Hi everyone!!
I am developing a generative feseability study definition with grasshopper for the studio im working on. The aim of this is to take just a curve from rhino and make automatically a building for a feseability studio. After a long trip, I can say that the definition does everything i want him to do but know im facing the optimization problem . I would like to optimize firstly % of units requierement such as : 25% of 1D, 40% of 2D, 35% of 3D, for example.
These are the development schematic steps I have taken to explain the problem
Step 0 : Originial curve ‘L’
- L
Step 1: 2 units are placed at the start and the end of the curve which makes the original curve L to be shorten (so L changes to ‘L1’ )
SlidersA for choosing which units go to the start and end point.
- L1
- L > L1
Step 2: L1 is evaluated and automatically a module of stairs is placed in the middle. So L1 is split into L2 and L3 (or if 2 stairs of modules had been placed, it will return L1, L2 and L3)
Sliders for choosing which units go besides and in front of the stairs
- L2 and L3
- L > L1 > L2 and L3
Step 3:
I have a database (list) of all combinations possible (done with itertools in Ghpython). This database are just the sum of façade lengths. Once step 1 and 2 finish, the resultant lengths of L1 and L2 are compared to the database, and filters them giving me a list of the combinations that fit inside the lenths L1 and L2. with a slider connected to this list i ‘choose’ the combination to be placed at L1 and L2.
Sliders for choosing the combination to be placed in L1 and L2
Step 4 - Optimization - I explain below…
Image 1 - Shows the different steps. this image is just for a simple case, which optimization is not probably requiered due to the small length and small combinations possible, but hope you guys get an idea of the potential of the definition…and problems I’m having
Image 2 - It’s a conceptual picture of the definition showing the ordering sliders A,B and C that i would like to optimize
I would like to optimize the 3 part-sliders
My Experience:
If I just plug in the sliders ‘C’ the optimization here is pretty strightfoward, this is because L1 and L2 are kept fixed during the optimization which is in charge of finding the best combination of combinations to fullfill the mix requierements.
But what i really want is galapagos to optimize the 3 part sliders, to have more control and diversity to find the best solutions for achieving the mix requierements.
So my question is if conceptually, this can be optimized? A variable-sliderA which at the same time is changing next variable B and C could be pluged in Galapagos.
the problem I see here is that when galapagos iterates over sliders 1and 2, it changes the list from where sliders 3 takes the combination to place the units… and if im not wrong i am changing the search field in every iteration of the optimization.
Any suggestions or/and clarification if this is the way to optimize properly?
Thanks in advance.
sorry for not uploading any gh files. Later on the discussion i can be more explicit if someone is interest for debate…