I have a question regarding the optimization of a list of randomized designs. Is it possible to have the following process: First, create a randomized design; second, optimize it for a certain number of iterations; third, evaluate/save the result; fourth, repeat the process with a new randomized design?
Thank you for your help.
Nothing from outer space in your request. Started writing code for things like these (C Alexander + Ill defined problems + Fuzzy Sets + … + HARNESS Project and the likes) a miiion years ago: now this is my core “app” for AEC Design (obviously strictly internal) . Think of it kinda a very big “add-on” to AECOSim.
I.e. doing some sort of MOO within Loop/Loops, storing something (volatile > persistent data) according some continuously variable/evolving Events/Validation rules (this means: maintaining some sort of History as well) … repeat. Some HAC Clustering is involved as well (for reasons hard to explain right now).
But I can hardly imagine a real-life thing like this done with components (but hope dies last - while should die first) most notably if you want an interactive procedure related with a MOO with many parameters (changing ones, mind). What this means? well … in real-life after some iterations solutions like these are heading towards a minimum entropy and not at all towards an ideal “state of matters” (this happens only in Planet Utopia).
This brings the 1++M question: when and where (not to mention why) we abandon our initial goals? (always we do that sooner or later - beause theory and reality have nothing in common).
have a look at the "anemone’ components, where you can do looping until an exit condition is met. A better alternative is Daniel Piker’s ‘Kangaroo’