An online tutorial of 1h30 proposed by Supply Chain of the Future Chair of the school of Ponts Paristech about the topic “ Multi-objectives maximization “, presented by Vincent Leclere who is a searcher in stochastic maximization at CERMICS, school of Ponts Paristech.
The function of the supply chain is to steer complex arbitrations in an uncertain environment on the whole values chain. This implies statement decisions dealing with multi objective organization…
For a mathematician, the word “MAXIMIZATION” means “found a solution with minimal cost among the ones satisfying the constraints”.
To do so, one must MODELIZE the problem, which means:
- Maximization variables (action’s lever).
- Strains to satisfy.
- Objective function (the cost of a solution).
It is often simultaneous objectives hardly comparable (for example, for the supply chain. Storage, costs, services, sustainability).
The standard solution consists in converting everything in the same unit (€) and necessitating coefficient determination in a random way (for example by comparing mobilized money to consumed cash / cost of one ton of CO2 versus cost of late delivery).
THE MULTI-OBJECTIVE MAXIMIZATION consists in keeping different objectives simultaneously to determine Pareto-efficient solutions: solutions for which we can’t improve an objective without losing on another. The choice between the different solutions is done by the operator who looks at the propositions done by the algorithm.
The goal of the tutorial is to specify these notions of solutions and to present few methods to handle multi-objective problems. This approach will be illustrated by short academic examples.
Read all about the Supply Chain of the Future Chair in the press