Bayesian Optimization for Hyperparameter Tuning (2D)

This 2D companion drives the same Gaussian-process surrogate and Upper-Confidence-Bound acquisition function as the 3D lab, drawn as a flat top-down heatmap instead of an orbiting surface: the posterior mean colors the grid, sampled trials appear as dots, and a pulsing marker shows the acquisition function's current top pick. Raise exploration (κ) to chase uncertain regions or lower it to exploit the best trial found so far, and toggle the hidden true landscape to see how closely the surrogate has converged.