Each result cited below is universally quantified over the data in its own statement, and is applied to the data named where it is used. Throughout, the vector operations of R m \mathbb{R}^{m} R m and their identities are those of Euclidean Space R n \mathbb{R}^n R n is a Real Vector Space , and the claims of Properties of the Euclidean Norm on R n \mathbb{R}^n R n : Square, Distance, Vanishing, Coordinate Bounds, Homogeneity, Triangle Inequalities and Expansion of a Square are used for the Euclidean norm: claim 2 for d E ( x , y ) = ∥ x − y ∥ d_{E}(x,y)=\lVert x-y\rVert d E ( x , y ) = ∥ x − y ∥ , claim 3 for the vanishing of the norm, claim 5 for ∥ λ x ∥ = ∣ λ ∣ ∥ x ∥ \lVert\lambda x\rVert=|\lambda|\,\lVert x\rVert ∥ λ x ∥ = ∣ λ ∣ ∥ x ∥ and claim 6 for the triangle inequality.
Claim 1. Let x , y ∈ R m x,y\in\mathbb{R}^{m} x , y ∈ R m . From x = ( x − y ) + y x=(x-y)+y x = ( x − y ) + y and the triangle inequality, ∥ x ∥ ≤ ∥ x − y ∥ + ∥ y ∥ \lVert x\rVert\le\lVert x-y\rVert+\lVert y\rVert ∥ x ∥ ≤ ∥ x − y ∥ + ∥ y ∥ , hence ∥ x ∥ − ∥ y ∥ ≤ ∥ x − y ∥ \lVert x\rVert-\lVert y\rVert\le\lVert x-y\rVert ∥ x ∥ − ∥ y ∥ ≤ ∥ x − y ∥ by claim 3 of Elementary Arithmetic in an Ordered Field . Exchanging x x x and y y y gives ∥ y ∥ − ∥ x ∥ ≤ ∥ y − x ∥ \lVert y\rVert-\lVert x\rVert\le\lVert y-x\rVert ∥ y ∥ − ∥ x ∥ ≤ ∥ y − x ∥ , and ∥ y − x ∥ = ∥ ( − 1 ) ( x − y ) ∥ = ∣ − 1 ∣ ∥ x − y ∥ = ∥ x − y ∥ \lVert y-x\rVert=\lVert(-1)(x-y)\rVert=|-1|\,\lVert x-y\rVert=\lVert x-y\rVert ∥ y − x ∥ = ∥( − 1 ) ( x − y )∥ = ∣ − 1∣ ∥ x − y ∥ = ∥ x − y ∥ by homogeneity. By claim 6 of Properties of the Absolute Value in an Ordered Field the two bounds give
∣ ∥ x ∥ − ∥ y ∥ ∣ ≤ ∥ x − y ∥ = d E ( x , y ) , \bigl|\,\lVert x\rVert-\lVert y\rVert\,\bigr|\le\lVert x-y\rVert=d_{E}(x,y), ∥ x ∥ − ∥ y ∥ ≤ ∥ x − y ∥ = d E ( x , y ) ,
that is, d R ( N ( x ) , N ( y ) ) ≤ 1 ⋅ d E ( x , y ) d_{\mathbb{R}}(N(x),N(y))\le 1\cdot d_{E}(x,y) d R ( N ( x ) , N ( y )) ≤ 1 ⋅ d E ( x , y ) . So N N N is Lipschitz with constant 1 1 1 in the sense of Lipschitz Map Between Metric Spaces ; it is uniformly continuous on R m \mathbb{R}^{m} R m by A Lipschitz Map is Uniformly Continuous and Continuous , hence continuous on R m \mathbb{R}^{m} R m by A Uniformly Continuous Map Between Metric Spaces Is Continuous .
