optimize_method.py

def optimize(population, rounds):
    population = sort_population(population)
    # Set our seed set to a pretty good default
    test_set = population[:10]
    # To compare other sets against
    highest_exposure = len(get_intersection(test_set))
    print "Baseline: %d" % highest_exposure
    print "=============="

    for i in range(rounds):
        # Find somebody to remove
        index = find_worst(test_set)
        # Find somebody to replace them
        replacement = get_replacement(population, test_set)

        # Here we get our new test set
        # We are using a form of simulated annealing to try
        # to avoid local maximas
        if random.random() < heat(i, rounds):
            new_set = remove_at_index(test_set, random.randrange(len(test_set)))
            new_set.append(replacement)
        else:
            new_set = remove_at_index(test_set, index)
            new_set.append(replacement)

        # Test it out
        exposure = len(get_intersection(new_set))
        if exposure > highest_exposure:
            highest_exposure = exposure
            test_set = new_set

    print "Conclusion: %d" % highest_exposure
    return test_set