Struct TensorTrain 

pub struct TensorTrain {
    pub cores: Vec<Array3<f64>>,
    pub shape: Vec<usize>,
    pub ranks: Vec<usize>,
}

Tensor-train representation of a d-dimensional array.

A[i1,...,id] ≈ G1[i1] * G2[i2] * ... * Gd[id]

where each core Gk has shape [r_{k-1}, n_k, r_k] and the boundary ranks satisfy r_0 = r_d = 1.

Fields

cores: Vec<Array3<f64>>

Cores: cores[k] has shape [r_{k-1}, n_k, r_k].

shape: Vec<usize>

Mode sizes n_1, ..., n_d.

ranks: Vec<usize>

Ranks r_0=1, r_1, ..., r_{d-1}, r_d=1.

Implementations

impl TensorTrain

pub fn new(cores: Vec<Array3<f64>>) -> Result<Self, InterpolateError>

Create a TT from pre-built cores.

Validates that shapes are consistent.

pub fn eval(&self, idx: &[usize]) -> Result<f64, InterpolateError>

Evaluate the TT at a multi-index idx = [i1, ..., id].

Returns the scalar G1[:,i1,:] * G2[:,i2,:] * ... * Gd[:,id,:] contracted to a single f64.

pub fn to_dense(&self) -> Result<ArrayD<f64>, InterpolateError>

Reconstruct the full dense tensor.

Warning: exponential cost in d — only practical for small tensors.

pub fn norm(&self) -> f64

Compute the Frobenius norm of the TT tensor via a transfer-matrix approach.

pub fn n_params(&self) -> usize

Number of parameters (stored f64 elements) in the TT format.

pub fn from_dense( tensor: &ArrayD<f64>, max_rank: usize, tol: f64, ) -> Result<Self, InterpolateError>

Convenience: build from a dense tensor with TT-SVD.

Trait Implementations

impl Clone for TensorTrain

fn clone(&self) -> TensorTrain

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TensorTrain

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

Formats the value using the given formatter.