The Lagrange multiplier theorem
ProvedFamousTheorems.lagrange_multiplier_theoremThe Lagrange multiplier theorem. Let be a real Banach space, , and let for in a finite index set. Suppose is a local extremum of on the level set , and that and all are strictly differentiable at . Then there are multipliers and , not all zero, with
This is the multiplier rule of constrained optimization. It is due to Lagrange for mechanics, and in this form, with a possibly vanishing , to Fritz John. When the are linearly independent one can take , which gives the familiar condition .
Formalization note. Mathlib's IsLocalExtrOn.exists_multipliers_of_hasStrictFDerivAt. Derivatives are elements of StrongDual ℝ E, the continuous dual of . The condition "not all zero" is (Λ, Λ₀) ≠ 0 for the pair in .
Preamble
import Mathlib
Formal statement
namespace FamousTheorems
theorem lagrange_multiplier_theorem {E : Type*} [NormedAddCommGroup E] [NormedSpace ℝ E] [CompleteSpace E] {φ : E → ℝ} {x₀ : E}
{φ' : StrongDual ℝ E} {ι : Type*} [Fintype ι] {f : ι → E → ℝ} {f' : ι → StrongDual ℝ E}
(hextr : IsLocalExtrOn φ {x | ∀ i, f i x = f i x₀} x₀) (hf' : ∀ i, HasStrictFDerivAt (f i) (f' i) x₀)
(hφ' : HasStrictFDerivAt φ φ' x₀) :
∃ (Λ : ι → ℝ) (Λ₀ : ℝ), (Λ, Λ₀) ≠ 0 ∧ ∑ i, Λ i • f' i + Λ₀ • φ' = 0 := by sorry
end FamousTheoremsSource
Listed in Mathlib's curated theorem manifests (docs/1000.yaml); formalized in Mathlib as `IsLocalExtrOn.exists_multipliers_of_hasStrictFDerivAt`. Proof here reduces to that Mathlib result.