Residual plateau after removing numerical and optimization effects

Determine how much of the residual plateau at the regular Reeb vector of the dP2 pentagon persists after sampling and optimization effects are removed, and whether increasing the optimization budget eliminates any further portion of the plateau.

Background

At the regular Reeb vector, where a Kähler–Einstein metric on dP2 is ruled out by the automorphism-group obstruction, the numerical polynomial fits exhibit a residual plateau. The authors emphasize that a plateau can also result from insufficient sampling, boundary effects, or optimization failure, and that the experiments do not separate these mechanisms completely. They therefore leave unresolved the limiting size of the plateau and the extent to which additional computational budget would reduce it.