Proper seeding can improve geometric fitting algorithms convergence.
Seeding by chaos optimization helps avoiding local minima.
Different one-dimensional maps effectiveness is shown on different geometries.
The logistic map is shown to be the best map for seeding Levenberg–Marquardt algorithm.