Help language development. Donate to The Perl Foundation

## Random::Choice cpan:TITSUKI last updated on 2022-02-27

Random-Choice-0.0.8/

# NAME

Random::Choice - A Raku alias method implementation

# SYNOPSIS

```use Random::Choice;

say choice(:size(8), :p([0.1, 0.1, 0.1, 0.7])); # (3 1 0 3 3 3 3 3)
say choice(:p([0.1, 0.1, 0.1, 0.7])); # 3
```

# DESCRIPTION

Random::Choice is a Raku alias method implementation. Alias method is an efficient algorithm for sampling from a discrete probability distribution.

## METHODS

### choice

Defined as:

``````multi sub choice(:@p! --> Int) is export
multi sub choice(Int :\$size!, :@p! --> List)
``````

Returns a sample which is an Int value or a List. Where `:@p` is the probabilities associated with each index and `:\$size` is the sample size.

# FAQ

## Is `Random::Choice` faster than Mix.roll?

The answer is YES when you roll a large biased dice or try to roll a dice many times; but NO when a biased dice is small or try to roll a dice few times.

Why? There are some possible reasons:

• `Random::Choice` employs O(N) + O(1) algorithm whereas `Mix.roll` employs O(N) + O(N) algorithm (rakudo 2018.12).

• `Mix.roll` is directly written in nqp. In general, nqp-powered code is faster than naive-Raku-powered code when they take small input.

• Both algorithms take O(N) initialization cost; however, the actual cost of `Mix.roll` is slightly less than `Random::Choice`.

A benchmark result is here (For more info, see `example/bench.p6`):

### The Comparison Table on the Benchmark

```\$ perl6 example/bench.p6
Benchmark:
Timing 1000 iterations of Mix(size=10, @p.elems=10) , Random::Choice(size=10, @p.elems=10)...
Mix(size=10, @p.elems=10) : 0.076 wallclock secs (0.086 usr 0.003 sys 0.089 cpu) @ 13154.606/s (n=1000)
Random::Choice(size=10, @p.elems=10): 0.122 wallclock secs (0.137 usr 0.008 sys 0.145 cpu) @ 8210.383/s (n=1000)
O--------------------------------------O---------O----------------------------O--------------------------------------O
|                                      | Rate    | Mix(size=10, @p.elems=10)  | Random::Choice(size=10, @p.elems=10) |
O======================================O=========O============================O======================================O
| Mix(size=10, @p.elems=10)            | 13155/s | --                         | -42%                                 |
| Random::Choice(size=10, @p.elems=10) | 8210/s  | 73%                        | --                                   |
O--------------------------------------O---------O----------------------------O--------------------------------------O
Benchmark:
Timing 1000 iterations of Mix(size=1000, @p.elems=10) , Random::Choice(size=1000, @p.elems=10)...
Mix(size=1000, @p.elems=10) : 1.879 wallclock secs (1.892 usr 0.000 sys 1.892 cpu) @ 532.130/s (n=1000)
Random::Choice(size=1000, @p.elems=10): 0.097 wallclock secs (0.099 usr 0.002 sys 0.101 cpu) @ 10361.621/s (n=1000)
O----------------------------------------O---------O------------------------------O----------------------------------------O
|                                        | Rate    | Mix(size=1000, @p.elems=10)  | Random::Choice(size=1000, @p.elems=10) |
O========================================O=========O==============================O========================================O
| Mix(size=1000, @p.elems=10)            | 532/s   | --                           | 2141%                                  |
| Random::Choice(size=1000, @p.elems=10) | 10362/s | -96%                         | --                                     |
O----------------------------------------O---------O------------------------------O----------------------------------------O
Benchmark:
Timing 1000 iterations of Mix(size=10, @p.elems=1000) , Random::Choice(size=10, @p.elems=1000)...
Mix(size=10, @p.elems=1000) : 2.576 wallclock secs (2.560 usr 0.020 sys 2.580 cpu) @ 388.182/s (n=1000)
Random::Choice(size=10, @p.elems=1000): 6.010 wallclock secs (6.015 usr 0.032 sys 6.047 cpu) @ 166.398/s (n=1000)
O----------------------------------------O-------O------------------------------O----------------------------------------O
|                                        | Rate  | Mix(size=10, @p.elems=1000)  | Random::Choice(size=10, @p.elems=1000) |
O========================================O=======O==============================O========================================O
| Mix(size=10, @p.elems=1000)            | 388/s | --                           | -57%                                   |
| Random::Choice(size=10, @p.elems=1000) | 166/s | 134%                         | --                                     |
O----------------------------------------O-------O------------------------------O----------------------------------------O
Benchmark:
Timing 1000 iterations of Mix(size=100, @p.elems=100), Random::Choice(size=100, @p.elems=100)...
Mix(size=100, @p.elems=100): 1.505 wallclock secs (1.511 usr 0.000 sys 1.511 cpu) @ 664.420/s (n=1000)
Random::Choice(size=100, @p.elems=100): 0.619 wallclock secs (0.624 usr 0.000 sys 0.624 cpu) @ 1616.535/s (n=1000)
O----------------------------------------O--------O-----------------------------O----------------------------------------O
|                                        | Rate   | Mix(size=100, @p.elems=100) | Random::Choice(size=100, @p.elems=100) |
O========================================O========O=============================O========================================O
| Mix(size=100, @p.elems=100)            | 664/s  | --                          | 146%                                   |
| Random::Choice(size=100, @p.elems=100) | 1617/s | -59%                        | --                                     |
O----------------------------------------O--------O-----------------------------O----------------------------------------O
Benchmark:
Timing 1000 iterations of Mix(size=1000, @p.elems=1000), Random::Choice(size=1000, @p.elems=1000)...
Mix(size=1000, @p.elems=1000): 135.720 wallclock secs (135.946 usr 0.288 sys 136.234 cpu) @ 7.368/s (n=1000)
Random::Choice(size=1000, @p.elems=1000): 6.022 wallclock secs (6.031 usr 0.028 sys 6.058 cpu) @ 166.058/s (n=1000)
O------------------------------------------O--------O-------------------------------O------------------------------------------O
|                                          | Rate   | Mix(size=1000, @p.elems=1000) | Random::Choice(size=1000, @p.elems=1000) |
O==========================================O========O===============================O==========================================O
| Mix(size=1000, @p.elems=1000)            | 7.37/s | --                            | 2158%                                    |
| Random::Choice(size=1000, @p.elems=1000) | 166/s  | -96%                          | --                                       |
O------------------------------------------O--------O-------------------------------O------------------------------------------O
```

### The Environment on the Benchmark

• `CPU` Ryzen7 5800X (8core)

• `OS` Debian11 bullseye

# AUTHOR

titsuki [email protected]