One can, however, catch some of these with unit tests, in maths just as well as in code. The unit tests for maths are tiny, trivial cases of the main theorem. F

One can, however, catch some of these with unit tests, in maths just as well as in code. The unit tests for maths are tiny, trivial cases of the main theorem. For instance, if I wrote down Fermat's Last Theorem as "x^n + y^n = z^n has no solutions for n > 1", I would run my unit tests by checking the boundary case n=2, and discovering the easy counterexample of (3,4,5).