Ні!
Ran into a problem in the Galapagos. I am currently looking for the optimal placement of rectangles in a limited area. Each rectangle can move along the X and Y axis, and rotate around itself. Also, each rectangle has a buffer zone along the long side. The condition of the algorithm is that the buffer zones of the rectangles do not intersect with each other and do not go beyond the limits of the zone. The problem is that Galapagos doesn’t work when, before Galapagos starts, the buffer zones already have an overlap. Can anyone help resolve this issue.
Thank you.
Galapagos problem.gh (37.2 KB)
once Galapagos tells you the best fit of your setup is for instance the following, what are you going to do with that information / insight / result ? 
when the Fitness function is defined as a 0/1 boolean (here you have the count of building footprints intersections must be zero and all the points at footprint corners must be inside the predefined area-curve, which gets multiplied by a constant), then -maximizing the Fitness- will give you a list of “equally reinstatable” results for which the fitness function is not zero
it’s like a digital 0/1 signal where everything where Fitness = 1 will be as good as any other… hypotetically, the following would be good solutions (there’s a tiny gap between each curve, so no intersections whatsoever, and all the curves lie inside the initial boundary)
in order to get the best from Galapagos I would advice to include in the Fitness some granular metrics to evaluate additional features which would make a given solution better than another one
the thing of providing 8 different multiples of 45° for rotations of a rectangle is a bit of an overkill, because after 0, 45, 90, 135 you just get 180 = 0
but of course, if you are planning to use that on buildings that have more complex shapes, then that is fine (it is fine also if rectangles have a “direction”, for instance like if entrance is always along edge at index 0 for each rectangle…)
I want to try to create a grid of points in the middle of the plot so that the buildings are placed more orthogonally to each other. It goes like this: The center of the building is attached to a random point, then the algorithm calculates whether the buildings and their buffer zones have intersected with each other, and whether the buildings have not gone beyond the plot. But I still don’t understand how to implement it better and whether it will work. Roughly speaking, replace the MD Slider node
