Quantum Properties

Yuuya Yoshida, Masahito Hayashi

The mixing property is a useful property for quantum dynamics as well as ergodicity. The decoupling property plays an important role in open quantum systems and quantum information. We show that these two properties are equivalent to each other for quantum dynamics. We derive several useful necessary and sufficient conditions for these conditions by using the ergodicity. Additionally, we discuss irreducibility and primitivity which are important properties used in Perron-Frobenius theory.

Yasunori Fujii, Kei‐ichi Maeda

Uman Khalid, Junaid ur Rehman, Haejoon Jung +3 more

The learning of fundamental quantum properties—namely coherence, discord, and entanglement—benchmarks the security, computational, and metrological capability of noisy intermediate-scale quantum (NISQ) communication, computing, and sensing networks. The current learning techniques vary widely for these fundamental quantum properties, including standard tomographic procedures that involve exhaustive optimization. Fortunately, the fundamentally distinct quantum properties feature an intricate connection. In this paper, we put forth the concept of universal quantum witness machines (UQWMs) to develop a unified framework for quantum property learning (QPL) of a quantum system. We first formulate the certification and quantification of quantum properties based on quantum witnesses. The witness-based certification method is experimentally accessible and resource-efficient but lacks reliability and generality. To universalize the scope and circumvent the unreliability, we transform the certification task into a classification task by employing UQWMs with classical machine learning to construct quantum property classifiers. This formalism offers a unifying perspective on the certification, quantification, and classification of these enigmatically linked fundamental quantum properties. To demonstrate our UQWM approach, we provide a comparative numerical analysis of quantum property quantification with quantum witnesses and classification performance analysis of quantum property classification with convolutional neural networks, specifically for$4 \times 4$quantum systems.

Harry Paul

Maximilian Schlosshauer

Sounak Bhowmik, Travis S. Humble, Himanshu Thapliyal

Quantum neural networks (QNN) hold immense potential for the future of quantum machine learning (QML). However, QNN security and robustness remain largely unexplored. In this work, we proposed novel Trojan attacks based on the quantum computing properties in a QNN-based binary classifier. Our proposed Quantum Properties Trojans (QuPTs) are based on the unitary property of quantum gates to insert noise and Hadamard gates to enable superposition to develop Trojans and attack QNNs. We showed that the proposed QuPTs are significantly stealthier and heavily impact the quantum circuits’ performance, specifically QNNs. The most impactful QuPT caused a deterioration of $23 \%$ accuracy of the compromised QNN under the experimental setup. To the best of our knowledge, this is the first work on the Trojan attack on a fully quantum neural network independent of any hybrid classical-quantum architecture.