BackTracking

class kona.algorithms.util.linesearch.BackTracking(optns={}, out_file=<open file '<stdout>', mode 'w'>)[source]

Bases: kona.algorithms.util.linesearch.LineSearch

Back-tracking line search.

Variables:
  • alpha_init (float) – Initial step size.
  • alpha_min (float) – Minimum step size.
  • rdtn_factor (float) – Reduction factor for the step size at each iteration.
  • p_dot_dfdx (float) – Value of \(\langle p, \nabla f \rangle\) at current step.
find_step_length(merit)[source]