TheoremBase

Existence comes from the extreme value theorem on the closed ball; localisation from comparing the value at a maximiser with the value at the centre against the strictly smaller maximum of the function on the closed annulus; compactness from Bolzano-Weierstrass applied to the perturbing vectors together with the sequential characterisation of closed sets.

Proof

The ball Bˉ\bar{B} is nonempty, since x^∈Bˉ\hat{x}\in\bar{B} by claim 1 of Elementary Properties of the Closed Ball in a Metric Space, and compact in Rn\mathbb{R}^{n} by claim 2 of A Closed Euclidean Ball is Convex and Compact. It is bounded and closed by claims 2 and 3 of Elementary Properties of the Closed Ball in a Metric Space. For p∈Rnp\in\mathbb{R}^{n} let ψp:Bˉ→R\psi_{p}:\bar{B}\to\mathbb{R} be given by ψp(x)=φ(x)+p⋅x\psi_{p}(x)=\varphi(x)+p\cdot x; thus M(p)M(p) is the set of points of Bˉ\bar{B} at which ψp\psi_{p} attains its greatest value on Bˉ\bar{B}. Throughout we use the bilinearity and symmetry of the dot product (The Dot Product on Rn\mathbb{R}^n Is Symmetric, Bilinear and Positive Definite) and the Cauchy-Schwarz inequality The Cauchy-Schwarz Inequality on Rn\mathbb{R}^n.

Step 1 (the perturbed functions are continuous). Let p∈Rnp\in\mathbb{R}^{n} and S⊆BˉS\subseteq\bar{B}. We check that the restriction of ψp\psi_{p} to SS satisfies the continuity hypothesis of Extreme Value Theorem on a Compact Subset of a Metric Space. Let x∈Sx\in S and let ε∈R\varepsilon\in\mathbb{R} with 0<ε0<\varepsilon. For y∈Bˉy\in\bar{B},

∣ψp(y)−ψp(x)∣≤∣φ(y)−φ(x)∣+∣p⋅(y−x)∣≤∣φ(y)−φ(x)∣+∥p∥ ∥y−x∥.\bigl|\psi_{p}(y)-\psi_{p}(x)\bigr|\le\bigl|\varphi(y)-\varphi(x)\bigr|+\bigl|p\cdot(y-x)\bigr|\le\bigl|\varphi(y)-\varphi(x)\bigr|+\lVert p\rVert\,\lVert y-x\rVert .

By (H1) there is a positive real δ1\delta_{1} such that every z∈Bˉz\in\bar{B} with ∥z−x∥<δ1\lVert z-x\rVert<\delta_{1} satisfies ∣φ(z)−φ(x)∣<ε/2|\varphi(z)-\varphi(x)|<\varepsilon/2. Let δ\delta be the smaller of δ1\delta_{1} and ε/(2∥p∥+2)\varepsilon/(2\lVert p\rVert+2), a positive real number. If y∈Sy\in S satisfies dE(x,y)=∥y−x∥<δd_{E}(x,y)=\lVert y-x\rVert<\delta, then

∣ψp(y)−ψp(x)∣<ε2+∥p∥ ε2∥p∥+2≤ε2+ε2=ε,\bigl|\psi_{p}(y)-\psi_{p}(x)\bigr|<\frac{\varepsilon}{2}+\lVert p\rVert\,\frac{\varepsilon}{2\lVert p\rVert+2}\le\frac{\varepsilon}{2}+\frac{\varepsilon}{2}=\varepsilon,

since 2∥p∥≤2∥p∥+22\lVert p\rVert\le 2\lVert p\rVert+2 gives ∥p∥/(2∥p∥+2)≤12\lVert p\rVert/(2\lVert p\rVert+2)\le\tfrac12.

