Struct Variable 

pub struct Variable<M: LieGroup> {
    pub value: M,
    pub fixed_indices: HashSet<usize>,
    pub bounds: HashMap<usize, (f64, f64)>,
    pub covariance: Option<Mat<f64>>,
}

Generic Variable struct that uses static dispatch with any manifold type.

This struct represents optimization variables that live on manifolds and provides type-safe operations for updating variables with tangent space perturbations.

Type Parameters

• M - The manifold type that implements the LieGroup trait

Examples

use apex_solver::core::variable::Variable;
use apex_solver::manifold::se2::SE2;
use apex_solver::manifold::rn::Rn;

// Create a Variable for SE2 manifold
let se2_value = SE2::from_xy_angle(1.0, 2.0, 0.5);
let se2_var = Variable::new(se2_value);

// Create a Variable for Euclidean space
let rn_value = Rn::from_vec(vec![1.0, 2.0, 3.0]);
let rn_var = Variable::new(rn_value);

Fields

value: M

The manifold value

fixed_indices: HashSet<usize>

Indices that should remain fixed during optimization

bounds: HashMap<usize, (f64, f64)>

Bounds constraints on the tangent space representation

covariance: Option<Mat<f64>>

Covariance matrix in the tangent space (uncertainty estimation)

This is None if covariance has not been computed. When present, it’s a square matrix of size tangent_dim x tangent_dim representing the uncertainty in the optimized variable’s tangent space.

For example, for SE3 this would be a 6×6 matrix representing uncertainty in [translation_x, translation_y, translation_z, rotation_x, rotation_y, rotation_z].

Implementations

impl<M> Variable<M>

where M: LieGroup + Clone + 'static, M::TangentVector: Tangent<M>,

pub fn new(value: M) -> Self

Create a new Variable from a manifold value.

Arguments

• value - The initial manifold value

Examples

use apex_solver::core::variable::Variable;
use apex_solver::manifold::se2::SE2;

let se2_value = SE2::from_xy_angle(1.0, 2.0, 0.5);
let variable = Variable::new(se2_value);

pub fn set_value(&mut self, value: M)

Set the manifold value.

Arguments

• value - The new manifold value

pub fn get_size(&self) -> usize

Get the degrees of freedom (tangent space dimension) of the variable.

This returns the dimension of the tangent space, which is the number of parameters that can be optimized for this manifold type.

Returns

The tangent space dimension (degrees of freedom)

pub fn plus(&self, tangent: &M::TangentVector) -> M

Plus operation: apply tangent space perturbation to the manifold value.

This method takes a tangent vector and returns a new manifold value by applying the manifold’s plus operation (typically the exponential map).

Arguments

• tangent - The tangent vector to apply as a perturbation

Returns

A new manifold value after applying the tangent perturbation

Examples

use apex_solver::core::variable::Variable;
use apex_solver::manifold::se2::{SE2, SE2Tangent};
use nalgebra as na;

let se2_value = SE2::from_xy_angle(1.0, 2.0, 0.0);
let variable = Variable::new(se2_value);

// Create a tangent vector: [dx, dy, dtheta]
let tangent = SE2Tangent::from(na::DVector::from(vec![0.1, 0.1, 0.1]));
let new_value = variable.plus(&tangent);

pub fn minus(&self, other: &Self) -> M::TangentVector

Minus operation: compute tangent space difference between two manifold values.

This method computes the tangent vector that would transform this variable’s value to the other variable’s value using the manifold’s minus operation (typically the logarithmic map).

Arguments

• other - The other variable to compute the difference to

Returns

A tangent vector representing the difference in tangent space

Examples

use apex_solver::core::variable::Variable;
use apex_solver::manifold::se2::SE2;

let se2_1 = SE2::from_xy_angle(2.0, 3.0, 0.5);
let se2_2 = SE2::from_xy_angle(1.0, 2.0, 0.0);
let var1 = Variable::new(se2_1);
let var2 = Variable::new(se2_2);

let difference = var1.minus(&var2);

pub fn get_covariance(&self) -> Option<&Mat<f64>>

Get the covariance matrix for this variable (if computed).

Returns None if covariance has not been computed.

Returns

Reference to the covariance matrix in tangent space

pub fn set_covariance(&mut self, cov: Mat<f64>)

Set the covariance matrix for this variable.

The covariance matrix should be square with dimension equal to the tangent space dimension of this variable.

Arguments

• cov - Covariance matrix in tangent space

pub fn clear_covariance(&mut self)

Clear the covariance matrix.

impl Variable<Rn>

pub fn to_vector(&self) -> DVector<f64>

Convert the Rn variable to a vector representation.

pub fn from_vector(values: DVector<f64>) -> Self

Create an Rn variable from a vector representation.

pub fn update_variable(&mut self, tangent_delta: DVector<f64>)

Update the Rn variable with bounds and fixed constraints.

Trait Implementations

impl<M: Clone + LieGroup> Clone for Variable<M>

fn clone(&self) -> Variable<M>

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<M: Debug + LieGroup> Debug for Variable<M>

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

Formats the value using the given formatter.