polydat 0.1.0

Polydat — generation kernel for deterministic variate generation in nb-rs (formerly nbrs-variates)
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33

// independent_samples.gk — Library kernel for independent distribution samples.
//
// A common need: generate multiple statistically independent fields
// from a single coordinate. This kernel takes one input and produces
// N independent quantiles by chaining hashes.

// Formal interface: one input, three independent quantiles.
(seed: u64) -> (q0: f64, q1: f64, q2: f64) := {
    // Each hash step produces an independent-seeming value.
    h0 := hash(seed)
    h1 := hash(h0)
    h2 := hash(h1)

    // Normalize each to [0, 1) independently.
    q0 := unit_interval(h0)
    q1 := unit_interval(h1)
    q2 := unit_interval(h2)
}

// Usage in another kernel:
//
//   init temp_lut = dist_normal(72.0, 5.0)
//   init wait_lut = dist_exponential(0.5)
//   init size_lut = dist_pareto(1.0, 2.0)
//
//   input cycle: u64
//   (q_temp, q_wait, q_size) := independent_samples(cycle)
//   temperature := lut_sample(q_temp, temp_lut)
//   wait_time := lut_sample(q_wait, wait_lut)
//   file_size := lut_sample(q_size, size_lut)
//
// Now temperature, wait_time, and file_size are statistically
// independent because they derive from different hash depths.