Quantum Neural Network Training of a Repeater Node

Abstract

The construction of robust and scalable quantum gates is a uniquely hard problem in the field of quantum computing. Real-world quantum computers suffer from many forms of noise, characterized by the decoherence and relaxation times of a quantum circuit, which make it very hard to construct efficient quantum algorithms. One example is a quantum repeater node, a circuit that swaps the states of two entangled input and output qubits. Robust quantum repeaters are a necessary building block of long-distance quantum networks. A solution exists for this problem, known as a swap gate, but its noise tolerance is poor. Machine learning may hold the key to efficient and robust quantum algorithm design, as demonstrated by its ability to learn to control other noisy and highly nonlinear systems. Here, a quantum neural network (QNN) is constructed to perform the swap operation and compare a trained QNN solution to the standard swap gate. The system of qubits and QNN is constructed in MATLAB and trained under ideal conditions before noise is artificially added to the system to test robustness. We find that the QNN easily generalizes for two qubits and can be scaled up to more qubits without additional training. We also find that as the number of qubits increases, the noise tolerance increases with it, meaning a sufficiently large system can produce extremely noise-tolerant results. This begins to explore the ability of neural networks to construct those robust systems.