Struct OnlineKernelMatrix 

pub struct OnlineKernelMatrix { /* private fields */ }

Incrementally updatable kernel matrix.

Supports efficient O(n) updates when adding new samples, avoiding O(n²) recomputation of the entire matrix.

Example

use tensorlogic_sklears_kernels::{OnlineKernelMatrix, RbfKernel, RbfKernelConfig, Kernel};

let kernel = RbfKernel::new(RbfKernelConfig::new(0.5)).unwrap();
let mut online = OnlineKernelMatrix::new(Box::new(kernel));

// Add samples incrementally
online.add_sample(vec![1.0, 2.0, 3.0]).unwrap();
online.add_sample(vec![4.0, 5.0, 6.0]).unwrap();
online.add_sample(vec![7.0, 8.0, 9.0]).unwrap();

// Get the kernel matrix
let matrix = online.get_matrix();
assert_eq!(matrix.len(), 3);

Implementations

impl OnlineKernelMatrix

pub fn new(kernel: Box<dyn Kernel>) -> Self

Create a new online kernel matrix.

pub fn with_config(kernel: Box<dyn Kernel>, config: OnlineConfig) -> Self

Create with custom configuration.

pub fn add_sample(&mut self, sample: Vec<f64>) -> Result<()>

Add a new sample and update the kernel matrix.

Time complexity: O(n) where n is current number of samples.

pub fn add_samples(&mut self, samples: Vec<Vec<f64>>) -> Result<()>

Add multiple samples at once (batch update).

More efficient than adding one by one due to better cache utilization.

pub fn remove_sample(&mut self, index: usize) -> Result<Vec<f64>>

Remove a sample by index and update the matrix.

Time complexity: O(n²) due to matrix restructuring.

pub fn get_matrix(&self) -> &Vec<Vec<f64>>

Get the current kernel matrix.

pub fn get_samples(&self) -> &Vec<Vec<f64>>

Get the current samples.

pub fn get(&self, i: usize, j: usize) -> Option<f64>

Get a specific kernel value.

pub fn len(&self) -> usize

Get the number of samples.

pub fn is_empty(&self) -> bool

Check if empty.

pub fn stats(&self) -> &OnlineStats

Get statistics.

pub fn clear(&mut self)

Reset the matrix.

pub fn kernel(&self) -> &dyn Kernel

Get the underlying kernel.

pub fn config(&self) -> &OnlineConfig

Get the configuration.

pub fn compute_with_sample( &self, query: &[f64], sample_idx: usize, ) -> Result<f64>

Compute kernel value between a query point and stored sample.

pub fn compute_with_all(&self, query: &[f64]) -> Result<Vec<f64>>

Compute kernel values between query and all stored samples.

pub fn to_matrix(&self) -> Vec<Vec<f64>>

Clone the current matrix as a standalone 2D vector.