Positive scaling preserves the coefficient-change objective
OpenQAlgorithms.Papers.Independent.Intakenext100.P36.positive_scalingLet , let , and let and be finite families. For invertible , put , , and
For every positive definite and every positive real , . Each positive definite admits a positive scalar multiple with and the same cost. Consequently,
The condition means precisely that all eigenvalues lie in . Empty families have zero cost.
This explains the sampling-domain normalization mathematically, without providing any sampling success or accuracy guarantee.
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P36
noncomputable section
open scoped Matrix ComplexConjugate ENNReal ComplexOrder
namespace QAlgorithms.Papers.Independent.Intakenext100.P36
local notation "Mat" => QAlgorithms.Papers.Independent.Information100.P50.M
local notation "Map" => QAlgorithms.Papers.Independent.Information100.P50.Lin
open QAlgorithms.Papers.Independent.Dependency50.P49 (traceDual)
/-- PDF p.17: scalar invariance and normalization of the positive domain.
For positive definite S, S <= I is equivalent to all eigenvalues belonging to (0,1].
Zero coefficient dimension is included, with empty objective equal to zero. -/
theorem positive_scaling {n k p : ℕ} (hn : 0 < n) (hk : 0 < k)
(E : Fin p → Matrix (Fin k) (Fin n) ℂ)
(F : Fin p → Matrix (Fin n) (Fin k) ℂ) :
(∀ S : Mat p, S.PosDef → ∀ c : ℝ, 0 < c →
changeCost E F ((c : ℂ) • S) = changeCost E F S) ∧
(∀ S : Mat p, S.PosDef → ∃ c : ℝ, 0 < c ∧
((c : ℂ) • S).PosDef ∧ (1 - (c : ℂ) • S).PosSemidef ∧
changeCost E F ((c : ℂ) • S) = changeCost E F S) ∧
sInf {v : ℝ≥0∞ | ∃ S : Mat p, S.PosDef ∧
v = ENNReal.ofReal (changeCost E F S)} =
sInf {v : ℝ≥0∞ | ∃ S : Mat p, S.PosDef ∧ (1 - S).PosSemidef ∧
v = ENNReal.ofReal (changeCost E F S)} := by sorry
end QAlgorithms.Papers.Independent.Intakenext100.P36
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
For every pair of positive natural numbers , every natural number , and arbitrary families and , indexed by , define, for each matrix ,
where is the Euclidean operator norm. The declaration asserts all three of the following: every positive-definite and every real satisfy ; for every positive-definite , there exists a real such that is positive definite, is positive semidefinite, and ; and
with both infima taken in the extended nonnegative reals and . No independence or representation condition is imposed on either family. The case is included: all displayed sums are empty, the cost is zero, the unique empty matrix satisfies the stated positivity conditions, and both infima are zero.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.