Weighted Brunn-Minkowski inequality on the real line
ProvedConvexOptimization.brunn_minkowski_real_line_weightedanalysisconvex-geometryinequalitiesmeasure-theory
Let , and let be nonempty Lebesgue-measurable sets. Their weighted Minkowski sum satisfies
This is the one-dimensional Brunn–Minkowski inequality used as the geometric input in the first layer-cake proof of the Prékopa–Leindler inequality.
Formalization Note Measures take values in the extended nonnegative reals. On the right, volume evaluates the outer measure of the sum even when that sum has not separately been proved measurable; this removes the source sum-measurability side condition. The source notes that its boundedness assumption is inessential, and the formal theorem omits it.
Preamble
import Mathlib open scoped RealInnerProductSpace ENNReal Pointwise open MeasureTheory Set
Formal statement
theorem ConvexOptimization.brunn_minkowski_real_line_weighted
(l : ℝ) (hl0 : 0 < l) (hl1 : l < 1)
(A B : Set ℝ) (hA : MeasurableSet A) (hB : MeasurableSet B)
(hAn : A.Nonempty) (hBn : B.Nonempty) :
ENNReal.ofReal (1 - l) * volume A + ENNReal.ofReal l * volume B ≤
volume ((1 - l) • A + l • B) := by
sorrySource
R. J. Gardner, The Brunn-Minkowski Inequality, https://faculty.gardner.wwu.edu/gorizia12.pdf, Theorem 2.1, equation (2), PDF p. 3; see also the compact-approximation proof there and the remarks immediately following the theorem that boundedness is inessential and the Minkowski sum may be nonmeasurable.