ProCreations's picture
Publish generalized convex exact native reproduction
2d1810a verified
Raw
History Blame Contribute Delete
882 Bytes
{
"assessment": "verified",
"claim": 3,
"convex_combinations_checked": 570,
"destructive_control": {
"effect": "the middle atom is never active, so the parameterization is not lean",
"minimum_domination_margin_on_129_points": 0.25,
"mutation": "raise only the y=0 intercept of r(y)=y\u00b2 by 1/2"
},
"exact_witness_inequalities_checked": 14250,
"finite_Y": [
"-1",
"-1/2",
"0",
"1/2",
"1"
],
"kernel": "Phi(x,y)=xy",
"lean_parameterizations": 20,
"literal_claim": "Theorem 4 (Section V.C) proves that the lean subset of finitely \u1ef8-convex functions forms a convex parameter space, which is what allows bilevel objectives to be rewritten as single-level problems solvable with standard first-order optimization (Section I).",
"unordered_pairs": 190,
"witness_identity": "at x=2ay_i+b, line_i-line_j=a(y_i-y_j)^2 >= 0"
}