Proof of claim 1. Let p∈Rnp\in\mathbb{R}^{n}. By step 1 with S=BˉS=\bar{B}, the function ψp\psi_{p} satisfies the hypotheses of Extreme Value Theorem on a Compact Subset of a Metric Space on the nonempty compact set Bˉ\bar{B}, so there is xmax⁡∈Bˉx_{\max}\in\bar{B} with ψp(y)≤ψp(xmax⁡)\psi_{p}(y)\le\psi_{p}(x_{\max}) for every y∈Bˉy\in\bar{B}; that is, xmax⁡∈M(p)x_{\max}\in M(p) and M(p)≠∅M(p)\ne\varnothing. Now let δ\delta be a positive real number. The origin 0Rn0_{\mathbb{R}^{n}} satisfies ∥0Rn∥=0≤δ\lVert 0_{\mathbb{R}^{n}}\rVert=0\le\delta by claim 3 of Properties of the Euclidean Norm on Rn\mathbb{R}^n: Square, Distance, Vanishing, Coordinate Bounds, Homogeneity, Triangle Inequalities and Expansion of a Square, so any element of M(0Rn)M(0_{\mathbb{R}^{n}}) belongs to KδK_{\delta}, and Kδ≠∅K_{\delta}\ne\varnothing.

Proof of claim 2. Let ρ∈R\rho\in\mathbb{R} with 0<ρ0<\rho.

Suppose first that r<ρr<\rho. Every x∈Bˉx\in\bar{B} satisfies dE(x^,x)≤r<ρd_{E}(\hat{x},x)\le r<\rho, so Bˉ⊆BdE(x^,ρ)\bar{B}\subseteq B_{d_{E}}(\hat{x},\rho); since Kδ⊆BˉK_{\delta}\subseteq\bar{B} for every positive real δ\delta, the choice δρ=1\delta_{\rho}=1 works.

Suppose now that ρ≤r\rho\le r, and put

A={x∈Bˉ:ρ≤∥x−x^∥}.A=\bigl\{x\in\bar{B}:\rho\le\lVert x-\hat{x}\rVert\bigr\}.

The set AA is nonempty: let e1∈Rne_{1}\in\mathbb{R}^{n} have first component 11 and all other components 00, so that ∥e1∥2=1\lVert e_{1}\rVert^{2}=1 and hence ∥e1∥=1\lVert e_{1}\rVert=1 by claim 1 of Properties of the Euclidean Norm on Rn\mathbb{R}^n: Square, Distance, Vanishing, Coordinate Bounds, Homogeneity, Triangle Inequalities and Expansion of a Square; then z=x^+ρ e1z=\hat{x}+\rho\,e_{1} satisfies ∥z−x^∥=∥ρ e1∥=ρ\lVert z-\hat{x}\rVert=\lVert\rho\,e_{1}\rVert=\rho by claim 5 there, so dE(x^,z)=ρ≤rd_{E}(\hat{x},z)=\rho\le r and z∈Az\in A.

The set AA is compact. Indeed A=Bˉ∩CA=\bar{B}\cap C where C=Rn∖BdE(x^,ρ)C=\mathbb{R}^{n}\setminus B_{d_{E}}(\hat{x},\rho), because dE(x^,x)=∥x−x^∥d_{E}(\hat{x},x)=\lVert x-\hat{x}\rVert and BdE(x^,ρ)B_{d_{E}}(\hat{x},\rho) consists of the points at distance less than ρ\rho from x^\hat{x}. The open ball is open by Open Ball in a Metric Space is Open, so CC is closed; Bˉ\bar{B} is closed; and the intersection of two closed sets is closed by Complements, Unions and Intersections of Closed Sets in a Topological Space. Moreover A⊆BˉA\subseteq\bar{B} is bounded, directly from Bounded Subset of a Metric Space. By The Heine-Borel Theorem in Rn\mathbb{R}^n, and Compactness of Closed Balls and the Unit Sphere, AA is compact.

By step 1 with p=0Rnp=0_{\mathbb{R}^{n}} and S=AS=A, and by Extreme Value Theorem on a Compact Subset of a Metric Space, there is xA∈Ax_{A}\in A with φ(x)≤φ(xA)\varphi(x)\le\varphi(x_{A}) for every x∈Ax\in A. Since 0<ρ≤∥xA−x^∥0<\rho\le\lVert x_{A}-\hat{x}\rVert we have xA≠x^x_{A}\ne\hat{x}, so (H2) gives φ(xA)<φ(x^)\varphi(x_{A})<\varphi(\hat{x}). Put

η=φ(x^)−φ(xA)>0,δρ=η2r>0.\eta=\varphi(\hat{x})-\varphi(x_{A})>0,\qquad \delta_{\rho}=\frac{\eta}{2r}>0 .

