KitWallace
commented
6 years ago New idea
1 make eps proportional to the path increment size 2 divide into two phases - a) to find a crossing point |f(i)-f(j)| < eps |i -j| > dz b) get nearest crossing value - increment i and j whilst near condition is true to find minimum distance, resume a) at i' = min-i + dz+1
2 doesnt work because approach to point is sometimes from decreasing j
I ve experimented with various forms of this algorithm - the current one works ok for the 3link curve but is poor for lissajous. I suspect the problem with the later is that the rate of advance various considerably along the curve but is more uniform for 3link.