Convexity from a Nonnegative Hessian Lower Bound
ProvedFunctionalInequalities.Analysis.BakryEmery.convexOn_of_hessian_ge_readablefunctional-inequalitieskls-conjecturekls-modularkls-readablelog-concave
This is the standard convexity implication from a pointwise Hessian lower bound.
Let be a real inner-product space and let be . Let and suppose that for every ,
Then is convex on all of :
The Hessian is represented by bilinearIteratedFDerivTwo, and no completeness or finite-dimensionality assumption is required. The result may be used to convert curvature-type Hessian hypotheses into global convexity.
Original Lean declaration header
Within namespace FunctionalInequalities.Analysis.BakryEmery. Shown in its original module context. Proof omitted.
theorem convexOn_of_hessian_ge {U : E → ℝ} (hU : ContDiff ℝ 2 U) {ρ : ℝ} (hρ : 0 ≤ ρ)
(hH : ∀ x v, ρ * ‖v‖ ^ 2 ≤ bilinearIteratedFDerivTwo ℝ U x v v) :
ConvexOn ℝ Set.univ U
Module navigation
Module: Analysis/BakryEmery/Curvature.
Next: integrable_weightedLaplacian_gamma.
Source excerpt, lines 39–41 · Frozen source 408915f.
Preamble
import Definitions.Def_KLSModular_FunctionalInequalities_Analysis_BakryEmery_Curvature
universe u
variable {E : Type u} [NormedAddCommGroup E] [InnerProductSpace ℝ E] [CompleteSpace E]
[FiniteDimensional ℝ E]
omit [CompleteSpace E] [FiniteDimensional ℝ E]
Formal statement
theorem FunctionalInequalities.Analysis.BakryEmery.convexOn_of_hessian_ge_readable {U : E → ℝ} (hU : ContDiff ℝ 2 U) {ρ : ℝ} (hρ : 0 ≤ ρ)
(hH : ∀ x v, ρ * ‖v‖ ^ 2 ≤ bilinearIteratedFDerivTwo ℝ U x v v) :
ConvexOn ℝ Set.univ U := by sorry
Source
Lean source: https://github.com/human-ai-mathematics/kls-lean/blob/408915f/formalization/FunctionalInequalities/Analysis/BakryEmery/Curvature.lean#L37-L99 (repository human-ai-mathematics/kls-lean, commit 408915f, module `FunctionalInequalities.Analysis.BakryEmery.Curvature`, declaration `FunctionalInequalities.Analysis.BakryEmery.convexOn_of_hessian_ge`; the repository is private at upload time).