QOBLIB Expands Shared Benchmarks for Quantum Optimization
A common library of ten optimization problem classes gives quantum and classical approaches comparable tasks and public results.
IBM engineers installing Quantum System One at Fraunhofer, used as hardware context for the shared optimization benchmark. Image credit“IBM Q System One (Fraunhofer) installation” by Holger Muench / IBM Research, CC BY 2.0. · https://creativecommons.org/licenses/by/2.0/
Comparing optimization machines gets harder when each tackles a different version of the problem. QOBLIB supplies shared problem definitions and public baselines so researchers can compare approaches on common ground. IBM's August 12 update describes a new website for the library and a growing collection of results across ten problem classes.
IBM reports more than 2,000 submissions and links the project's foundational research. That makes the library a resource for comparison, rather than evidence that a quantum solver wins.
Any claimed lead still depends on solution quality, runtime, hardware access and how the comparison was made. Reproducible results on identical instances, with the full classical and quantum resource costs disclosed, are what would make those comparisons convincing.