Sums of self-concordant functions
ProvedConvexOptimization.self_concordant_addconvex-optimizationinterior-pointself-concordance
Self-concordance is preserved by addition.
Let be open and let and both be self-concordant on — convex, , and satisfying for every line restriction with . Then is self-concordant on :
The proof is the elementary inequality for the pieces, and the practical consequence is a calculus: a barrier assembled from many self-concordant terms is self-concordant, so one never verifies the third-derivative condition for a composite function directly.
Formalization Note Openness of Ω is an explicit hypothesis, and it is genuinely needed: at a boundary point the iterated derivatives of the line restriction are junk values and the inequality would be asserted about meaningless quantities. Source: B&V §9.6.1, p. 497.
Preamble
import Mathlib import Definitions.Def_ConvexOptimization_selfConcordance open scoped RealInnerProductSpace ENNReal open MeasureTheory
Formal statement
theorem ConvexOptimization.self_concordant_add {n : ℕ} (Ω : Set (EuclideanSpace ℝ (Fin n)))
(hΩo : IsOpen Ω) (f h : EuclideanSpace ℝ (Fin n) → ℝ)
(hf : IsSelfConcordantOn Ω f) (hh : IsSelfConcordantOn Ω h) :
IsSelfConcordantOn Ω (f + h) := by
sorry
Source
Boyd & Vandenberghe 2004, Convex Optimization, Cambridge University Press (seventh printing with corrections, 2009), https://web.stanford.edu/~boyd/cvxbook/, pp. 499, §9.6.1 (self-concordance is preserved by addition). Openness of the domain is an explicit hypothesis: the line-restriction derivatives carry junk values at boundary points