Documentation

IntroductoryLecturesOnConvexOptimization_Nesterov_2004.Chap01.Definition_1_4_17

noncomputable def secondOrderTaylorModelAt {E : Type u} [NormedAddCommGroup E] [InnerProductSpace ℝ E] [CompleteSpace E] (f : E → ℝ) (x : E) :
E → ℝ

Definition 1.4.17: the second-order Taylor model of f at x.

Instances For
    @[simp]
    theorem secondOrderTaylorModelAt_apply {E : Type u} [NormedAddCommGroup E] [InnerProductSpace ℝ E] [CompleteSpace E] (f : E → ℝ) (x y : E) :
    secondOrderTaylorModelAt f x y = f x + inner ℝ (gradient f x) (y - x) + 1 / 2 * inner ℝ ((hessian f x) (y - x)) (y - x)

    Evaluating secondOrderTaylorModelAt f x recovers the displayed quadratic formula.

    theorem secondOrderTaylorModelAt_apply_hessianMatrix {n : ℕ} (f : EuclideanSpace ℝ (Fin n) → ℝ) (x y : EuclideanSpace ℝ (Fin n)) :
    secondOrderTaylorModelAt f x y = f x + inner ℝ (gradient f x) (y - x) + 1 / 2 * inner ℝ ((Matrix.toEuclideanLin (hessianMatrix f x)) (y - x)) (y - x)

    On ℝⁿ, the quadratic Taylor model can be rewritten using the Hessian matrix ∇² f x acting on the displacement y - x.