MPhys project supporting repository
A classical neural network structure and behaviour was modified to exhibit quantum behaviour by mimicking a quantum system of spins in the magnetic field acting as a qubit circuit. The main manipulations were done on the activation function, which has the structure of the hydrogen radial orbit superposition. With certain amplitudes of different orbitals (determined using the Monte Carlo technique) good learning rate is achieved alongside with dispersion in the intermediate qubit position. Latest implementation includes the CNOT gate acting on 2 qubit system.
Credits: The code is based on the work done by Michael Nielsen, accessible at: https://github.com/mnielsen/neural-networks-and-deep-learning