Random property-based testing is useful for that sort of thing, as used in the QuickCheck library (most popular in Haskell/Erlang but there are ports to pretty

Random property-based testing is useful for that sort of thing, as used in the QuickCheck library (most popular in Haskell/Erlang but there are ports to pretty much every language in existence). One property you could use for your example would be to have a simpler, slower reference implementation that you trust and test that it produces the same output as the one you've written. This isn't a verification but is very effective at snuffing out bugs. But you are right that testing does not verify the correctness of code (unless you are writing something like a boolean -> boolean function), but often property-based testing is a stepping stone to verification since there is a quicker feedback cycle between write code->find bugs->fix code->repeat than the equivalent write code->fail to prove it works->wonder if its broken or if you just don't know how to prove it.