Skip to content

DX generator

This section will guide you through the steps to install NextRNGBook and begin applying its functionality. If you're new to this package, follow the instructions below to get started smoothly.

Installation

To install NextRNGBook, you can use the Python package manager pip in your terminal or command line interface:

pip install nextrngbook

Import the package

After installing NextRNGBook, you can start by importing the package in your Python script. You will need to import both NextRNGBook and NumPy. NextRNGBook is designed to work with the underlying uniform random number generators (RNGs), while NumPy is used for higher-level operations such as generating random arrays, matrices, and distributions.

from nextrngbook.dx_generator import DX
from numpy.random import Generator

Create DX generators

NextRNGBook provides the DX function to initialize a specific DX generator from the DX generator family, which includes over 10,000 built-in DX generators. By calling this function, you can create an instance of a 32-bit DX generator that can be used for further operations.

>>> DX()
_DXGenerator(bb=24184, pp=2147483647, kk=47, ss=2, log10_period=438.6000061035156)

DX(dx_id, seed) takes two parameters:

  • dx_id: Selects a specific DX generator from the DX generator family (over 10,000 options).
  • seed: Sets the RNG state for reproducibility, provided that the seed is not None.
>>> DX(dx_id=10000, seed=308)
_DXGenerator(bb=646323, pp=2147483647, kk=47, ss=2, log10_period=438.6000061035156)

In short, dx_id determines the specific DX generator, and seed ensures reproducibility. For more details on potential issues with dx_id values and reproducibility, refer to the API Reference section.

View DX generator info

To obtain information about the created DX generator, you can print it out.

rng = DX(dx_id=9999)
print(rng)

DX-47-2 generator
Multiplier = 646173
Modulus    = 2147483647
The log₁₀(period) of the PRNG is 438.6

Use NumPy's Generator class with DX generator

After creating a DX generator with DX(), you can easily connect it to NumPy's Generator class.

>>> rng = Generator(rng)
>>> rng
Generator(_DXGenerator) at ...

Generate random numbers

Once you have connected the DX generator to NumPy's Generator class, you can begin generating random numbers. This step enables you to create random numbers based on your desired distribution or for various operations. To generate random numbers, simply call the appropriate method from the Generator class, such as integers(), random(), or others, depending on your specific needs. For more information about the available methods, you can consult the NumPy documentation. Here are some examples of generating random numbers using NumPy’s Generator with the DX generator.

# sampling from distributions
print(rng.normal(0, 1, 20)) # generate twenty N(0, 1) data
print(rng.uniform(0, 1, 10)) # generate ten U(0, 1) data

# randomly choose
print(rng.choice(["A", "B", "C", "D", "E"], size=30)) # choose thirty elements with replacement

# randomly shuffle
sample_lst = ["A", "B", "C", "D", "E"]
rng.shuffle(sample_lst)
print(sample_lst)

[ 1.85811761  1.51038657 -0.46972854  2.54010018  1.3581735  -0.19969388
-0.76451271  0.19767763 -0.80287183 -0.45321281  1.30631922  1.62328021        
-0.87913752 -0.25868414 -0.47459769 -0.11657998  1.16520699 -0.08221444        
0.72956116 -0.80021773]
[0.49784149 0.08103851 0.51156899 0.31261863 0.0444274  0.35562973
0.90995614 0.64610098 0.69553556 0.1983629 ]
['D' 'B' 'B' 'A' 'A' 'A' 'E' 'C' 'D' 'E' 'C' 'D' 'E' 'E' 'B' 'D' 'B' 'B'        
'B' 'A' 'E' 'D' 'E' 'B' 'B' 'D' 'E' 'B' 'A' 'B']
['A', 'E', 'C', 'D', 'B']

Parallel random number generation

To enable parallel computation, you can create multiple DX generators with different dx_id values. This approach enables the creation of multiple low-correlation generators, reducing dependencies between random sequences in parallel processes.

from nextrngbook.dx_generator import DX
from numpy.random import Generator

# Create multiple DX generators with different dx_id values
generators = [Generator(DX(dx_id=i)) for i in range(14760, 14769)]

Extend usage with other libraries

NumPy’s Generator, beyond basic random number generation, can be integrated with libraries such as SciPy for scientific computing and SymPy for symbolic computation. For more information, refer to their respective documentation.