HypercubeChallenge is the Python library built by Michael Green as a submission for the Hypercube Challenge.
build an array of vectors in an n-dimensional hypercube, whose vectors satisfy a system of non-linear constraints.
The current method is an iterative process which, from a set of random vectors within the hypercube, linearly interpolates between good vectors and bad vectors to generate new vectors, until the number of required accepted vectors is reached. An initial set of random vectors is generated as an array called test_values. test_values is then passed through a cleaning function to extract vectors which satisfy the constraints provided, and place these vectors in a separate array called report_values to be reported.
Now that we have two arrays, the arrays are passed through a generator function. For each vector in the test_values array, the generator function pairs this vector with a random vector from the report_values array. Linear interpolation is then used to replace the vector in the test_values array with a new vector which is between the good vector and bad vector. These new vectors are then passed back to the cleaning function for evaluation. This process of cleaning and generating is repeated until there are no bad vectors.
From the command prompt, the latest version this library can be installed via pip and git
pip install git+https://github.com/1mikegrn/HypercubeChallenge
Where the setup file will automatically check dependencies and install
to the main module library. Once installed, calling HyperCMD <input_file> <output_file> <n_results> will execute the calculation from the
command line.
This library also has a colab jupyter notebook, from which calculations can be executed without any necessary local downloads. See the included repo-badge at the top of this page to be directed there.