Abstract
We prove computability and complexity results for an original model of computation called the continuous space machine. Our model is inspired by the theory of Fourier optics. We prove our model can simulate analog recurrent neural networks, thus establishing a lower bound on its computational power. We also define a $\Theta(\log_{2}n)$ unordered search algorithm with our model.
Keywords: continuous space machine, unconventional model of computation, analog computation, optical computing, computability, computational complexity, analog recurrent neural network, Fourier transform, binary search, unordered search
|