§5, p. 356 — and
ProvedExpConeIPM.Curvature.witness_memconic-optimizationexponential-coneinterior-point-methodsp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1
Let be the exponential cone. The points
satisfy and .
These are the base point and the direction at which Dahl and Andersen exhibit the failure of negative curvature for the exponential-cone barrier; the definition of negative curvature needs the base point in the interior of the cone and the direction in the cone.
Formalization Note The paper writes ; this statement asserts the stronger interior membership, which the definition of negative curvature requires and which holds since . Coordinates are 0-based: the paper's are (x 0, x 1, x 2).
Preamble
import Mathlib import Definitions.Def_ExpConeIPM_Curvature_ExpCone import Definitions.Def_ExpConeIPM_Curvature_NegativeCurvature open scoped InnerProductSpace
Formal statement
namespace ExpConeIPM.Curvature
/-- **§5, p. 356 — the witness.** `x̂ := (1, e^{−2}, 0)` lies in `int(Kexp)` (the paper writes
`x̂ ∈ Kexp`; the definition of negative curvature needs, and this states, the interior), and
`û := (1, 0, 0) ∈ Kexp \ int(Kexp)`. -/
theorem witness_mem :
!₂[1, Real.exp (-2), 0] ∈ interior Kexp ∧
!₂[(1 : ℝ), 0, 0] ∈ Kexp ∧ !₂[(1 : ℝ), 0, 0] ∉ interior Kexp := by sorry
end ExpConeIPM.Curvature
Source
Dahl, Andersen, A primal-dual interior-point algorithm for nonsymmetric exponential-cone optimization, Math. Program. 194 (2022), p. 356, §5
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.