Claim 2. For y ∈ R m y\in\mathbb{R}^{m} y ∈ R m one has d E ( 0 , y ) = ∥ 0 − y ∥ = ∥ y ∥ d_{E}(0,y)=\lVert 0-y\rVert=\lVert y\rVert d E ( 0 , y ) = ∥ 0 − y ∥ = ∥ y ∥ , so B ˉ \bar{B} B ˉ and U U U are exactly the closed ball and the open ball of ( R m , d E ) (\mathbb{R}^{m},d_{E}) ( R m , d E ) with centre 0 0 0 and radius 1 1 1 , in the sense of Closed Ball in a Metric Space and Open Ball in a Metric Space . Since ∥ 0 ∥ = 0 ≤ 1 \lVert 0\rVert=0\le 1 ∥ 0 ∥ = 0 ≤ 1 , the set B ˉ \bar{B} B ˉ contains 0 0 0 and is nonempty. By claims 2 and 3 of Elementary Properties of the Closed Ball in a Metric Space , B ˉ \bar{B} B ˉ is bounded in ( R m , d E ) (\mathbb{R}^{m},d_{E}) ( R m , d E ) and closed in ( R m , T ) (\mathbb{R}^{m},\mathcal{T}) ( R m , T ) ; therefore B ˉ \bar{B} B ˉ is compact in ( R m , T ) (\mathbb{R}^{m},\mathcal{T}) ( R m , T ) by The Heine-Borel Theorem in R n \mathbb{R}^n R n , and Compactness of Closed Balls and the Unit Sphere . By claim 2 of The Restriction of a Metric to a Subset Induces the Subspace Topology the subsets of B ˉ \bar{B} B ˉ open in the metric space ( B ˉ , d B ˉ ) (\bar{B},d_{\bar{B}}) ( B ˉ , d B ˉ ) are exactly the members of the subspace topology, so by claim 2 of Compactness in a Subspace Agrees with Compactness in the Ambient Space the metric space ( B ˉ , d B ˉ ) (\bar{B},d_{\bar{B}}) ( B ˉ , d B ˉ ) is compact.
By Open Ball in a Metric Space is Open , U ∈ T U\in\mathcal{T} U ∈ T ; by claim 1 of Elementary Properties of the Closed Ball in a Metric Space , U ⊆ B ˉ U\subseteq\bar{B} U ⊆ B ˉ ; and U = U ∩ B ˉ U=U\cap\bar{B} U = U ∩ B ˉ is open in ( B ˉ , d B ˉ ) (\bar{B},d_{\bar{B}}) ( B ˉ , d B ˉ ) by claim 2 of The Restriction of a Metric to a Subset Induces the Subspace Topology .
Claim 3. Write a x = ( 1 + ∥ x ∥ ) − 1 a_{x}=(1+\lVert x\rVert)^{-1} a x = ( 1 + ∥ x ∥ ) − 1 for x ∈ R m x\in\mathbb{R}^{m} x ∈ R m ; since 0 ≤ ∥ x ∥ 0\le\lVert x\rVert 0 ≤ ∥ x ∥ we have 1 ≤ 1 + ∥ x ∥ 1\le 1+\lVert x\rVert 1 ≤ 1 + ∥ x ∥ by claim 3 of Elementary Arithmetic in an Ordered Field , and 0 < 1 0<1 0 < 1 by claim 6 of Elementary Order Arithmetic in an Ordered Field , so 0 < 1 + ∥ x ∥ 0<1+\lVert x\rVert 0 < 1 + ∥ x ∥ by claim 2 of that lemma. Hence a x a_{x} a x is defined and 0 < a x 0<a_{x} 0 < a x by claim 7 of Elementary Order Arithmetic in an Ordered Field ; and claim 5 of Elementary Arithmetic in an Ordered Field , applied to 1 ≤ 1 + ∥ x ∥ 1\le 1+\lVert x\rVert 1 ≤ 1 + ∥ x ∥ with the nonnegative factor a x a_{x} a x , gives a x = a x ⋅ 1 ≤ a x ( 1 + ∥ x ∥ ) = 1 a_{x}=a_{x}\cdot 1\le a_{x}(1+\lVert x\rVert)=1 a x = a x ⋅ 1 ≤ a x ( 1 + ∥ x ∥) = 1 . By homogeneity, ∥ h ( x ) ∥ = a x ∥ x ∥ \lVert h(x)\rVert=a_{x}\lVert x\rVert ∥ h ( x )∥ = a x ∥ x ∥ , and from ∥ x ∥ < 1 + ∥ x ∥ \lVert x\rVert<1+\lVert x\rVert ∥ x ∥ < 1 + ∥ x ∥ we get a x ∥ x ∥ < 1 a_{x}\lVert x\rVert<1 a x ∥ x ∥ < 1 on multiplying by the positive number a x a_{x} a x , using claim 10 of Elementary Order Arithmetic in an Ordered Field . Hence h ( x ) ∈ U h(x)\in U h ( x ) ∈ U .
