optimized_clr.py

import numpy as np


# decorator that validates inputs for the funding mechanisms
def funding_mechanism(func):
    def wrapper(self, contribution_matrix: np.ndarray, *args, **kwargs):
        # make sure the incoming contribution array is valid

        if not isinstance(contribution_matrix, np.ndarray):
            raise TypeError("contribution_matrix must be a numpy array")

        if len(contribution_matrix.shape) != 2:
            raise ValueError("contribution_matrix must be a 2D array")

        return func(self, contribution_matrix, *args, **kwargs)

    wrapper.__name__ = func.__name__
    return wrapper


class RoundManager:
    # TODO: docstring

    def __init__(self, matching_funds: float):
        self.matching_funds = matching_funds

        # useful for comparing all funding mechanisms
        self.funding_mechanisms = [self.quadratic_funding, self.one_dollar_one_vote, self.one_person_one_vote]

    def _calculate_matching(self, votes_per_public_good: np.array):
        assert len(votes_per_public_good.shape) == 1
        return np.array(votes_per_public_good) / np.sum(votes_per_public_good) * self.matching_funds

    @funding_mechanism
    def quadratic_funding(self, contribution_matrix: np.ndarray):
        # if not np.all(contribution_matrix[contribution_matrix > 0] >= 1):
            # raise ValueError("Quadratic funding requires each non-zero donation is >= 1 (because sqrt(x) when x is between 0 and 1 is larger than x).")

        # quadratic funding requires each donation is >= 1
        contribution_matrix[contribution_matrix < 1] = 0

        votes = np.power(np.sum(np.sqrt(contribution_matrix), axis=0), 2)
        return self._calculate_matching(votes)

    @funding_mechanism
    def one_dollar_one_vote(self, contribution_matrix: np.ndarray):
        votes = np.sum(contribution_matrix, axis=0)
        return self._calculate_matching(votes)

    @funding_mechanism
    def one_person_one_vote(self, contribution_matrix: np.ndarray):
        contribution_matrix = contribution_matrix > 0
        votes = np.sum(contribution_matrix, axis=0)
        return self._calculate_matching(votes)

how the matrix is loaded from a CSV: https://gist.github.com/nollied/a7996277ffa558a854d799d0e936bbb1