Register

Stochastic Optimization for Large-Scale Systems

Scientific and technical environment of the training course

Centre international de rencontres mathématiques - UMS 822

COURSE DIRECTOR

Patrick FOULON

Senior researcher

UMS 822

LOCATION

MARSEILLE (13)

ORGANISATION

4 days - From 4 to 12 attendees
From Monday (2:00 pm) to Friday (12:00 am)
Training course in English
CIRM is a scientific and conference centre. Onsite accommodation is strongly encouraged to benefit from all exchanges taking place during the day and in the evening. Contact: interface@cirm-math.fr 

TRAINING FEES

2000 Euros

AT THE END OF THE TRAINING COURSE

Satisfaction survey from trainees
A certificate of training is delivered.

COURSE DATE

19278 : from monday 04/11/2019 to friday 08/11/2019

January February March April
May June July August
Sept. Oct. Nov.
19278
Dec.

OBJECTIVE

- Learn and know how to put into practice resolution methods of optimization problems that are large scale due to uncertainties and to temporal and spatial dimensions

PUBLICS

Researchers and engineers working in the field of energy, transport, finance, etc.

Attendees are invited to download and fill out the survey from our web site to adapt the programme to their expectations.

PRE-REQUIREMENT

Knowledge of continuous optimization (linear programming, convexity, duality) and probability calculus (random variable, conditional expectation)

TRAINING PROGRAMME

Sensors and data abound. This spatial and temporal information is supposed to allow a better management in new energy systems, transport or eco-industrial parks, to quote a few examples. This leads to problems of large-scale optimization, the formulation of which is delicate. Indeed, one needs to take into account at least three dimensions: temporal (stages of decision, dynamics, inertia), spatial (different decision units connected by flows), stochastic (various scenarios are possible) and thus risk. This leads to multi-stage stochastic optimization problems which enter the class of large-scale systems.

- Stochastic programming: from one-stage to multistage problems
- Stochastic optimal control and dynamic programming
- Stochastic Dual Dynamic Programming Algorithm
- Advanced decomposition methods
- Practical works in Julia
- Interactive session on industry case studies

Detailed programme available on CIRM website.

This 4-day training session alternates courses, computing and interactive case studies sessions.

EQUIPMENT

Attendees should bring their own laptop, equipped with the free software Julia. Computers will be also available onsite for those in need.

SPEAKERS

P. Carpentier, J.-P. Chancelier, M. De Lara and V. Leclère (Professors)
+ d'infos