tensor4all-tensorci

tensor4all-tensorci

Tensor Cross Interpolation algorithms. TCI2 (primary, actively maintained) and TCI1 (legacy).

Key Types

• crossinterpolate2() — main entry point for two-site TCI
• TCI2Options — tolerance, pivot strategy, and convergence settings
• CachedFunction — wrapper that caches function evaluations to avoid redundant calls

Example

use tensor4all_tensorci::{crossinterpolate2, TCI2Options};
use tensor4all_simplett::AbstractTensorTrain;

// Function to interpolate: f(i, j) = (i + j + 1) as f64
let f = |idx: &Vec<usize>| (idx[0] + idx[1] + 1) as f64;

// Run TCI on a 4x4 grid
let (tci, _ranks, errors) = crossinterpolate2::<f64, _, fn(&[Vec<usize>]) -> Vec<f64>>(
    f,
    None,
    vec![4, 4],         // local dimensions
    vec![vec![3, 3]],   // initial pivot (0-indexed); pick where f is large
    TCI2Options {
        tolerance: 1e-10,
        ..Default::default()
    },
).unwrap();

// Verify convergence
assert!(*errors.last().unwrap() < 1e-10);

// Verify interpolation accuracy
let tt = tci.to_tensor_train().unwrap();
let val = tt.evaluate(&[2, 3]).unwrap();  // f(2, 3) = 2 + 3 + 1 = 6.0
assert!((val - 6.0).abs() < 1e-10);

Documentation

• User Guide: TCI
• API Reference