Lemma 3.6, p. 270, unconstrained (X = ℝⁿ) — f(x − ∇f(x)/β) − f(y) ≤ ∇f(x)⊤(x − y) − ‖∇f(x)‖²/(2β)
OpenConvexOptAlg.NesterovSmooth.lemma_3_6_unconstrainedconvex-optimizationgradient-stepp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-bookp2o-v1smoothness
Let be convex and -smooth with . Then for all ,
This is Lemma 3.6 of the book in the case : the projection is the identity, so and the gradient mapping equals . The proof of Theorem 3.19 applies it twice per iteration, once with and once with .
Formalization Note The gradient is an explicit map with ; convexity is Mathlib's ConvexOn ℝ Set.univ f. The hypothesis is implicit in the book (the step ) and is stated.
Preamble
import Mathlib import Definitions.Def_ConvexOptAlg_NesterovSmooth_Defs open scoped InnerProductSpace
Formal statement
namespace ConvexOptAlg.NesterovSmooth
/-- Lemma 3.6 (Bubeck, arXiv:1405.4980v2, p. 270) in its unconstrained version (X = ℝⁿ), as used in
the proof of Theorem 3.19 (p. 294): for a convex β-smooth `f` on `ℝⁿ` with gradient map `g`,
`x⁺ = x − (1/β)∇f(x)` and `g_X(x) = β(x − x⁺) = ∇f(x)`, for all `x, y`,
`f(x⁺) − f(y) ≤ ∇f(x)⊤(x − y) − (1/(2β))‖∇f(x)‖²`. -/
theorem lemma_3_6_unconstrained {n : ℕ} (f : EuclideanSpace ℝ (Fin n) → ℝ)
(g : EuclideanSpace ℝ (Fin n) → EuclideanSpace ℝ (Fin n)) (β : ℝ) (hβ : 0 < β)
(hconv : ConvexOn ℝ Set.univ f) (hf : IsBetaSmooth f g β) (x y : EuclideanSpace ℝ (Fin n)) :
f (x - (1 / β) • g x) - f y ≤ ⟪g x, x - y⟫_ℝ - 1 / (2 * β) * ‖g x‖ ^ 2 := by sorry
end ConvexOptAlg.NesterovSmooth
Source
Bubeck, arXiv:1405.4980v2, Lemma 3.6, p. 270, in the unconstrained form used in the proof of Theorem 3.19, p. 294