Struct QuantumSVMModel 

pub struct QuantumSVMModel { /* private fields */ }

Quantum-Enhanced Support Vector Machine

Implements a kernel SVM where the kernel is computed via a quantum-inspired random Fourier feature map (Rahimi-Recht approximation of the RBF kernel). The feature map φ(x) = √(2/D) · [cos(ω₁ᵀx + b₁), …, cos(ω_Dᵀx + b_D)]ᵀ approximates exp(-‖x - z‖² / (2σ²)) for random ωᵢ ~ N(0, I/σ²) and bᵢ ~ Uniform[0, 2π].

Training Algorithm

Binary classification is performed via a coordinate-descent dual SVM solver (simplified SMO) on the quantum kernel matrix, respecting the box constraint 0 ≤ αᵢ ≤ C for all support vectors.

Example

use scirs2_core::ndarray::{Array1, Array2};
use scirs2_spatial::quantum_inspired::algorithms::quantum_machine_learning::QuantumSVMModel;

// Two-class separable dataset (4 points, 2 features)
let x = Array2::from_shape_vec((4, 2), vec![
    0.0, 0.0,
    1.0, 0.0,
    0.0, 1.0,
    5.0, 5.0,
])?;
let y = Array1::from_vec(vec![1.0, 1.0, 1.0, -1.0]);

let mut model = QuantumSVMModel::new(4, 1.0);
model.fit(&x, &y)?;

let x_test = Array2::from_shape_vec((1, 2), vec![0.5, 0.5])?;
let preds = model.predict(&x_test)?;
assert_eq!(preds.len(), 1);

Implementations

impl QuantumSVMModel

pub fn new(n_qubits: usize, regularization: f64) -> Self

Construct a new QuantumSVMModel.

Arguments

• n_qubits - Number of qubits; the random Fourier feature dimension is D = 2^n_qubits (clamped to [4, 256] for practical use).
• regularization - The SVM regularisation constant C > 0.

Panics

Does not panic — invalid parameters are caught in [fit].

pub fn n_qubits(&self) -> usize

Return the number of qubits controlling the feature map dimension.

pub fn regularization(&self) -> f64

Return the regularisation parameter C.

pub fn num_support_vectors(&self) -> usize

Return the number of support vectors (0 before training).

pub fn fit(&mut self, x: &Array2<f64>, y: &Array1<f64>) -> SpatialResult<()>

Train the model on labelled data.

Arguments

• x - Training features matrix of shape (n_samples, n_features).
• y - Binary labels {+1, -1} of shape (n_samples,).

Errors

Returns SpatialError::InvalidInput for empty inputs, shape mismatch, invalid label values, or non-positive regularization.

pub fn predict(&self, x: &Array2<f64>) -> SpatialResult<Array1<f64>>

Predict class labels for new samples.

Arguments

• x - Test features matrix of shape (n_test, n_features).

Returns

Binary predictions {+1, -1} of shape (n_test,).

Errors

Returns error if the model has not been fitted or if x has a different number of features than the training data.

Trait Implementations

impl Clone for QuantumSVMModel

fn clone(&self) -> QuantumSVMModel

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for QuantumSVMModel

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

Formats the value using the given formatter.