`rng`

Controls the seed and generator of the global random number stream.

Category: Mathematics > Random Number Generation

Syntax

rng(seed)

Seeds the Mersenne Twister generator with a nonnegative integer.

rng('default')

Restores the startup settings: Mersenne Twister, seed 0.

rng('shuffle')

Seeds from the current time, giving a different sequence each run.

rng(seed, generator)

Chooses the generator, for example 'twister'.

s = rng

Returns the current settings as a struct, to restore later with rng(s).

Description

rng sets the state of the global stream, which is what rand, randn, randi and randperm use. Seeding with the same value before a computation makes it reproducible: it is the first line of most scripts that use random data.

To run a piece of code and then go back to the previous random state, save s = rng first and call rng(s) afterward.

Input arguments

seed: Nonnegative integer, 'default' or 'shuffle'.
generator: Generator name: 'twister' (or 'mt19937ar').
s: Struct with fields Type, Seed and State, as returned by rng.

Output arguments

s: The settings in effect before the call.

Example

Reproducible numbers and saved state

rng(42);
a = rand;
rng(42);
b = rand;
a == b
s = rng;
x = rand(1, 2);
rng(s);
isequal(x, rand(1, 2))
rng('default');
rand

Output

ans = logical
   1
ans = logical
   1
ans = 0.8147

Notes

Only the Mersenne Twister generator ('twister') is available. The rand values for a given seed match MATLAB's; randn, randi and randperm may not.

rng