rge256-core 1.0.7

RGE-256: ARX-based pseudorandom number generator with geometric entropy from Recursive Division Tree analysis

RGE-256: Geometric Entropy PRNG

A novel ARX-based pseudorandom number generator (PRNG) with rotation schedules derived from Recursive Division Tree (RDT) entropy constants.

Features

• 256-bit State: 8 × 32-bit words for large period and good distribution
• ARX Architecture: Add-Rotate-XOR operations for efficient mixing
• RDT-Derived Constants: Rotation schedule based on mathematical entropy constants
• Multiple Output Formats: 32-bit, 64-bit, floats, ranges, bytes
• State Serialization: Save and restore generator state
• Utility Functions: Shuffle, choice, sample

Installation

pip install rge256-core

Quick Start

from rge256_core import RGE256

# Create a generator with a seed
rng = RGE256(seed=12345)

# Generate random numbers
print(rng.next32())           # Random 32-bit integer
print(rng.next_float())       # Random float in [0, 1)
print(rng.next_range(1, 100)) # Random int in range [1, 100]

# Generate random bytes
random_bytes = rng.next_bytes(16)
print(random_bytes.hex())

# Shuffle a list
deck = list(range(52))
shuffled = rng.shuffle(deck)

API Reference

Constructor

RGE256(
    seed=None,          # int, bytes, str, or None (uses system entropy)
    rounds=3,           # Number of mixing rounds (1-10)
    zetas=(1.585, 1.926, 1.262),  # RDT entropy constants
    domain="rge256-default"       # Domain separation string
)

Methods

Method	Returns	Description
next32()	int	Random 32-bit unsigned integer
next64()	int	Random 64-bit unsigned integer
next_float()	float	Random float in [0, 1)
next_double()	float	High-precision random double
next_range(min, max)	int	Random integer in [min, max]
next_bytes(n)	bytes	n random bytes
shuffle(list)	list	Fisher-Yates shuffled copy
choice(list)	any	Random element from list
sample(list, k)	list	k unique random elements

Design Notes

RGE-256 is a statistical PRNG, not a cryptographically secure RNG. Do not use for cryptographic applications.

Author

Steven Reid - ORCID: 0009-0003-9132-3410

License

MIT License