Résumé de section
-
Contact:
- this page address: http://tinyurl.com/m2msiam-invmeth
- email: elise ( dot ) arnaud ( at ) univ-grenoble-alpes ( dot ) fr
deadlines:
- jan, 16: written report on personal research
- jan, 16: completed python notebook (for MSIAM students)
- oral presentations will be held on jan, 16
schedule for presentations:
- 11h15 : Juliette, Estelle, Marcelin - EnKF
- 11h35 : Thierry, Constantin - Variational Sensitivity Analysis
- 11h55 : Nicolas, Adrien - Automatic differentiation
- 12h15 : Ksenia, Michi, Maxime - Hybrid data assimilation
Calendar:
- https://edt.grenoble-inp.fr
- Planned courses (at 11h15, room D208 unless otherwise specified) :
- Oct 3 (room H105), 10 , 17, 24
- Nov 7, 21, 28
- Dec 5, 12, 19
- Jan 9, 16
Assessment (see more details below):
- All of you: you are requested to fill a self-assessment sheet before every class (see the homework section below)
- If you are enrolled for 1.75 ECTS (e.g., Ensimag 3A):
- A1- Personal research, using web and easy articles, written report (5 to 10 pages) and oral defense (15 minutes).
- If you are enrolled for 3 ECTS in a Professional track (e.g., MSIAM track IM student):
- A1- Personal research, using web and easy articles, written report (5 to 10 pages) and oral defense (15 minutes).
- B2- Practical work using Python, written report and commented source code.
- If you are enrolled for 3 ECTS in a Research track (e.g., MSIAM other students):
- A2- Personal research, using state-of-the-art research articles, written report (10 to 20 pages) and oral defense (15 minutes).
- B1- Practical work using Python, commented source code only.
-
-
Latex template for the report
-
-
-
-
-
Here you will find the course chapters, web and articles resources for your research.
-
-
-
-
-
-
-
-
-
-
This work is called "A" in the assessment process described above.
There are 2 levels:
- "A1" (normal)
- "A2" (in depth).
Your assessment will consist in:- written report assessment
- oral "question and answer" (individually, even if you worked in pairs)
A1 work (1.75 ECTS, e.g., Ensimag students)
Proposed subjects:- Ensemble Kalman filter (EnKF) and extensions
- Adjoint-based sensitivity analysis (VarSA)
- Global sensitivity analysis (GSA)
- Variational assimilation with non-linear models (VarNL)
- Particle filters (PF) especially for high dimension
Produce a short report of your research: 5 to 10 pages.
A2 work (3 ECTS, e.g., MSIAM students)
Proposed subjects:- Ensemble Kalman filter (EnKF) and extension
- Ensemble transform Kalman filter (ETKF)
- SEEK filter
- Adjoint-based sensitivity analysis (VarSA)
- Global sensitivity analysis (GSA)
- Automatic differentiation (AD)
- Variational assimilation with non-linear models (VarNL)
- Particle filters (PF), especially for high dimension
- Hybrid variational/filtering assimilation
- Reduced order assimilation
- Model error control in DA
Produce a report of your research: 10 to 20 pages. -
This work is called "B" in the assessment process described above.There are 2 levels:
- "B1" (normal)
- "B2" (in depth)
Ensimag (1.75 ECTS) students can still do it, for extra credit.B1 (MSIAM all tracks except IM)Produce your commented source code.B2 (MSIAM track IM) (other tracks can still do it for extra credit)Produce a written report (with figures) and your commented source code.Coding is done in Python. To open and work with the notebook, please install and use Jupyter -
Please find here suggested resources adapted to the various subjects. You are free to use them and/or find other resources on your own.
Links:
- Ensemble Kalman filter: https://en.wikipedia.org/wiki/Ensemble_Kalman_filter
- Automatic differentiation:
- https://en.wikipedia.org/wiki/Automatic_differentiation
- http://alexey.radul.name/ideas/2013/introduction-to-automatic-differentiation/
- https://www-sop.inria.fr/tropics/ad/whatisad.html
- Filtering assimilation with non-linear models: https://en.wikipedia.org/wiki/Extended_Kalman_filter
- Particle filters: https://en.wikipedia.org/wiki/Particle_filter
Pdf of tutorials, simple and research articles are below, grouped by subject:
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-