Proof of Theorem 1.14, p. 0:26 — z̄⊗z⊗z̄ + z⊗z̄⊗z = A with z = x + √2 y, so rank_ℝ(A) ≤ 2
ProvedTensorNP.Rank.real_decompositionp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1tensor-rank
Let , , regarded in , and put and . Let be the tensor . Then
and consequently .
This is the upper half of Theorem 1.14: over the reals, is a sum of two outer products.
Formalization Note The identity is an equality of real arrays, with 's rational entries cast to ; the rank bound is about the same cast tensor.
Preamble
import Mathlib import Definitions.Def_TensorNP_Rank_Defs import Definitions.Def_TensorNP_Rank_Construction
Formal statement
namespace TensorNP.Rank
theorem real_decomposition :
let x : Fin 2 → ℝ := fun i => (vecX i : ℝ)
let y : Fin 2 → ℝ := fun i => (vecY i : ℝ)
let z : Fin 2 → ℝ := x + Real.sqrt 2 • y
let zbar : Fin 2 → ℝ := x - Real.sqrt 2 • y
(outer zbar z zbar + outer z zbar z = fun i j k => (tensorA i j k : ℝ)) ∧
rankOver ℝ (fun i j k => (tensorA i j k : ℝ)) ≤ 2 := by sorry
end TensorNP.Rank
Source
Hillar & Lim, Most Tensor Problems Are NP-Hard, J. ACM 60(6) (2013), Art. 45 (final version, arXiv:0911.1393v5), p. 0:26, proof of Theorem 1.14, first display
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.