New Algorithm Cuts Quantum Computing Errors by Simplifying Qudit Operations

Researchers at the National University of Science and Technology 'MISIS' have developed a new algorithm to simplify quantum computing operations. The method breaks down complex qudit-based transformations into shorter, more efficient sequences. This breakthrough could improve performance on trapped-ion and superconducting quantum systems.

The team focused on decomposing unitary transformations for qudit-based quantum computers. Their algorithm ensures that any single-qudit operation requires no more than four pulses, cutting down on errors and processing time. In most cases, even fewer pulses are needed, optimising hardware resources.

To test the approach, they compared it across different trapped-ion species and superconducting qudits. Methods like QSearch, QSweep, LocQRPass, and LocAdaPass were implemented using libraries such as BQSKit, MQT, NetworkX, and NumPy. The results showed significant improvements over existing techniques, particularly for systems with arbitrary selection rules.

Real-world testing on OVHcloud's quantum processors highlighted practical challenges. Noise and reduced gate accuracy on actual QPUs differed from emulator predictions, requiring adjustments for efficiency. The study also stressed the need to consider fidelity, gate count, and circuit depth—not just execution time—when evaluating performance.

The new algorithm offers a more efficient way to decompose qudit operations, reducing pulse requirements and potential errors. Validated across multiple platforms, it addresses real-world limitations that ideal simulations often overlook. These findings could help optimise quantum computing tasks in both trapped-ion and superconducting systems.