You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The algorithm should not depend on the scale of the rooms. If both polygons and respective doors are equally scaled, the algorithm should produce similar results with similar speed.
Steps to reproduce
None.
Actual results
The algorithm converges in a similar way with regard to the number of iterations. However, the number of iterations per second decreases as rooms are scaled up. It should be further investigated what part of the algorithms slows it down.
Expected results
Number of iterations and iterations per seconds should not be worse when scaling rooms up.
Notes for the thesis
The reason why scaling up the rooms yields worse results is still unknown.
Update 1
Picking a random intersection now operates in O(1) rather than in O(n) with "n" being the length of a chosen line.
Update 2
Method AddNodeGreedily was changed in a way that it does not search the whole space but rather it tries every i-th point. Where i is chosen in a way that the number of points is left the same. It increases the speed of iterations and has little effect on convergence rate.
ConfigurationSpacesGenerator is currently very slow when scaling up the rooms and must be reworked.
Description
The algorithm should not depend on the scale of the rooms. If both polygons and respective doors are equally scaled, the algorithm should produce similar results with similar speed.
Steps to reproduce
None.
Actual results
The algorithm converges in a similar way with regard to the number of iterations. However, the number of iterations per second decreases as rooms are scaled up. It should be further investigated what part of the algorithms slows it down.
Expected results
Number of iterations and iterations per seconds should not be worse when scaling rooms up.
Notes for the thesis
The reason why scaling up the rooms yields worse results is still unknown.
Update 1
Picking a random intersection now operates in O(1) rather than in O(n) with "n" being the length of a chosen line.
Update 2
Method
AddNodeGreedily
was changed in a way that it does not search the whole space but rather it tries every i-th point. Where i is chosen in a way that the number of points is left the same. It increases the speed of iterations and has little effect on convergence rate.ConfigurationSpacesGenerator
is currently very slow when scaling up the rooms and must be reworked.Benchmark
The text was updated successfully, but these errors were encountered: