Probe, Then Commit: Why Solver Tuning Finally Grew Up
Opening — Why this matters now Constraint programming (CP) has always promised elegance: state the problem, let the solver do the work. In practice, however, seasoned users know the uncomfortable truth—solver performance lives or dies by hyperparameters most people neither understand nor have time to tune. As problem instances grow larger and solver configurations explode combinatorially, manual tuning has become less of an art and more of a liability. The paper Hyperparameter Optimization of Constraint Programming Solvers confronts this reality head-on, proposing a framework that finally treats solver configuration as what it is: a resource allocation problem under uncertainty. ...