Given a direction, perform a line search to find a direction to descend.
Given a direction, perform a line search to find a direction to descend. At the moment, this just executes backtracking, so it does not fulfill the wolfe conditions.
the current state
The objective
The step direction
stepSize
Implements the Orthant-wise Limited Memory QuasiNewton method, which is a variant of LBFGS that handles L1 regularization.
Paper is Andrew and Gao (2007) Scalable Training of L1-Regularized Log-Linear Models