The unprovable algorithm usually use randomness
and/or rely for their correctness on unproved
hypotheses, ...
As an example, some algorithms rely on the Riemann Hypothesis for their correctness. ... yet are observed to be the fastest in practice
for numbers of substantial size.
The Pollard Rho algorithm is not a fast one, running in expected time proportional to the square root of the factor found, which is usually (but not always) the smallest factor. however, this effectively relies on the birthday problem, and is only a heuristic, not proven. It's also not guaranteed to factor the number at all. The provable algorithms are usually deterministic,
but their key feature is that their correctness is
unconditional.
And their run time is proven without using unproven hypotheses.Similarly, the complexity of the quadratic sieve is heuristic because there is no known way to prove that (sqrt(n) + i)^2 - n results in the same distribution of smooth integers than that of a random sequence of the same size. Dixon's algorithm [1] turns the quadratic sieve provable (but much slower in practice) by sampling randomly instead of sequentially. Sieving is thus no longer possible, but we get a subexponential running time of L_n[1/2, 2 sqrt(2)]. The hyperelliptic curve factorization algorithm [2, 3] is another probabilistic but provable algorithm with a more efficient but not provable counterpart, the elliptic curve method [4].
The blog post only seems to care about deterministic factoring algorithms, though. There are indeed no known deterministic factorization algorithms with better than exponential complexity.
[1] http://www.ams.org/journals/mcom/1981-36-153/S0025-5718-1981...
[2] https://math.dartmouth.edu/~carlp/hyperI.pdf
[3] https://math.dartmouth.edu/~carlp/PDF/paper129.pdf
[4] https://openaccess.leidenuniv.nl/bitstream/handle/1887/3826/...
It's therefore not far fetched that some algorithms might in fact always work correctly, but at the same time a proof for their correctness doesn't exist. You can never be sure (literally!) that you're not just confusing it with an algorithm that's wrong but happens to work on all problem instances you've thrown at it so far.
IIUC, when the article says unprovable, what it actually means is unproven. To correctly describe an algorithm as unprovable, you would have to have a proof of its unprovability, which would be quite something.
In fact, singling out ZFC gives you another example of this kind of result: if ZF is consistent, then so are ZF+C and ZF-C.
Oh, I see. I read "A is consistent with B" as "A+B is consistent", but I can see the validity of the reading "A+B is 'as consistent as' A".
In a deterministic algorithm, this bound will always exist. In a non-deterministic algorithm (e.g. one that uses a random walk), this bound will not. The funny thing is that in the real world (and for big data) the latter algorithms generally run faster than the former.
If you're wondering why a non-deterministic algorithm can't have bounds, it's simple! Because it would range from a trivial best case of I get really lucky and randomly hit every "good" factor in constant time O(1) to the trivial worst cast of I get really unlucky and keep guessing the wrong numbers ad infinitum.