TheoremBase

Proof

By Convex Subset of Rn\mathbb{R}^n we must show that t x+(1−t) y∈Ωδt\,x+(1-t)\,y\in\Omega^{\delta} whenever x,y∈Ωδx,y\in\Omega^{\delta} and t∈Rt\in\mathbb{R} satisfies 0≤t0\le t and t≤1t\le 1. (If Ωδ\Omega^{\delta} is empty there is nothing to prove.)

Fix such xx, yy and tt and put z=t x+(1−t) yz=t\,x+(1-t)\,y. By the definition of Ωδ\Omega^{\delta} it suffices to prove that Bˉ(z,δ)⊆Ω\bar B(z,\delta)\subseteq\Omega.

Let w∈Bˉ(z,δ)w\in\bar B(z,\delta), so that d(z,w)≤δd(z,w)\le\delta by Closed Ball in a Metric Space, and put v=w−zv=w-z. By claim 2 of Properties of the Euclidean Norm on Rn\mathbb{R}^n: Square, Distance, Vanishing, Coordinate Bounds, Homogeneity, Triangle Inequalities and Expansion of a Square we have ∥v∥=∥w−z∥=d(w,z)\lVert v\rVert=\lVert w-z\rVert=d(w,z), and d(w,z)=d(z,w)d(w,z)=d(z,w) by the symmetry clause 3 of a metric; hence ∥v∥≤δ\lVert v\rVert\le\delta.

Step 1 (the translates of xx and yy lie in Ω\Omega). By claim 2 of Properties of the Euclidean Norm on Rn\mathbb{R}^n: Square, Distance, Vanishing, Coordinate Bounds, Homogeneity, Triangle Inequalities and Expansion of a Square, d(x,x+v)=∥v∥≤δd(x,x+v)=\lVert v\rVert\le\delta, so x+v∈Bˉ(x,δ)x+v\in\bar B(x,\delta); since x∈Ωδx\in\Omega^{\delta} we have Bˉ(x,δ)⊆Ω\bar B(x,\delta)\subseteq\Omega, whence x+v∈Ωx+v\in\Omega. The same argument gives y+v∈Ωy+v\in\Omega.

Step 2 (ww is the corresponding convex combination). In the real vector space Rn\mathbb{R}^{n} of Euclidean Space Rn\mathbb{R}^n is a Real Vector Space, distributivity of scalar multiplication over vector addition together with associativity and commutativity of addition gives

t (x+v)+(1−t) (y+v)=(t x+(1−t) y)+(t v+(1−t) v),t\,(x+v)+(1-t)\,(y+v)=\bigl(t\,x+(1-t)\,y\bigr)+\bigl(t\,v+(1-t)\,v\bigr),

and distributivity of scalar multiplication over addition of scalars gives t v+(1−t) v=(t+(1−t))v=1 v=vt\,v+(1-t)\,v=\bigl(t+(1-t)\bigr)v=1\,v=v. Hence

t (x+v)+(1−t) (y+v)=z+v.t\,(x+v)+(1-t)\,(y+v)=z+v.

Moreover v=w−z=w+(−z)v=w-z=w+(-z) by claim 3 of Euclidean Space Rn\mathbb{R}^n is a Real Vector Space, so z+v=z+(w+(−z))=w+(z+(−z))=wz+v=z+\bigl(w+(-z)\bigr)=w+\bigl(z+(-z)\bigr)=w by associativity, commutativity, claim 2 of Euclidean Space Rn\mathbb{R}^n is a Real Vector Space and claim 1 of the same result, which identifies the zero vector.

Step 3 (conclusion). By Steps 1 and 2, ww is the convex combination t (x+v)+(1−t) (y+v)t\,(x+v)+(1-t)\,(y+v) of the two points x+v,y+v∈Ωx+v,y+v\in\Omega; since Ω\Omega is convex, w∈Ωw\in\Omega. As w∈Bˉ(z,δ)w\in\bar B(z,\delta) was arbitrary, Bˉ(z,δ)⊆Ω\bar B(z,\delta)\subseteq\Omega, that is, z∈Ωδz\in\Omega^{\delta}. Therefore Ωδ\Omega^{\delta} is convex.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…