weyl

README

Middle Square Weyl Sequence RNG for Go

A new implementation of John von Neumann's middle square random number generator (RNG). A Weyl sequence is utilized to keep the generator running through a long period.

Paper

Using

To start using Weyl, install Go and run go get:

$ go get -u github.com/tidwall/weyl

Create a new RNG

rng := rand.New(weyl.NewSource(time.Now().UnixNano()))

Use the RNG just as you would use the standard math/rand.

rng.Int63()  // random int64
rng.Float64() // random float64

Performance

Weyl RNG:

BenchmarkInt63Threadsafe-8     	100000000	        16.3 ns/op
BenchmarkInt63Unthreadsafe-8   	300000000	         5.32 ns/op
BenchmarkIntn1000-8            	100000000	        16.0 ns/op
BenchmarkInt63n1000-8          	50000000	        25.7 ns/op
BenchmarkInt31n1000-8          	100000000	        14.6 ns/op
BenchmarkFloat32-8             	100000000	        11.9 ns/op
BenchmarkFloat64-8             	200000000	         8.44 ns/op
BenchmarkPerm3-8               	20000000	        78.0 ns/op
BenchmarkPerm30-8              	 2000000	       600 ns/op
BenchmarkRead3-8               	100000000	        14.8 ns/op
BenchmarkRead64-8              	10000000	       135 ns/op
BenchmarkRead1000-8            	 1000000	      1801 ns/op

Go standard math/rand:

BenchmarkInt63Threadsafe-8     	100000000	        21.4 ns/op
BenchmarkInt63Unthreadsafe-8   	200000000	         6.74 ns/op
BenchmarkIntn1000-8            	100000000	        17.0 ns/op
BenchmarkInt63n1000-8          	50000000	        26.4 ns/op
BenchmarkInt31n1000-8          	100000000	        14.4 ns/op
BenchmarkFloat32-8             	100000000	        12.9 ns/op
BenchmarkFloat64-8             	200000000	         9.77 ns/op
BenchmarkPerm3-8               	20000000	        80.6 ns/op
BenchmarkPerm30-8              	 2000000	       658 ns/op
BenchmarkRead3-8               	100000000	        14.6 ns/op
BenchmarkRead64-8              	10000000	       147 ns/op
BenchmarkRead1000-8            	 1000000	      2112 ns/op

Contact

Josh Baker @tidwall

License

Weyl source code is available under the ISC License.

Documentation

Index

• func Int63() int64
• func NewSource(seed int64) rand.Source
• func NewSource64(seed int64) rand.Source64
• type Generator
  • func (w *Generator) Float64() float64
  • func (w *Generator) Int63() int64
  • func (w *Generator) Seed(seed int64)
  • func (w *Generator) Uint64() uint64

Functions

func Int63

func Int63() int64

Int63 returns a random int64

func NewSource

func NewSource(seed int64) rand.Source

NewSource returns a new rand.Source for used with rand.New()

func NewSource64

func NewSource64(seed int64) rand.Source64

NewSource64 returns a new rand.Source64

Types

type Generator

type Generator struct {
	// contains filtered or unexported fields
}

Generator represents an random number generator

func (*Generator) Float64

func (w *Generator) Float64() float64

Float64 returns a random float64

func (*Generator) Int63

func (w *Generator) Int63() int64

Int63 returns a random Int63 returns a random int64 that is always a positive number.

func (*Generator) Seed

func (w *Generator) Seed(seed int64)

Seed or re-seed the generator

func (*Generator) Uint64

func (w *Generator) Uint64() uint64

Uint64 returns a random uint64