Characterizing Quantum Evolutions Via a Recommender System

We explore the possibility of predicting the properties of quantum evolutions via matrix factorization algorithm, a particular type of the recommender system (RS). A system undergoing a quantum evolution can be characterized in several ways. Here we choose (i) quantum correlations characterized by measures such as entropy, negativity, or discord, and (ii) state-fidelity. Using quantum registers with up to 10 qubits, we demonstrate that an RS can efficiently characterize both unitary and nonunitary evolutions. After carrying out a detailed performance-analysis of the RS in two-qubits, we show that it can be used to distinguish a clean database of quantum correlations from a noisy one. Moreover, we find that the RS may bring about a significant computational advantage for estimating the quantum discord of a general quantum state, for which no simple closed-form expression exists. Finally, we show that the RS can efficiently characterize systems undergoing nonunitary evolutions in terms of both reduced quantum correlations and state-fidelity.