Proposition 11.12.12. (Gaussian boundary limit for a finite maximum). Let be a finite-dimensional Euclidean space, let be linearly independent with , and put , and . Let be orthogonal projection. Suppose Set , and for . This set of active indices is nonempty. If and are the standard normalized Gaussian densities and , then The convergence is uniform for in compact subsets of . No monotonicity of the approaching parameters is assumed. Moreover for . If for a nonzero , the inequality is strict.
Argument in the manuscript
Proof. Every is nonpositive. Since , a maximizing index there has all these coefficients zero, proving that is nonempty. For , every inactive linear form in tends to as ; active forms annihilate . Consequently, for each fixed , Write . The function is -Lipschitz and subadditive. For , its nonnegativity and give the integrable bound Dominated convergence now gives the limit of the integral. Orthogonal decomposition factors the normalized Gaussian into the product of the two Gaussian densities; integration in gives exactly (11.12.3). Both sides, as functions of , have Lipschitz constant at most . A finite net on a compact subset of upgrades pointwise convergence to uniform convergence. Equivalently, any failure of uniform convergence would give a sequence of parameters and a convergent sequence of points contradicting that bound and the already proved pointwise limit.
Finally, , and integration gives monotonicity. If , the difference is a nonnegative continuous function of and is positive at : its value there is . It is therefore positive on a nonempty open set of positive Gaussian measure. Its integral is strictly positive. This proves strictness without a separate strict-convexity argument. ◻