Colin945![]() Member ![]() Posts: 49 Joined: 2/26/2005 Location: Calgary ![]() | Hi Barry, I've read through the SW documentation and all it does is refer to the Fast Search without explaining the settings. What would be very useful is a document that lists each setting, what it does, and how/if it should be adjusted based on total permutations. The documentation implies there is some 'intelligence' being used to narrow the search space but it is clearly important to select enough samples, appropriately spread through the search space, to give a reasonable chance of finding a good optimum. I agree that for a lot of this optimization all one wants is a 'good' strategy, not necessarily the 'best'. That's where having a bit of an appreciation of the search approach being used is helpful. Do we build a proxy model? I'm not trying to discover any proprietary inventions that Nirvana has Any light you can throw on this area should be of use to all of us using SW. Many thanks, Colin |