hours) taught by specialists in the field and of student presentations. The target audience is young researchers and students specializing in extreme value theory.

Instructors and content

The two instructors are Jonathan A. Tawn (Lancaster University) and Clément Dombry (Université de Franche-Comté, Besançon). The first course will be on recent advances in multivariate extreme value modelling.


  1. Fundamental methods: componentwise maximum and threshold approaches. Multivariate point process and regular variation, angular measure, inference methods.
  2. More sophisticated methods: Tail dependence, measures of asymptotic dependence, asymptotically (in)dependent extremes; coefficient of tail dependence, (hidden) regular variation. Extrapolation at different rates in different margins. Inference methods. Second order structure.
  3. Conditional extreme values and the Heffernan–Tawn model: theory and methods.
  4. Implications for modelling extremes of univariate Markov processes.
  5. Applications to finance, river flows and rainfall, ocean waves and heatwaves.


The second will be focused on max-stable processes

  1. Theory of max stable processes: Theoretical bases, models for max-stable processes, extremal and subextremal functions, hitting scenario. Exact and conditional simulation of max-stable processes.

Students are expected to have a background in extreme value theory, typically a first course in extreme value theory and notions of spatial extremes or max-stable processes.

EDMA- Mathematics Doctoral Course

Participants attending both the winter school and the workshop can obtain 2 ECTS credits for the doctoral course (offered by EDMA); please email us if such is the case. An additional administrative fee of 100CHF will be charged by EPFL for the credit transfer for external auditors. A written report will be asked from the participants in addition to active participation in the activities (courses, student presentations, poster session and speed meetings).


Participants will have the opportunity to give a presentation, subject to time availability. There will also be a poster session during the main workshop.


You can find here the book of abstracts for the winter school.

Speed meeting

We will organize a speed-meeting to give PhD students the possibility to exchange with the speakers of the workshop. This is a great opportunity for participants to get feedback and foster new ideas on their research. Each meeting will be 15 minutes long. Students should summarize their research problem and question related either to their own work or to the research of the professor (from an article, for example). This summary will be sent in advance to the senior participants who agree to participate in the speed meetings to increase the fruitfulness of the discussion.

Fees and financial support

We have limited financial support for PhD students. This would cover accommodation for five days, lunches, coffee breaks and the lectures. Given the funding availability, the latter will be competitive and a selection will be done by the organization committee. Priority will be given to PhD student giving a presentation (talk or poster).

The registration fee is 200 CHF for PhD students receiving financial support (with accommodation) and 150 CHF for all others. This also covers the workshop fees. The payment form will be available soon.


Registration is now closed for both the workshop and the winter school.

We no longer accept applications for financial support at this stage (PhD students had to postulate before the end of October for their funding application to be considered).

Students presenting a poster or giving a contributed talk during the week can provide an abstract upon registering or in the Fall with the payment form. Please avoid multiple submissions.

More information will made available in due time on this website.