Better yet: repeatedly divide and then take the inverse.