Module optimizer 

Optimization solvers for nonlinear least squares problems.

This module provides various optimization algorithms specifically designed for nonlinear least squares problems commonly found in computer vision:

• Levenberg-Marquardt algorithm
• Gauss-Newton algorithm
• Dog Leg algorithm

Re-exports

pub use dog_leg::DogLeg;

pub use gauss_newton::GaussNewton;

pub use levenberg_marquardt::LevenbergMarquardt;

pub use crate::observers::OptObserver;

pub use crate::observers::OptObserverVec;

Modules

dog_leg

Dog Leg trust region optimization algorithm implementation.

gauss_newton

Gauss-Newton optimization algorithm implementation.

levenberg_marquardt

Levenberg-Marquardt algorithm implementation.

Structs

ConvergenceInfo

Detailed convergence information.

ConvergenceParams

Parameters for convergence checking, shared across optimizers.

InitializedState

Result of optimization state initialization, shared by all optimizers.

IterationStats

Per-iteration statistics for detailed logging (Ceres-style output).

OptimizerSummary

Unified summary statistics for all optimizer types.

SolverResult

Result of a solver execution.

Enums

OptimizationStatus

Status of an optimization process

OptimizerError

Optimizer-specific error types for apex-solver

OptimizerType

Type of optimization solver algorithm to use

Traits

Solver

Core trait for optimization solvers.

Functions

apply_negative_parameter_step

Apply negative parameter step to rollback variables.

apply_parameter_step

Apply parameter update step to all variables.

build_solver_result

Build a SolverResult from common optimization loop outputs.

check_convergence

Check convergence criteria common to all optimizers.

compute_cost

compute_parameter_norm

Compute total parameter vector norm ||x|| across all variables.

compute_step_quality

Compute step quality ratio (actual vs predicted reduction).

create_jacobi_scaling

Create Jacobi scaling diagonal matrix from Jacobian column norms.

create_linear_solver

Create the appropriate linear solver based on configuration.

create_optimizer_summary

Create an OptimizerSummary from common optimization loop outputs.

initialize_optimization_state

Initialize optimization state from problem and initial parameters.

notify_observers

Notify observers with current optimization state.

process_jacobian

Process Jacobian by applying Jacobi scaling (created on first iteration).

Type Aliases

OptimizerResult

Result type for optimizer operations