Pcg64Dxsm Class

public sealed class Pcg64Dxsm : RandomSource<ScalarGenerator64<Pcg64DxsmEngine>>, 
	IRandomSourceFactory<Pcg64Dxsm>, IAdvanceable, IJumpable

Public NotInheritable Class Pcg64Dxsm
	Inherits RandomSource(Of ScalarGenerator64(Of Pcg64DxsmEngine))
	Implements IRandomSourceFactory(Of Pcg64Dxsm), IAdvanceable, 
	IJumpable

public ref class Pcg64Dxsm sealed : public RandomSource<ScalarGenerator64<Pcg64DxsmEngine>>, 
	IRandomSourceFactory<Pcg64Dxsm^>, IAdvanceable, IJumpable

[<SealedAttribute>]
type Pcg64Dxsm = 
    class
        inherit RandomSource<ScalarGenerator64<Pcg64DxsmEngine>>
        interface IRandomSourceFactory<Pcg64Dxsm>
        interface IAdvanceable
        interface IJumpable
    end

Inheritance

Object  →  RandomSource<ScalarGenerator64<Pcg64DxsmEngine>>  →  Pcg64Dxsm

Implements

IAdvanceable, IJumpable, IRandomSourceFactory<Pcg64Dxsm>

Use the Pcg64Dxsm class to represent a pseudo-random number generator that uses the PCG64 DXSM (Double XorShift Multiply) algorithm by Melissa O'Neill.

This generator features a 128-bit state with a period of 2¹²⁸. It passes the strictest statistical tests including BigCrush and PractRand, and provides excellent statistical properties and performance.

Algorithm details:

• LCG: 128-bit state
• Multiplier: 0xDA942042E4DD58B5 (64-bit "cheap" multiplier)
• Output function: DXSM (Double Xorshift Multiply)
• Output ordering: output-then-advance (outputs from state before advancing)

The DXSM variant provides the best overall statistical quality of the PCG family. It uses a 64-bit "cheap" multiplier which provides excellent performance while the DXSM output function ensures superior statistical properties.

Use case: This is the recommended PCG64 variant for new code requiring the highest statistical quality. It matches NumPy 1.18+ PCG64DXSM implementation and is the default generator in NumPy's modern random API.

Compatibility note: This generator will produce different sequences than Pcg64 even with the same seed, due to the different multiplier and output function.

Reference: Melissa E. O'Neill. "PCG: A Family of Simple Fast Space-Efficient Statistically Good Algorithms for Random Number Generation." Harvey Mudd College, 2014.

See also: https://numpy.org/doc/stable/reference/random/bit_generators/pcg64dxsm.html

Reference