Inverse. Let x ∈ R m x\in\mathbb{R}^{m} x ∈ R m , put y = h ( x ) y=h(x) y = h ( x ) and s = ∥ y ∥ = a x ∥ x ∥ s=\lVert y\rVert=a_{x}\lVert x\rVert s = ∥ y ∥ = a x ∥ x ∥ . Then
1 − s = ( 1 + ∥ x ∥ − ∥ x ∥ ) a x = a x , 1-s=\bigl(1+\lVert x\rVert-\lVert x\rVert\bigr)a_{x}=a_{x}, 1 − s = ( 1 + ∥ x ∥ − ∥ x ∥ ) a x = a x ,
so ( 1 − s ) − 1 = 1 + ∥ x ∥ (1-s)^{-1}=1+\lVert x\rVert ( 1 − s ) − 1 = 1 + ∥ x ∥ and g ( y ) = ( 1 + ∥ x ∥ ) a x x = x g(y)=(1+\lVert x\rVert)a_{x}x=x g ( y ) = ( 1 + ∥ x ∥) a x x = x . Thus g ( h ( x ) ) = x g(h(x))=x g ( h ( x )) = x for every x ∈ R m x\in\mathbb{R}^{m} x ∈ R m , so h h h is injective. Conversely let y ∈ U y\in U y ∈ U , put x = g ( y ) x=g(y) x = g ( y ) and t = ∥ x ∥ = ( 1 − ∥ y ∥ ) − 1 ∥ y ∥ t=\lVert x\rVert=(1-\lVert y\rVert)^{-1}\lVert y\rVert t = ∥ x ∥ = ( 1 − ∥ y ∥ ) − 1 ∥ y ∥ , the factor ( 1 − ∥ y ∥ ) − 1 (1-\lVert y\rVert)^{-1} ( 1 − ∥ y ∥ ) − 1 being positive. Then
1 + t = ( 1 − ∥ y ∥ + ∥ y ∥ ) ( 1 − ∥ y ∥ ) − 1 = ( 1 − ∥ y ∥ ) − 1 , 1+t=\bigl(1-\lVert y\rVert+\lVert y\rVert\bigr)(1-\lVert y\rVert)^{-1}=(1-\lVert y\rVert)^{-1}, 1 + t = ( 1 − ∥ y ∥ + ∥ y ∥ ) ( 1 − ∥ y ∥ ) − 1 = ( 1 − ∥ y ∥ ) − 1 ,
so a x = 1 − ∥ y ∥ a_{x}=1-\lVert y\rVert a x = 1 − ∥ y ∥ and h ( x ) = ( 1 − ∥ y ∥ ) ( 1 − ∥ y ∥ ) − 1 y = y h(x)=(1-\lVert y\rVert)(1-\lVert y\rVert)^{-1}y=y h ( x ) = ( 1 − ∥ y ∥) ( 1 − ∥ y ∥ ) − 1 y = y . Hence h h h maps R m \mathbb{R}^{m} R m onto U U U and g g g is its inverse.
h h h is Lipschitz with constant 2 2 2 . Let x , y ∈ R m x,y\in\mathbb{R}^{m} x , y ∈ R m and write a = a x a=a_{x} a = a x , b = a y b=a_{y} b = a y . Then h ( x ) − h ( y ) = a ( x − y ) + ( a − b ) y h(x)-h(y)=a(x-y)+(a-b)y h ( x ) − h ( y ) = a ( x − y ) + ( a − b ) y , and
a − b = ( ( 1 + ∥ y ∥ ) − ( 1 + ∥ x ∥ ) ) a b = a b ( ∥ y ∥ − ∥ x ∥ ) . a-b=\bigl((1+\lVert y\rVert)-(1+\lVert x\rVert)\bigr)ab=ab\bigl(\lVert y\rVert-\lVert x\rVert\bigr). a − b = ( ( 1 + ∥ y ∥) − ( 1 + ∥ x ∥) ) ab = ab ( ∥ y ∥ − ∥ x ∥ ) .
Since a a a and b b b are positive, ∣ a − b ∣ = a b ∣ ∥ y ∥ − ∥ x ∥ ∣ |a-b|=ab\,\bigl|\lVert y\rVert-\lVert x\rVert\bigr| ∣ a − b ∣ = ab ∥ y ∥ − ∥ x ∥ . Now 0 < a ≤ 1 0<a\le 1 0 < a ≤ 1 , and b ∥ y ∥ = ∥ h ( y ) ∥ ≤ 1 b\lVert y\rVert=\lVert h(y)\rVert\le 1 b ∥ y ∥ = ∥ h ( y )∥ ≤ 1 by the first paragraph of the present claim; hence, using claim 1,
∣ a − b ∣ ∥ y ∥ = a ( b ∥ y ∥ ) ∣ ∥ y ∥ − ∥ x ∥ ∣ ≤ ∣ ∥ x ∥ − ∥ y ∥ ∣ ≤ ∥ x − y ∥ . |a-b|\,\lVert y\rVert=a\,(b\lVert y\rVert)\,\bigl|\lVert y\rVert-\lVert x\rVert\bigr|\le\bigl|\lVert x\rVert-\lVert y\rVert\bigr|\le\lVert x-y\rVert . ∣ a − b ∣ ∥ y ∥ = a ( b ∥ y ∥) ∥ y ∥ − ∥ x ∥ ≤ ∥ x ∥ − ∥ y ∥ ≤ ∥ x − y ∥ .
