I'm messing about with a snub geometry solver. I originally solved the script manually with sliders (tedious), galapagos wasn't quite accurate enough so I created a 2 level loop using hoopsnake which boosted the accuracy to grasshopper's capabilities, however, this is time dependent.
The current convergence method is parabolic, but Hoopsnake's ability to be paused mid-loop or even just manipulated during the loop presents the opportunity to manually change the step factor while the loop is running (but this obviously needs user input during the looping process).
I'm thinking that it would be possible to define a more dramatic convergence curve that could increase the solving rate and hopefully generalise it, so the curve is not dependent on the starting values. I don't think I'll touch this for a couple of weeks, but I'm open to any suggestions...??
Hoopsnake
566 members
Description
Update:
Hoopsnake is now Open Source!
Check out the repo in Github!
Fork it and fix it :)
Hoopsnake is a component that allows feedback loops within Grasshopper.
Snub Geometry
by Daniel Prohasky
Nov 9, 2011
I'm messing about with a snub geometry solver. I originally solved the script manually with sliders (tedious), galapagos wasn't quite accurate enough so I created a 2 level loop using hoopsnake which boosted the accuracy to grasshopper's capabilities, however, this is time dependent.
The current convergence method is parabolic, but Hoopsnake's ability to be paused mid-loop or even just manipulated during the loop presents the opportunity to manually change the step factor while the loop is running (but this obviously needs user input during the looping process).
I'm thinking that it would be possible to define a more dramatic convergence curve that could increase the solving rate and hopefully generalise it, so the curve is not dependent on the starting values. I don't think I'll touch this for a couple of weeks, but I'm open to any suggestions...??
Snub cube: HS_2LevSolver.gh
Thank you for any suggestions :)