numir.random

RNG

Declaration

class RNG;

get

Declaration

static @property ref auto get();
setSeed

Declaration

static void setSeed(uint seed);

generate

Declaration

auto generate(V, size_t N)(V var, size_t[N] length...);

general function for random slice generation with global RNG

normal

Declaration

auto normal(E = double, size_t N)(size_t[N] length...);

uniform

Declaration

auto uniform(E = double, size_t N)(size_t[N] length...);

Examples

      import mir.ndslice : all;
import std.algorithm : sum;
import mir.random.variable : BernoulliVariable;
auto bs = BernoulliVariable!double(0.25).generate(100).sum;
assert(0 < bs && bs < 50, "maybe fail");

// pre-defined random variables (normal and uniform)
RNG.setSeed(1);
auto r0 = normal(3, 4).slice;
assert(r0.shape == [3, 4]);
RNG.setSeed(0);
auto r1 = normal(3, 4).slice;
assert(r0 != r1);

RNG.setSeed(0);
auto r2 = normal(3, 4).slice;
assert(r1 == r2);

auto u = uniform(3, 4).slice;
assert(u.shape == [3, 4]);
assert(u.all!(a => (0 <= a && a < 1)));

    

permutation

Declaration

auto permutation(T...)(T t);

generate a sequence as same as numir.core.arange but shuffled

Examples

      import numir : arange;
import mir.ndslice.sorting : sort;
import numir.testing : approxEqual;
import std.stdio;
auto ps1 = permutation(100);
auto ps2 = permutation(100);
assert(ps1 != ps2, "maybe fail at 1%");
assert(ps1.sort() == arange(100));

auto ps3 = permutation(1, 10, 0.1);
auto ps4 = permutation(1, 10, 0.1);
assert(ps3 != ps4);
assert(ps4.sort.approxEqual(arange(1, 10, 0.1)));

    