Appendix, proof of Lemma 4.2, p. 40 — T(−Q + (Te₁+w)(Te₁+w)ᵀ/(1+e₁ᵀTw))T is (31) at ŵ = Tw
ProvedConicQuadIPM.NTScaling.rot_scaling_conjugateconic-optimizationnesterov-todd-scalingp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1second-order-cone
Let , let , be the matrices (18), (19) of the rotated quadratic cone and . For every ,
Combined with (63) this shows that the rotated-cone matrix (34) is the conjugate by of the quadratic-cone NT matrix (31), from which (34) follows.
Formalization Note Vectors are Fin d → ℝ with 0-based indices (the paper's is coord x 0); is tailSq x 1, a sum of squares (Euclidean, never Mathlib's sup norm); is x ⬝ᵥ s. The page writes the denominator as , a printed slip for ; the Lean uses . Both sides use Mathlib's total inverse, so the identity is stated for every .
Preamble
import Mathlib import Definitions.Def_ConicQuadIPM_NTScaling_Setting import Definitions.Def_ConicQuadIPM_NTScaling_Scaling open Matrix
Formal statement
namespace ConicQuadIPM.NTScaling
/-- Appendix, proof of Lemma 4.2, p. 40 (rotated quadratic cone): conjugating the matrix (34)
by `T` gives the quadratic-cone matrix (31) at `ŵ = Tw`, i.e. `T W T = Ŵ`. -/
theorem rot_scaling_conjugate (d : ℕ) (hd : 2 ≤ d) (w : Fin d → ℝ) :
Tmat .rot d * Wrot d w * Tmat .rot d = WquadOuter d (Tmat .rot d *ᵥ w) := by sorry
end ConicQuadIPM.NTScaling
Source
Andersen, Roos & Terlaky, On implementing a primal-dual interior-point method for conic quadratic optimization, Math. Program. (2003), DOI 10.1007/s10107-002-0349-3; authors' preprint of 18 Dec 2000, p. 40, Appendix, proof of Lemma 4.2 (end)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.