Pseudo Random Number Generator

PRNG

- splitmix64
- xoshiro256+
- xoshiro256++
- xoshiro256**

splitmix64.h

#ifndef SPLITMIX64_H
#define SPLITMIX64_H

/*  Written in 2015 by Sebastiano Vigna ([email protected])

To the extent possible under law, the author has dedicated all copyright
and related and neighboring rights to this software to the public domain
worldwide. This software is distributed without any warranty.

See <http://creativecommons.org/publicdomain/zero/1.0/>. */

#include <stdint.h>

/* This is a fixed-increment version of Java 8's SplittableRandom generator
   See http://dx.doi.org/10.1145/2714064.2660195 and
   http://docs.oracle.com/javase/8/docs/api/java/util/SplittableRandom.html

   It is a very fast generator passing BigCrush, and it can be useful if
   for some reason you absolutely want 64 bits of state. */

#ifdef __cplusplus
extern "C" {
#endif

// The state can be seeded with any value.
static inline uint64_t* _splitmix64_x() {
    static uint64_t x;
    return &x;
}

static uint64_t splitmix64_next() {
    uint64_t z = (*_splitmix64_x() += 0x9e3779b97f4a7c15);
    z = (z ^ (z >> 30)) * 0xbf58476d1ce4e5b9;
    z = (z ^ (z >> 27)) * 0x94d049bb133111eb;
    return z ^ (z >> 31);
}

static inline void splitmix64_seed(uint64_t seed) {
    *_splitmix64_x() = seed;
}

static inline void splitmix64_shuffle() {
    for (int32_t i = 0; i < 8; ++i)
        splitmix64_next();
}

#ifdef __cplusplus
}
#endif

#endif // SPLITMIX64_H

xoshiro256p.h

#ifndef XOSHIRO256P_H
#define XOSHIRO256P_H

/*  Written in 2018 by David Blackman and Sebastiano Vigna ([email protected])

To the extent possible under law, the author has dedicated all copyright
and related and neighboring rights to this software to the public domain
worldwide. This software is distributed without any warranty.

See <http://creativecommons.org/publicdomain/zero/1.0/>. */

#include <stdint.h>

/* This is xoshiro256+ 1.0, our best and fastest generator for floating-point
   numbers. We suggest to use its upper bits for floating-point
   generation, as it is slightly faster than xoshiro256++/xoshiro256**. It
   passes all tests we are aware of except for the lowest three bits,
   which might fail linearity tests (and just those), so if low linear
   complexity is not considered an issue (as it is usually the case) it
   can be used to generate 64-bit outputs, too.

   We suggest to use a sign test to extract a random Boolean value, and
   right shifts to extract subsets of bits.

   The state must be seeded so that it is not everywhere zero. If you have
   a 64-bit seed, we suggest to seed a splitmix64 generator and use its
   output to fill s. */

#ifdef __cplusplus
extern "C" {
#endif

static inline uint64_t _xoshiro256p_rotl(const uint64_t x, int k) {
	return (x << k) | (x >> (64 - k));
}


static inline uint64_t* _xoshiro256p_s() {
	static uint64_t s[4];
	return s;
}

static uint64_t xoshiro256p_next(void) {
	uint64_t* s = _xoshiro256p_s();

	const uint64_t result = s[0] + s[3];

	const uint64_t t = s[1] << 17;

	s[2] ^= s[0];
	s[3] ^= s[1];
	s[1] ^= s[2];
	s[0] ^= s[3];

	s[2] ^= t;

	s[3] = _xoshiro256p_rotl(s[3], 45);

	return result;
}


/* This is the jump function for the generator. It is equivalent
   to 2^128 calls to next(); it can be used to generate 2^128
   non-overlapping subsequences for parallel computations. */

static void xoshiro256p_jump(void) {
	uint64_t* s = _xoshiro256p_s();

	static const uint64_t JUMP[] = { 0x180ec6d33cfd0aba, 0xd5a61266f0c9392c, 0xa9582618e03fc9aa, 0x39abdc4529b1661c };

	uint64_t s0 = 0;
	uint64_t s1 = 0;
	uint64_t s2 = 0;
	uint64_t s3 = 0;
	for(int i = 0; i < sizeof JUMP / sizeof *JUMP; i++)
		for(int b = 0; b < 64; b++) {
			if (JUMP[i] & UINT64_C(1) << b) {
				s0 ^= s[0];
				s1 ^= s[1];
				s2 ^= s[2];
				s3 ^= s[3];
			}
			xoshiro256p_next();
		}
		
	s[0] = s0;
	s[1] = s1;
	s[2] = s2;
	s[3] = s3;
}


/* This is the long-jump function for the generator. It is equivalent to
   2^192 calls to next(); it can be used to generate 2^64 starting points,
   from each of which jump() will generate 2^64 non-overlapping
   subsequences for parallel distributed computations. */

static void xoshiro256p_long_jump(void) {
	uint64_t* s = _xoshiro256p_s();

