Projektnummer

0072677

Zusammenfassung

It is a curios paradox, often catching researchers off guard, that those entrenched in computerintensive sciences struggle with reproducibility. Against a common assumption, the use of computers does not guarantee the repetition of exact results. While computing power, algorithms and infrastructures have evolved swiftly, the adaption of scientific standards of its use have lagged behind. Should we consider a result reproducible only if it accommodates variations in hardware, software, parameters, and training data? Or quite the opposite? Who, then, has – or presumes to have – the authority to ...