Let δ\delta be real with 0<δ≤δρ0<\delta\le\delta_{\rho} and let x∈Kδx\in K_{\delta}, say x∈M(p)x\in M(p) with ∥p∥≤δ\lVert p\rVert\le\delta. Since x^∈Bˉ\hat{x}\in\bar{B} we have φ(x^)+p⋅x^≤φ(x)+p⋅x\varphi(\hat{x})+p\cdot\hat{x}\le\varphi(x)+p\cdot x, hence, using Cauchy-Schwarz and ∥x−x^∥=dE(x^,x)≤r\lVert x-\hat{x}\rVert=d_{E}(\hat{x},x)\le r,

φ(x) ≥ φ(x^)−p⋅(x−x^) ≥ φ(x^)−∥p∥ ∥x−x^∥ ≥ φ(x^)−δ r ≥ φ(x^)−η2 > φ(x^)−η=φ(xA).\varphi(x)\ \ge\ \varphi(\hat{x})-p\cdot(x-\hat{x})\ \ge\ \varphi(\hat{x})-\lVert p\rVert\,\lVert x-\hat{x}\rVert\ \ge\ \varphi(\hat{x})-\delta\,r\ \ge\ \varphi(\hat{x})-\frac{\eta}{2}\ >\ \varphi(\hat{x})-\eta=\varphi(x_{A}).

So φ(xA)<φ(x)\varphi(x_{A})<\varphi(x), which rules out x∈Ax\in A; as x∈Bˉx\in\bar{B}, this forces ∥x−x^∥<ρ\lVert x-\hat{x}\rVert<\rho, that is, x∈BdE(x^,ρ)x\in B_{d_{E}}(\hat{x},\rho). Hence Kδ⊆BdE(x^,ρ)K_{\delta}\subseteq B_{d_{E}}(\hat{x},\rho), proving claim 2.

Proof of claim 3. Let δ\delta be a positive real number. Since Kδ⊆BˉK_{\delta}\subseteq\bar{B}, the set KδK_{\delta} is bounded. We show that it is closed, using Sequential Characterization of Closed Subsets of a Metric Space.

Let (x(q))q∈N(x^{(q)})_{q\in\mathbb{N}} be a sequence with x(q)∈Kδx^{(q)}\in K_{\delta} for every qq, converging to a point x∈Rnx\in\mathbb{R}^{n}, and for each qq choose p(q)∈Rnp^{(q)}\in\mathbb{R}^{n} with ∥p(q)∥≤δ\lVert p^{(q)}\rVert\le\delta and x(q)∈M(p(q))x^{(q)}\in M(p^{(q)}). As Bˉ\bar{B} is closed, x∈Bˉx\in\bar{B}.

The closed ball BˉdE(0Rn,δ)\bar{B}_{d_{E}}(0_{\mathbb{R}^{n}},\delta) is bounded and closed by claims 2 and 3 of Elementary Properties of the Closed Ball in a Metric Space, and ∥p(q)∥=dE(0Rn,p(q))≤δ\lVert p^{(q)}\rVert=d_{E}(0_{\mathbb{R}^{n}},p^{(q)})\le\delta places every p(q)p^{(q)} in it. By The Bolzano-Weierstrass Theorem in Rn\mathbb{R}^n there are p∈Rnp\in\mathbb{R}^{n} and a strictly increasing sequence (qt)t∈N(q_{t})_{t\in\mathbb{N}} in N\mathbb{N} such that (p(qt))t∈N(p^{(q_{t})})_{t\in\mathbb{N}} converges to pp; by Sequential Characterization of Closed Subsets of a Metric Space, p∈BˉdE(0Rn,δ)p\in\bar{B}_{d_{E}}(0_{\mathbb{R}^{n}},\delta), that is, ∥p∥≤δ\lVert p\rVert\le\delta. The subsequence (x(qt))t∈N(x^{(q_{t})})_{t\in\mathbb{N}} converges to xx.

