KANQAS: Kolmogorov-Arnold Network for Quantum Architecture Search
Abstract
Quantum architecture search (QAS) is a promising direction for optimization and automated design of quantum circuits towards quantum advantage. Recent techniques in QAS focus on machine learning-based approaches from reinforcement learning, like deep Q-network. While multi-layer perceptron-based deep Q-networks have been applied for QAS, their interpretability remains challenging due to the high number of parameters. In this work, we evaluate the practicality of Kolmogorov-Arnold Networks (KANs) in QAS problems, analyzing their efficiency in the task of quantum state preparation and quantum chemistry. In quantum state preparation, our results show that in a noiseless scenario, the probability of success and the number of optimal quantum circuit configurations to generate the multi-qubit maximally entangled states are $2 imes$ to $5 imes$ higher than Multi-Layer perceptions (MLPs). Moreover, in noisy scenarios, KAN can achieve a better fidelity in approximating maximally entangled state than MLPs, where the performance of the MLP significantly depends on the choice of activation function. In tackling quantum chemistry problems, we enhance the recently proposed QAS algorithm by integrating Curriculum Reinforcement Learning (CRL) with a KAN structure instead of the traditional MLP. This modification allows us to design a parameterized quantum circuit that contains fewer 2-qubit gates and has a shallower depth, thereby improving the efficiency of finding the ground state of a chemical Hamiltonian. Further investigation reveals that KAN requires a significantly smaller number of learnable parameters compared to MLPs; however, the average time of executing each episode for KAN is higher.
AI-Generated Overview
Here is a brief overview of the text from the scientific paper titled "KANQAS: Kolmogorov-Arnold Network for Quantum Architecture Search":
-
Research Focus: The study investigates the effectiveness of Kolmogorov-Arnold Networks (KANs) in Quantum Architecture Search (QAS), specifically for optimizing quantum circuits aimed at achieving quantum advantage.
-
Methodology: The authors replace the traditional Multi-Layer Perceptrons (MLPs) in reinforcement learning-assisted QAS frameworks with KANs, and evaluate their performance in two primary tasks—quantum state preparation and quantum chemistry problems—under both noiseless and noisy scenarios.
-
Results: KANs demonstrated a 2× to 5× increase in the probability of success for generating multi-qubit maximally entangled states compared to MLPs in noiseless conditions. In noisy settings, KANs outperformed MLPs in fidelity for approximating maximally entangled states. Furthermore, KANs produced parameterized quantum circuits that were more compact with fewer 2-qubit gates compared to MLPs.
-
Key Contribution(s): This paper introduces KANQAS, a framework that effectively integrates KANs into QAS, showcasing their ability to achieve reliable and efficient quantum circuit designs with fewer learnable parameters than traditional MLPs.
-
Significance: By demonstrating that KANs can optimize quantum architecture searches more efficiently than MLPs, this research contributes a novel approach towards enhancing the design and implementation of quantum circuits, which is essential for the advancement of quantum computing technologies.
-
Broader Applications: The findings pave the way for future research into using KANs for various applications in quantum chemistry, algorithm automation, and machine learning, as well as potential scalability to larger quantum systems and exploration of their interpretability in complex tasks.