Module samplers

Samplers are containers for sampling data or convergence trajectories; or recording epochs/iterations for later tabulation, plotting or analysis.

Container	Description
SampleVec	a reservoir sampler for periodic sampling
SampleDeque	a reservoir sampler with guaranteed front and back minimum samples. Ideal for plots where the initial and final points are important to capture
ProgressBar	prints a progress bar
Tabulate	displays a simple table using println style formatting

Example

use stepwise::{VectorExt as _, Driver, fixed_iters, Step, samplers::SampleDeque};

// create a container with 10 initial, and 10 final samples,
// and 500 samples evenly taken from the 1,000,000 iterations
// this allows minimal memory use, and fast plotting
let mut sampler = SampleDeque::with_sizes(10, 500, 10);

type Point = (usize, Vec<f64>);  // (iteration number, X-vector)

let (_solved, _step) = fixed_iters(gradient_descent, 1_000_000)
    .on_step(|algo, step| {
        let point: Point = (step.iteration(), algo.x().to_vec());
        sampler.sample(point);
   })
   .solve()
   .unwrap();


// plot_data will have 500 evenly spaced samples, with the first and
// last 10 items guaranteed to be captured
let plot_data : Vec<Point> = sampler.into_unordered_iter().collect();

Structs

ProgressBar

A progress bar which can display to Stderr or Stdout

SampleDeque

A segmented vector that divides elements into three regions: head, middle, and tail.

SampleVec

Tabulate

A Tabulator or table generator which can display to Stderr or Stdout

Enums

SamplingOutcome

Indicates the action taken by sampling.

Traits

Sampler

Sampler trait