Solving NP-complete Problems Using Quantum Weightless Neuron Nodes

Despite neural networks have super-Turing computing power, there is no known algorithm for obtaining a classical neural networks that solves NP-complete problems in polynomial time. However this paper shows that a quantum neural networks model coupled with a non-unitary operator can solve 3-SAT in polynomial time. The proposed method uses a network circuit to represent a Boolean logic function and a non-unitary operator to decide the satisfiability. The parameters of the network is set deterministically and manually, accordingly to the problem at hand with neither quantum nor classical learning.