Enhancing Circuit Trainability with Selective Gate Activation Strategy

Quantum computing has shown promise in solving complex Techniques such as layerwise training (Skolik et al. problems in domains such as quantum chemistry, optimization, 2021) and parameter initialization schemes based on symmetry and machine learning, leveraging Variational Quantum considerations (Pesah et al. 2021) have been proposed Algorithms (VQAs) such as Quantum Approximate to achieve this. Optimization Algorithms (QAOA) (Farhi, Goldstone, and Local cost functions, selective parameter training, and Gutmann 2014; Pagano et al. 2020), Variational Quantum structured initialization methods have shown promise in mitigating Eigensolvers (VQE) (Kandala et al. 2017; Tilly et al. 2022), trainability challenges without significantly compromising and recently, quantum neural networks (QNNs) (Schuld and circuit expressibility. Moreover, techniques like symmetric Killoran 2019; Killoran et al. 2019) as a hybrid quantumclassical pruning (Wang et al. 2023), which leverage circuit framework in the Noisy Intermediate-Scale Quantum symmetries to reduce the effective parameter space, have (NISQ) era.