DEoptim strategy selection

Question

I am not sure that this is particularly programming related, but was looking at a demo of parallel optimisation in a financial portfolio context using DEoptim, shown here.

Which I found by looking at the presentation given here and here.

And it seems that the strategy used is strategy=6, is there any particular reason why this one is "better" than the others? does it get to the global minimum faster? is it particularly good for portfolio optimisation problems?

Also as a separate note, how does one load into R the rda files that are suggested in the demo, that are used i.e. here. Becasue when clicking on download I get the 10y_return.gz file, but don't know how to read it into R...help on that would be appreciated.

Answer 1

strategy=6 is adaptive differential evolution by Zhang and Sanderson (2009). If the control parameter c is non-zero, then the crossover and mutation will be adjusted during the optimization. This can often speed convergence on particularly difficult objective functions/constraints, as is the case with the constrained portfolio optimization examples.

I get 10y_return.rda when I click the download link. If you truly get a .gz file, then you just need to unzip it first.