Struct ProximalPointSolver 

pub struct ProximalPointSolver {
    pub lambda: f64,
    pub max_iter: usize,
    pub tol: f64,
}

Proximal point algorithm solver for minimising a proper lower-semicontinuous function.

Fields

lambda: f64

Regularisation parameter λ > 0.

max_iter: usize

Maximum number of iterations.

tol: f64

Convergence tolerance.

Implementations

impl ProximalPointSolver

pub fn new(lambda: f64, max_iter: usize, tol: f64) -> Self

Create a new proximal point solver.

pub fn prox_step(&self, f: impl Fn(&[f64]) -> f64 + Copy, v: &[f64]) -> Vec<f64>

Approximate the proximal operator prox_{λf}(v) via gradient descent on f(x) + ‖x-v‖²/(2λ).

pub fn solve( &self, f: impl Fn(&[f64]) -> f64 + Copy, x0: &[f64], ) -> Vec<Vec<f64>>

Run the proximal point algorithm starting from x0. Returns the sequence of iterates.

pub fn has_converged(&self, iterates: &[Vec<f64>]) -> bool

Check convergence: last step size is smaller than tolerance.