Enum argmin::gradientdescent::GDGammaUpdate

pub enum GDGammaUpdate<'a> {
    Constant(f64),
    BarzilaiBorwein,
    BacktrackingLineSearch(BacktrackingLineSearch<'a>),
}

Gradient Descent gamma update method

Variants

Constant(f64)

Constant gamma

BarzilaiBorwein

Gamma updated according to TODO Apparently this only works if the cost function is convex and the derivative of the cost function is Lipschitz. TODO: More detailed description (formula)

BacktrackingLineSearch(BacktrackingLineSearch<'a>)

Backtracking line search