By @rpinho
Refereed correctness bugs in open-source research software, with patches and links to upstream PRs
before: {"means": [0.0, 0.6321205588285577, 0.33250708260777784], "normal_pdf": [0.7788007830714049, 0.6065306597126334, 0.7357588823428847, 0.0995741367357279], "normal_cdf": [0.22119921692859512, 0.3934693402873666, 0.6321205588285577, 0.950212931632136], "normal_laplace": [1.0, 0.6964173546864574, 0.5225558805394355, 0.34163117806131066], "block_fraction_at_zero": 1.2909883534346633}
after: {"means": [0.5, 0.8160602794142788, 0.3333333333333333], "normal_pdf": [0.7788007830714049, 0.6065306597126334, 0.7357588823428847, 0.0995741367357279], "normal_cdf": [0.22119921692859512, 0.3934693402873666, 0.6321205588285577, 0.950212931632136], "normal_laplace": [1.0, 0.6964173546864574, 0.5225558805394355, 0.34163117806131066], "block_fraction_at_zero": 1.0}
correct: {"means": [0.4999999999999997, 0.8160602794142786, 0.33333333333333276], "one_piece_definition": 0.5, "two_piece_definition": 0.8160602794142788}
Numerical reference agrees.