ABSTRACT
As an example of a proposed alternative to floats, try the iterative loop using double precision interval arithmetic, which consumes 128 bits per value instead of 64. They had better be worth their high price:
They were not worth it. This answer is typical of interval arithmetic: A correct but useless result that says the answer is: Somewhere on the real number line. Now you can start to see why interval arithmetic has attracted a very limited fan club.
