Derivative Free Optimization / Optimisation sans Gradient

 

 (V04 course, season 2016/2017)

Common course between M2 AMS et M2 Optimisation (Paris Saclay)

 Anne Auger (INRIA), Laurent Dumas (UVSQ)



 
Optimization problems are encountered in many fields of engineering for which the associated cost function may be of various type : black box or explicit, with continuous or discrete variables, costly or not to compute, etc…

In many cases, the gradient of the cost function is not easy or even impossible to compute or it can exhibit many local minima leading to consider Derivative Free Optimization (DFO) methods.

This course deals with a large number of Derivative Free Optimization methods that have been recently developped, either local or global, deterministic or stochastic. It will be illustrated by various examples issued from industrial or medical fields.

 
 
Previous courses :

 

 cours 2011/2012, cours 2012/2013, cours 2013/2014  cours 2014/2015 course 2015/2016

 

 Outline

-PART 1 : STOCHASTIC METHODS (see page of Anne Auger)

 

1) General introduction / Motivation for DFO

2) Stochastic algorithms framework (essentially ES)

3) General principles for step size and covariance matrix adaptation. CMA-ES algorithm.

4) Comparisons between CMA/PSO/NEWUOA/BFGS

 

-PARTIE 2 : DETERMINSTIC METHODS

 

1) Local methods: direct methods (Pattern Search, Nelder Mead, MDS), trust region methods (NEWUOA)

2) Global methods : response surface methods  (RBF, kriging)

 

 
 
Schedule

 

 

Vendredi 26 novembre 2016, 14h00-17h15 (ENSTA, A. Auger) 

Vendredi 03 décembre 2016, 14h00-17h15 (ENSTA, A. Auger) 

Vendredi 10 décembre 2016, 14h00-17h15 (ENSTA, A. Auger) 

Vendredi 17 décembre 2016, 14h00-17h15 (ENSTA, A. Auger) 

Vendredi 06 janvier 2017, 14h00-17h15 (ENSTA, A. Auger) 

Vendredi 13 janvier 2017, 14h00-17h15 (ENSTA, L. Dumas) : direct methods

Vendredi 20 janvier 2017, 14h00-18h15 (new schedule, room1224  and for computer session 1212 ENSTA, L. Dumas) : trust region methods

Vendredi 27 janvier 2017, 14h00-18h15 (new schedule ENSTA, L. Dumas) : response surface methods (slides in french)

Vendredi 03 février 2017, 14h00-18h15 (new schedule ENSTA, L. Dumas) : other DFO methods+ exercices

 

Exam : vendredi 10 février2017, 14h00-17h00 : sujet (+ Matlab code)

 

 
 
Scilab/ Matlab scripts (deterministic part)

 

 

Pattern Search : pattern.sci

Nelder Nead : NelderMead.sci

MDS : MDS.sci

Trust region method (including the Lagrange interpolation in 2D) : Trust.zip (in Matlab)

 

 

 Bibliography (determinsitic part)

A. Conn, K. Scheinberg and L. Vincente, Introduction to Derivative Free Optimization, SIAM, 2009.

A review article on DFO : Journal Global Opt. 2013

 

An article on the non convergence of Nelder Mead : SIAM J. Opt.1998

 

An article on the convergence  of MDS:  PhD1989

A review article on metamodels RBF and Kriging : INRIA report, 2009

Description of the DIRECT method : DIRECT.pdf (Journal of Optimization theory and application, 1993)