Enhancing classical solutions for optimization problems with quantum-assisted methods proves to be difficult. It is unclear which algorithms are suited for which problems and how to implement them in a real-world setting. Mathematical formulation, encoding, decomposition and the selection of a hybrid algorithm and its hyperparameters require a vast amount of expert knowledge from many disciplines including physics, computer science, mathematics and engineering.

Good, reliable solution paths need to be found and automated to lower the obstacles for end users to apply hybrid quantum-classical methods. Within the QuaST project, we develop a decision tree framework that allows to explore the different options at the various levels of the optimization pipeline (formulation, encoding, decomposition, algorithm & hyperparameter selection) and give recommendations based on application-centric and application-specific metrics for the solution quality. As a starting point, the capacitated vehicle problem – relevant in countless logistic problems such as efficient garbage collection – is investigated in concrete detail. Ultimately, this framework aims to enable end users to identify valid quantum-enhanced solution paths automatically while preserving the freedom of researchers to include new and enhanced methods with a direct focus on their benefit for the application.