A new qubit mapping approach called QARMA-R achieves a 97-100% reduction in core-to-core operations, minimizing the main sources of noise and decoherence in modular quantum systems. Researchers at Bukyong National University in South Korea have developed QARMA-R, which leverages attention-based deep reinforcement learning to optimize the physical layout and operation of quantum circuits. Experimental results show that QARMA-R reduces inter-core communication by an average of 86% and outperforms highly optimized configurations within the Qiskit framework. This advancement, detailed in Physics Applied, enables the execution of larger quantum algorithms on limited hardware and contributes to the development of scalable quantum computing architectures.
QARMA-R achieves an average 86% inter-core communication reduction
This performance improvement exceeds existing tools for managing quantum information flow between processing units. This innovation combines attention-based mechanisms with graph neural networks to learn optimal qubit allocation, routing, and reuse strategies. QARMA-R incorporates dynamic qubit reuse to further increase efficiency. QARMA itself achieves 97-100% reduction in core-to-core operations when compared to traditional modular qubit mapping while maintaining 15% and 40% improvements for large-scale circuits without reuse. These improvements are important for scaling quantum computing architectures, as costly inter-core operations and quantum state transfer currently limit the size and complexity of the algorithms that can be executed.
Although modular quantum computing holds promise for scalability, it currently faces limitations due to physical constraints in interconnecting multiple quantum processing units. Existing methods for assigning qubits to physical locations (known as mapping) suffer from significant communication overhead between these processing units, which is a major source of errors and decoherence. This application of artificial intelligence optimizes the physical placement and operation of quantum circuits, as well as the algorithms themselves.