Fix y∈Bˉy\in\bar{B} and let ε∈R\varepsilon\in\mathbb{R} with 0<ε0<\varepsilon. By (H1) there is a positive real δ2\delta_{2} such that every z∈Bˉz\in\bar{B} with ∥z−x∥<δ2\lVert z-x\rVert<\delta_{2} satisfies ∣φ(z)−φ(x)∣<ε/4|\varphi(z)-\varphi(x)|<\varepsilon/4. By convergence, choose t∈Nt\in\mathbb{N} so large that

∥x(qt)−x∥<δ2,∥x(qt)−x∥<ε4δ+4,∥p(qt)−p∥<ε4∥x∥+4∥y∥+4.\lVert x^{(q_{t})}-x\rVert<\delta_{2},\qquad \lVert x^{(q_{t})}-x\rVert<\frac{\varepsilon}{4\delta+4},\qquad \lVert p^{(q_{t})}-p\rVert<\frac{\varepsilon}{4\lVert x\rVert+4\lVert y\rVert+4}.

Writing p(qt)⋅x(qt)−p⋅x=p(qt)⋅(x(qt)−x)+(p(qt)−p)⋅xp^{(q_{t})}\cdot x^{(q_{t})}-p\cdot x=p^{(q_{t})}\cdot(x^{(q_{t})}-x)+(p^{(q_{t})}-p)\cdot x and using Cauchy-Schwarz,

∣ψp(qt)(x(qt))−ψp(x)∣≤∣φ(x(qt))−φ(x)∣+δ ∥x(qt)−x∥+∥p(qt)−p∥ ∥x∥<3ε4,\bigl|\psi_{p^{(q_{t})}}(x^{(q_{t})})-\psi_{p}(x)\bigr|\le\bigl|\varphi(x^{(q_{t})})-\varphi(x)\bigr|+\delta\,\lVert x^{(q_{t})}-x\rVert+\lVert p^{(q_{t})}-p\rVert\,\lVert x\rVert<\frac{3\varepsilon}{4},

while ∣p(qt)⋅y−p⋅y∣≤∥p(qt)−p∥ ∥y∥<ε/4\bigl|p^{(q_{t})}\cdot y-p\cdot y\bigr|\le\lVert p^{(q_{t})}-p\rVert\,\lVert y\rVert<\varepsilon/4. Since x(qt)∈M(p(qt))x^{(q_{t})}\in M(p^{(q_{t})}) and y∈Bˉy\in\bar{B},

φ(y)+p(qt)⋅y ≤ φ(x(qt))+p(qt)⋅x(qt),\varphi(y)+p^{(q_{t})}\cdot y\ \le\ \varphi(x^{(q_{t})})+p^{(q_{t})}\cdot x^{(q_{t})} ,

and combining the three displays,

φ(y)+p⋅y < φ(y)+p(qt)⋅y+ε4 ≤ φ(x(qt))+p(qt)⋅x(qt)+ε4 < φ(x)+p⋅x+ε.\varphi(y)+p\cdot y\ <\ \varphi(y)+p^{(q_{t})}\cdot y+\frac{\varepsilon}{4}\ \le\ \varphi(x^{(q_{t})})+p^{(q_{t})}\cdot x^{(q_{t})}+\frac{\varepsilon}{4}\ <\ \varphi(x)+p\cdot x+\varepsilon .

As ε\varepsilon was an arbitrary positive real number, φ(y)+p⋅y≤φ(x)+p⋅x\varphi(y)+p\cdot y\le\varphi(x)+p\cdot x: were the reverse strict inequality to hold, taking ε\varepsilon to be half of the positive difference would contradict the display. Since y∈Bˉy\in\bar{B} was arbitrary, x∈M(p)x\in M(p) with ∥p∥≤δ\lVert p\rVert\le\delta, so x∈Kδx\in K_{\delta}.

By Sequential Characterization of Closed Subsets of a Metric Space the set KδK_{\delta} is closed, and being bounded it is compact by The Heine-Borel Theorem in Rn\mathbb{R}^n, and Compactness of Closed Balls and the Unit Sphere. This proves claim 3.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…