geopsy.org

Dear Marc

Do you have any comparison between different methods of inversion?
I mean comparison between GA, SA and Neighborhood algorithm.
I want to know, what is the necessity of using Neighborhood method beside usual Genetic Algorithm or Simulated Annealing.
Does your algorithm improves the model results or your algorithm just is easier and faster to run?

Regards;
Ahmad

This is an interesting question. Up to now, I've not seen any robust comparison of all these meta-heuristic inversion methods. You can add Particle Swarm Optimization, Ant/Bee Colony Optimization, tens of other variants. There are some benchmark on relatively simple problems (special mathematical functions with multiple minima) but there is nothing (as far as I'm aware of) which involves a real inversion problem with data scatter, errors, bias,... If you manage codes for one or several of these inversion methods, it could be nice to adapt them to surface wave inversion in the same way as I did for NA.