Enum Merit 

pub enum Merit {
    L1,
    AugmentedLagrangian,
}

Merit function of the line search.

Variants

L1

f + Σ μᵢ·violationᵢ with Powell’s per-row penalties μᵢ ← max(1.1|λᵢ|, ½(μᵢ + 1.1|λᵢ|)) (default).

AugmentedLagrangian

SNOPT’s augmented Lagrangian f − πᵀ(c − s) + ½Σρᵢ(cᵢ − sᵢ)², searched jointly in the variables, the multiplier estimate π and the slacks s (Gill, Murray and Saunders 2005, §2.7). The penalties ρ are the smallest for which the linear model of the merit falls by ½pᵀBp over the unit step, twice the paper’s slope rule, and may decrease under its damping rule. It is differentiable, so it needs no second-order correction; the combination is rejected. Steps whose elastic QP leaves rows violated use the ℓ1 merit.

Implementations

impl Merit

pub fn as_str(self) -> &'static str

Stable lowercase name, as used by the Python binding.

pub fn parse(name: &str) -> Result<Self>

Parses Self::as_str.

Trait Implementations

impl Clone for Merit

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for Merit

impl Debug for Merit

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for Merit

fn default() -> Self

Returns the “default value” for a type.

impl Eq for Merit

impl PartialEq for Merit

fn eq(&self, other: &Self) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for Merit