2D Response Surface

A heatmap of the GP posterior mean over two input parameters, with all others held fixed. This reveals how pairs of parameters jointly influence the outcome — including interactions that 1D slices can't capture.

Color encodes the predicted value (yellow = high, purple = low). Training points are overlaid as dots. A second panel shows the posterior standard deviation — brighter regions indicate higher model uncertainty, suggesting where additional trials would be most informative.

The two most important parameters (by kernel lengthscale) are selected by default.

Interactivity:

• Axis dropdowns — choose which two parameters to plot
• Outcome selector — switch between objectives
• Sliders — adjust held-fixed values for non-plotted parameters
• Pivoting — click a training point to pivot: all sliders snap to that trial's parameter values and the surface redraws around that operating point. Nearby dots stay opaque; distant ones fade based on kernel correlation