Re: Multi-objective vs Stochastic!

Michael wrote:

> Thanks! Okay, so if uncertainty is to be considered, what are the
> pluses and minuses when it comes to using Monte-Carlo or Stochastic
> Programming?

The big minus of Monte Carlo is that in every scenario you assume
to be fully clarevoyant, so you will never hedge properly. This makes
it very hard to assemble the scenario solutions into an implementable
policy. Stochastic optimization gives you flexibility to react to changing
environments.

Stein Wallace gives a very simple thought experiment: If one scenario
gives as the optimal solution to invest in one truck, and another scenario
wants to use two trucks, it is hard to see the optimally hedged solution,
which might be not to use any trucks at all and rely on a fleet of
stationwagons instead, which can be adjusted up or down once you
receive better information down the road.

I still recommend that you read a book on stochastic optimization.
The above and related issues are featured fairly prominently in any
of them.