Quantum automatic control and quantum learning machine

Abstract

We consider a novel approach of automatic control for quantum information processing. Based on the quantum automatic control theory, we propose a quantum learning machine (QLM) which can be used to develop a quantum algorithm. The proposal for QLM is based on the following hypotheses : H.1) QLM is non-deterministic. H.2) QLM eventually works a given task as maximizing the fidelity for the target states, and H.3) when the fidelity is optimized, the operation that QLM performs will be a quantum algorithm for the given task. In this work, we suggest an optimal learning method and illustrate that QLM can learn single quantum gates, in particular quantum NOT-gate. Going one step further, we show that QLM can find Deutsch algorithm that is different from but equivalent to the original one.