	static const uint64_t LONG_JUMP[] = { 0x76e15d3efefdcbbf, 0xc5004e441c522fb3, 0x77710069854ee241, 0x39109bb02acbe635 };

	uint64_t s0 = 0;
	uint64_t s1 = 0;
	uint64_t s2 = 0;
	uint64_t s3 = 0;
	for(int i = 0; i < sizeof LONG_JUMP / sizeof *LONG_JUMP; i++)
		for(int b = 0; b < 64; b++) {
			if (LONG_JUMP[i] & UINT64_C(1) << b) {
				s0 ^= s[0];
				s1 ^= s[1];
				s2 ^= s[2];
				s3 ^= s[3];
			}
			xoshiro256p_next();
		}
		
	s[0] = s0;
	s[1] = s1;
	s[2] = s2;
	s[3] = s3;
}

static inline void xoshiro256p_seed(uint64_t s0, uint64_t s1, uint64_t s2, uint64_t s3) {
	uint64_t* s = _xoshiro256p_s();
	s[0] = s0;
	s[1] = s1;
	s[2] = s2;
	s[3] = s3;
}

#ifdef __cplusplus
}
#endif

#endif // XOSHIRO256P_H

xoshiro256pp.h

#ifndef XOSHIRO256PP_H
#define XOSHIRO256PP_H

/*  Written in 2019 by David Blackman and Sebastiano Vigna ([email protected])

To the extent possible under law, the author has dedicated all copyright
and related and neighboring rights to this software to the public domain
worldwide. This software is distributed without any warranty.

See <http://creativecommons.org/publicdomain/zero/1.0/>. */

#include <stdint.h>

/* This is xoshiro256++ 1.0, one of our all-purpose, rock-solid generators.
   It has excellent (sub-ns) speed, a state (256 bits) that is large
   enough for any parallel application, and it passes all tests we are
   aware of.

   For generating just floating-point numbers, xoshiro256+ is even faster.

   The state must be seeded so that it is not everywhere zero. If you have
   a 64-bit seed, we suggest to seed a splitmix64 generator and use its
   output to fill s. */

#ifdef __cplusplus
extern "C" {
#endif

static inline uint64_t _xoshiro256pp_rotl(const uint64_t x, int k) {
    return (x << k) | (x >> (64 - k));
}


static inline uint64_t* _xoshiro256pp_s() {
    static uint64_t s[4];
    return s;
}

static uint64_t xoshiro256pp_next(void) {
    uint64_t* s = _xoshiro256pp_s();

    const uint64_t result = _xoshiro256pp_rotl(s[0] + s[3], 23) + s[0];

    const uint64_t t = s[1] << 17;

    s[2] ^= s[0];
    s[3] ^= s[1];
    s[1] ^= s[2];
    s[0] ^= s[3];

    s[2] ^= t;

    s[3] = _xoshiro256pp_rotl(s[3], 45);

    return result;
}


/* This is the jump function for the generator. It is equivalent
    to 2^128 calls to next(); it can be used to generate 2^128
    non-overlapping subsequences for parallel computations. */

static void xoshiro256pp_jump(void) {
    uint64_t* s = _xoshiro256pp_s();

    static const uint64_t JUMP[] = {0x180ec6d33cfd0aba, 0xd5a61266f0c9392c, 0xa9582618e03fc9aa, 0x39abdc4529b1661c};

    uint64_t s0 = 0;
    uint64_t s1 = 0;
    uint64_t s2 = 0;
    uint64_t s3 = 0;
    for (int i = 0; i < sizeof JUMP / sizeof * JUMP; i++)
        for (int b = 0; b < 64; b++) {
            if (JUMP[i] & UINT64_C(1) << b) {
                s0 ^= s[0];
                s1 ^= s[1];
                s2 ^= s[2];
                s3 ^= s[3];
            }
            xoshiro256pp_next();
        }

    s[0] = s0;
    s[1] = s1;
    s[2] = s2;
    s[3] = s3;
}



/* This is the long-jump function for the generator. It is equivalent to
    2^192 calls to next(); it can be used to generate 2^64 starting points,
    from each of which jump() will generate 2^64 non-overlapping
    subsequences for parallel distributed computations. */

static void xoshiro256pp_long_jump(void) {
    uint64_t* s = _xoshiro256pp_s();

    static const uint64_t LONG_JUMP[] = {0x76e15d3efefdcbbf, 0xc5004e441c522fb3, 0x77710069854ee241, 0x39109bb02acbe635};

    uint64_t s0 = 0;
    uint64_t s1 = 0;
    uint64_t s2 = 0;
    uint64_t s3 = 0;
    for (int i = 0; i < sizeof LONG_JUMP / sizeof * LONG_JUMP; i++)
        for (int b = 0; b < 64; b++) {
            if (LONG_JUMP[i] & UINT64_C(1) << b) {
                s0 ^= s[0];
                s1 ^= s[1];
                s2 ^= s[2];
                s3 ^= s[3];
            }
            xoshiro256pp_next();
        }

    s[0] = s0;
    s[1] = s1;
    s[2] = s2;
    s[3] = s3;
}

static inline void xoshiro256pp_seed(uint64_t s0, uint64_t s1, uint64_t s2, uint64_t s3) {
    uint64_t* s = _xoshiro256pp_s();
    s[0] = s0;
    s[1] = s1;
    s[2] = s2;
    s[3] = s3;
}

