A Representative Framework for Implementing Quantum Finite Automata on Real Devices
Aliya Khadieva, Özlem Salehi, Abuzer Yakaryılmaz
TL;DR
The paper tackles implementing quantum finite automata for the language $MOD_p$ on gate-based quantum computers in the NISQ era. It develops device-independent circuit decompositions for uniformly controlled rotations and hardware-aware adaptations to IBM Quantum backends, including pseudo-rotations and linear-nearest-neighbor optimizations. Key contributions include tunable trade-offs between CNOT-cost and rotation-precision (e.g., $2^t + \frac{d}{2}(192(\log_2 d - t) - 768)$ for $0<t<\log_2 d - 4$) and practical demonstrations showing substantial reductions in CNOT counts and circuit depth, along with improved discrimination between members and non-members for $MOD_p$ (e.g., $p=11$, $p=37$). The findings offer actionable guidance for deploying QFA on NISQ devices and underscore the importance of topology-aware transpilation and basis-gate alignment.
Abstract
We present a framework for the implementation of quantum finite automata algorithms designed for the language $ MOD_p = \{ a^{i\cdot p } \mid i \geq 0 \}$ on gate-based quantum computers. First, we compile the known theoretical results from the literature to reduce the number of CNOT gates. Second, we demonstrate techniques for modifying the algorithms based on the basis gates of available quantum hardware in order to reduce circuit depth. Lastly, we explore how the number of CNOT gates may be reduced further if the topology of the qubits is known.
