Struct GraphSeparableModel 

pub struct GraphSeparableModel {
    pub alpha_keys: Vec<String>,
    /* private fields */
}

Implements varpro’s SeparableNonlinearModel trait for an arbitrary separable spectrafit graph.

Fields

alpha_keys: Vec<String>

Names of alpha keys, for mapping results back.

Implementations

impl GraphSeparableModel

pub fn new( graph: &FitGraphSpec, dataset: &MeasurementSpec, all_params: &HashMap<String, ParameterSpec>, ) -> Result<Self, ModelError>

Build from a compiled graph, one dataset, and the full parameter map.

all_params maps "node_id.param_name" → ParameterSpec.

Errors

Returns [ModelError::ParameterNotInModel] when graph fails to compile, or [ModelError::IncorrectParameterCount] if the initial set_params call (used to prime the Φ cache) receives a parameter vector whose length does not match the model’s own count.

Trait Implementations

impl Debug for GraphSeparableModel

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

Formats the value using the given formatter.

impl SeparableNonlinearModel for GraphSeparableModel

type ScalarType = f64

The scalar number type for this model, which should be a real or complex number type, commonly either f64 or f32.

type Error = ModelError

The associated error type that can occur when the model or the derivative is evaluated. If this model does not need (or for performance reasons does not want) to return an error, it is possible to specify std::convert::Infallible as the associated Error type.

fn parameter_count(&self) -> usize

Returns the number of nonlinear parameters that this model depends on.

fn base_function_count(&self) -> usize

Returns the number of basis functions that this model depends on.

fn output_len(&self) -> usize

Returns the dimension $n$ of the output of the model $\vec{f}(\vec{x},\vec{\alpha},\vec{c}) \in \mathbb{R}^n$. This is also the dimension of every single basis function.

fn set_params( &mut self, parameters: OVector<f64, Dyn>, ) -> Result<(), Self::Error>

Set the nonlinear parameters $\vec{\alpha}$ of the model to the given vector .

fn params(&self) -> OVector<f64, Dyn>

Get the currently set nonlinear parameters of the model, i.e. the vector $\vec{\alpha}$.

fn eval(&self) -> Result<OMatrix<f64, Dyn, Dyn>, Self::Error>

Evaluate the basis functions of the model at the currently set parameters $\vec{\alpha}$ and return them in matrix form. The columns of this matrix are the evaluated basis functions. See below for a detailed explanation.

fn eval_partial_deriv( &self, derivative_index: usize, ) -> Result<OMatrix<f64, Dyn, Dyn>, Self::Error>

Evaluate the partial derivatives for the basis functions at the currently set parameters and return them in matrix form.