Terry McCracken wrote:bob wrote:Terry McCracken wrote:bob wrote:Guenther wrote:Terry McCracken wrote:melajara wrote:See
http://www.kurzweilai.net/d-wave-system ... ng-barrier
At 1000 qubits, the new processor considers 2^1000 possibilities simultaneously, a search space which dwarfs the 2^512 possibilities available to the 512-qubit D-Wave Two. ”In fact, the new search space contains far more possibilities than there are particles in the observable universe.”
D-Wave is mentioning some applications here,
http://www.dwavesys.com/quantum-computing/applications
I was wondering if some of them could be bent to target chess.
So far, I was sure I would never see chess solved in my lifetime. With quantum computers gaining traction, I'm not so sure anymore.
Two thirds voted never?! Is this the nineteenth or twenty-first century?
Before quantum computing will have any impact on chess programming it will be solved by other means so the answer to the poll is 'never'.
I'm not so sure it will ever be solved by any means. To solve it, ALL paths have to be searched to an endpoint conclusion of some sort. That is an absolutely impossible number of nodes thanks to the rules of the game (50 move rule is one issue). With the longest possible game something like 5500 moves, alpha/beta would choke on sort(38^5500) nodes. I don't even know how one would think about storing such a thing while searching it.
No machine present has the potential and programming may have to start from scratch when a quantum computer with the potential is created.
I think a machine with the potential may be created well before 2050.
Even with quantum computing, I don't see how such a tree is going to be searched in parallel or stored (38^5500 positions)...
We'll have to wait and see. With a massive quantum computer it should be doable but such a machine has to be created first.
We might have AGI before we have such a quantum monstrosity. Then I will be more concerned, if I'm still around, what the AGI could do. A hell of a lot more than us.
The entire tree does not have to be searched if there is a forced mate that is closer.
Suppose (for instance) that there is a forced mate in 200.
In such a case, 38^200 nodes would be all that are needed.
For some algorithms (e.g. factoring a number) a quantum computer can change NP into P.
A 1000 qubit quantum computer makes RSA and other cryptography schemes unsafe. Since it is not unlikely that our banks use RSA variants, that is something to actually worry about (should a malevolent person gain access to such a computer). The factoring algorithm does not simply try every possible factor, but instead utilizes properties of modular arithmetic.
https://en.wikipedia.org/wiki/Shor%27s_algorithm
The number 56,153 was factored on a quantum computer with 4 qubits.
Nikesh S. Dattani and Nathaniel Bryans. "Quantum factorization of 56153 with only 4 qubits."
Quantum computers also have a special black box search algorithm, but it only reduces search effort to sqrt(nodes).
https://en.wikipedia.org/wiki/Grover%27s_algorithm
It remains to be seen if there exists any algorithm which can turn chess search into P from NP. But until Shor's algorithm, factoring was assumed to be NP (exponential as a function of the number of bits of the product of two large primes).