Hybrid Quantum Solvers in Production: how to succeed in the NISQ era?
Eneko Osaba, Esther Villar-Rodriguez, Aitor Gomez-Tejedor, Izaskun Oregi
TL;DR
This work surveys hybrid quantum–classical solvers and classifies them using established taxonomies, arguing that such hybrids will persist beyond fault-tolerance. It focuses on two production-ready solvers, LeapBQM and Quantagonia's QHS, and benchmarks them with the QOPTLib suite across TSP, VRP, BPP, and MCP. Experimental results indicate that QHS generally outperforms LeapBQM and shows stronger scalability, particularly for non-MCP problems, while MCP is solvable optimally by both. A key finding is that the placement of quantum resources (exploration vs exploitation) significantly affects performance, underscoring the need for diverse, designer-controlled hybrid architectures in the NISQ era and beyond.
Abstract
Hybrid quantum computing is considered the present and the future within the field of quantum computing. Far from being a passing fad, this trend cannot be considered just a stopgap to address the limitations of NISQ-era devices. The foundations linking both computing paradigms will remain robust over time. The contribution of this work is twofold: first, we describe and categorize some of the most frequently used hybrid solvers, resorting to two different taxonomies recently published in the literature. Secondly, we put a special focus on two solvers that are currently deployed in real production and that have demonstrated to be near the real industry. These solvers are the LeapHybridBQMSampler contained in D-Wave's Hybrid Solver Service and Quantagonia's Hybrid Solver. We analyze the performance of both methods using as benchmarks four combinatorial optimization problems.