quantum-random 1.2.4

Quantum random numbers

Quantum random numbers in Python

This package brings the ANU quantum random numbers to Python 3.7 to 3.10.

The default pseudo-random generator in Python is replaced by calls to the ANU API that serves real quantum random numbers.

Install

pip install quantum-random

Optionally, for NumPy support,

pip install quantum-random[numpy]

Note that the NumPy integration is not well-tested and is only available for Python 3.8 and 3.9.

Setup: passing your API key

ANU now requires you to use an API key. You can get a free (trial) or paid key from here.

You can pass your key to qrandom in three ways:

1. By setting the environment variable QRANDOM_API_KEY.
2. By running qrandom-init to save your key in an INI config file that is stored in a subdirectory of your default home config directory (as specified by XDG, e.g., /home/<your-username>/.config/qrandom/).
3. By running qrandom-init to save your key in an INI file in a directory of your choice set by QRANDOM_CONFIG_DIR.

qrandom will look for the key in the order above. The qrandom-init utility is interactive and comes installed with qrandom.

Usage

Just import qrandom and use it like you'd use the standard Python random module. For example,

>>> import qrandom

>>> qrandom.random()
0.15357449726583722

>>> qrandom.sample(range(10), 2)
[6, 4]

>>> qrandom.gauss(0.0, 1.0)
-0.8370871276247828

Alternatively, you can import QuantumRandom from qrandom and use the class directly (just like random.Random).

Under the hood, batches of quantum numbers are fetched from the API as needed and each batch contains 1024 numbers. If you wish to pre-fetch more, use qrandom.fill(n), where n is the number of batches.

Optionally, if you have installed the NumPy integration,

>>> from qrandom.numpy import quantum_rng

>>> qrng = quantum_rng()

>>> qrng.random((3, 3))  # use like numpy.random.default_rng()
array([[0.37220278, 0.24337193, 0.67534826],
       [0.209068  , 0.25108681, 0.49201691],
       [0.35894084, 0.72219929, 0.55388594]])

Tests

To run the tests locally, you will need poetry and Python 3.7-3.10 (i.e., multiple versions of Python installed and seen by tox using, for example, pyenv). Then,

poetry install
poetry run tox

See here for a visualisation and a Kolmogorov–Smirnov test.

Notes on implementation

The qrandom module exposes a class derived from random.Random with a random() method that outputs quantum floats in the range [0, 1) (converted from 64-bit ints). Overriding random.Random.random is sufficient to make the qrandom module behave mostly like the random module as described in the Python docs. The exceptions at the moment are getrandbits() and randbytes() that are not available in qrandom. Because getrandbits() is not available, randrange() cannot produce arbitrarily long sequences. Finally, the user is warned when seed() is called because there is no state. For the same reason, getstate() and setstate() are not implemented.

NumPy support is provided using RandomGen.

License

See LICENCE.