#ifdef __cplusplus
}
#endif

#endif // XOSHIRO256PP_H

xoshiro256ss.h

#ifndef XOSHIRO256SS_H
#define XOSHIRO256SS_H

/*  Written in 2018 by David Blackman and Sebastiano Vigna ([email protected])

To the extent possible under law, the author has dedicated all copyright
and related and neighboring rights to this software to the public domain
worldwide. This software is distributed without any warranty.

See <http://creativecommons.org/publicdomain/zero/1.0/>. */

#include <stdint.h>

/* This is xoshiro256** 1.0, one of our all-purpose, rock-solid
   generators. It has excellent (sub-ns) speed, a state (256 bits) that is
   large enough for any parallel application, and it passes all tests we
   are aware of.

   For generating just floating-point numbers, xoshiro256+ is even faster.

   The state must be seeded so that it is not everywhere zero. If you have
   a 64-bit seed, we suggest to seed a splitmix64 generator and use its
   output to fill s. */

#ifdef __cplusplus
extern "C" {
#endif

static inline uint64_t _xoshiro256ss_rotl(const uint64_t x, int k) {
    return (x << k) | (x >> (64 - k));
}


static inline uint64_t* _xoshiro256ss_s(void) {
    static uint64_t s[4];
    return s;
}

static uint64_t xoshiro256ss_next(void) {
    uint64_t* s = _xoshiro256ss_s();

    const uint64_t result = _xoshiro256ss_rotl(s[1] * 5, 7) * 9;

    const uint64_t t = s[1] << 17;

    s[2] ^= s[0];
    s[3] ^= s[1];
    s[1] ^= s[2];
    s[0] ^= s[3];

    s[2] ^= t;

    s[3] = _xoshiro256ss_rotl(s[3], 45);

    return result;
}


/* This is the jump function for the generator. It is equivalent
    to 2^128 calls to next(); it can be used to generate 2^128
    non-overlapping subsequences for parallel computations. */

static void xoshiro256ss_jump(void) {
    uint64_t* s = _xoshiro256ss_s();

    static const uint64_t JUMP[] = {0x180ec6d33cfd0aba, 0xd5a61266f0c9392c, 0xa9582618e03fc9aa, 0x39abdc4529b1661c};

    uint64_t s0 = 0;
    uint64_t s1 = 0;
    uint64_t s2 = 0;
    uint64_t s3 = 0;
    for (int i = 0; i < sizeof JUMP / sizeof * JUMP; i++)
        for (int b = 0; b < 64; b++) {
            if (JUMP[i] & UINT64_C(1) << b) {
                s0 ^= s[0];
                s1 ^= s[1];
                s2 ^= s[2];
                s3 ^= s[3];
            }
            xoshiro256ss_next();
        }

    s[0] = s0;
    s[1] = s1;
    s[2] = s2;
    s[3] = s3;
}



/* This is the long-jump function for the generator. It is equivalent to
    2^192 calls to next(); it can be used to generate 2^64 starting points,
    from each of which jump() will generate 2^64 non-overlapping
    subsequences for parallel distributed computations. */

static void xoshiro256ss_long_jump(void) {
    uint64_t* s = _xoshiro256ss_s();

    static const uint64_t LONG_JUMP[] = {0x76e15d3efefdcbbf, 0xc5004e441c522fb3, 0x77710069854ee241, 0x39109bb02acbe635};

    uint64_t s0 = 0;
    uint64_t s1 = 0;
    uint64_t s2 = 0;
    uint64_t s3 = 0;
    for (int i = 0; i < sizeof LONG_JUMP / sizeof * LONG_JUMP; i++)
        for (int b = 0; b < 64; b++) {
            if (LONG_JUMP[i] & UINT64_C(1) << b) {
                s0 ^= s[0];
                s1 ^= s[1];
                s2 ^= s[2];
                s3 ^= s[3];
            }
            xoshiro256ss_next();
        }

    s[0] = s0;
    s[1] = s1;
    s[2] = s2;
    s[3] = s3;
}

static inline void xoshiro256ss_seed(uint64_t s0, uint64_t s1, uint64_t s2, uint64_t s3) {
    uint64_t* s = _xoshiro256ss_s();
    s[0] = s0;
    s[1] = s1;
    s[2] = s2;
    s[3] = s3;
}

#ifdef __cplusplus
}
#endif

#endif // XOSHIRO256SS_H_