By the triangle inequality and homogeneity, ∥ h ( x ) − h ( y ) ∥ ≤ a ∥ x − y ∥ + ∣ a − b ∣ ∥ y ∥ ≤ 2 ∥ x − y ∥ \lVert h(x)-h(y)\rVert\le a\lVert x-y\rVert+|a-b|\,\lVert y\rVert\le 2\lVert x-y\rVert ∥ h ( x ) − h ( y )∥ ≤ a ∥ x − y ∥ + ∣ a − b ∣ ∥ y ∥ ≤ 2 ∥ x − y ∥ , that is, d E ( h ( x ) , h ( y ) ) ≤ 2 d E ( x , y ) d_{E}(h(x),h(y))\le 2\,d_{E}(x,y) d E ( h ( x ) , h ( y )) ≤ 2 d E ( x , y ) . So h h h is Lipschitz with constant 2 2 2 , hence continuous on R m \mathbb{R}^{m} R m by A Lipschitz Map is Uniformly Continuous and Continuous and A Uniformly Continuous Map Between Metric Spaces Is Continuous .
g g g is continuous on U U U . Let y 0 ∈ U y_{0}\in U y 0 ∈ U , put s 0 = ∥ y 0 ∥ < 1 s_{0}=\lVert y_{0}\rVert<1 s 0 = ∥ y 0 ∥ < 1 , c = 2 − 1 ( 1 − s 0 ) > 0 c=2^{-1}(1-s_{0})>0 c = 2 − 1 ( 1 − s 0 ) > 0 and δ 0 = c \delta_{0}=c δ 0 = c . Let y ∈ U y\in U y ∈ U satisfy d E ( y , y 0 ) < δ 0 d_{E}(y,y_{0})<\delta_{0} d E ( y , y 0 ) < δ 0 . By claim 1, ∥ y ∥ ≤ s 0 + ∥ y − y 0 ∥ < s 0 + c \lVert y\rVert\le s_{0}+\lVert y-y_{0}\rVert<s_{0}+c ∥ y ∥ ≤ s 0 + ∥ y − y 0 ∥ < s 0 + c , so 1 − ∥ y ∥ > 1 − s 0 − c = c 1-\lVert y\rVert>1-s_{0}-c=c 1 − ∥ y ∥ > 1 − s 0 − c = c and therefore β : = ( 1 − ∥ y ∥ ) − 1 < c − 1 \beta:=(1-\lVert y\rVert)^{-1}<c^{-1} β := ( 1 − ∥ y ∥ ) − 1 < c − 1 . Put β 0 = ( 1 − s 0 ) − 1 \beta_{0}=(1-s_{0})^{-1} β 0 = ( 1 − s 0 ) − 1 . As above,
β − β 0 = ( ( 1 − s 0 ) − ( 1 − ∥ y ∥ ) ) β β 0 = β β 0 ( ∥ y ∥ − s 0 ) , \beta-\beta_{0}=\bigl((1-s_{0})-(1-\lVert y\rVert)\bigr)\beta\beta_{0}=\beta\beta_{0}\bigl(\lVert y\rVert-s_{0}\bigr), β − β 0 = ( ( 1 − s 0 ) − ( 1 − ∥ y ∥) ) β β 0 = β β 0 ( ∥ y ∥ − s 0 ) ,
so ∣ β − β 0 ∣ ≤ c − 1 β 0 ∥ y − y 0 ∥ |\beta-\beta_{0}|\le c^{-1}\beta_{0}\,\lVert y-y_{0}\rVert ∣ β − β 0 ∣ ≤ c − 1 β 0 ∥ y − y 0 ∥ by claim 1. Since g ( y ) − g ( y 0 ) = β ( y − y 0 ) + ( β − β 0 ) y 0 g(y)-g(y_{0})=\beta(y-y_{0})+(\beta-\beta_{0})y_{0} g ( y ) − g ( y 0 ) = β ( y − y 0 ) + ( β − β 0 ) y 0 and ∥ y 0 ∥ ≤ 1 \lVert y_{0}\rVert\le 1 ∥ y 0 ∥ ≤ 1 , the triangle inequality and homogeneity give
d E ( g ( y ) , g ( y 0 ) ) = ∥ g ( y ) − g ( y 0 ) ∥ ≤ L ∥ y − y 0 ∥ , L = c − 1 + c − 1 β 0 > 0. d_{E}(g(y),g(y_{0}))=\lVert g(y)-g(y_{0})\rVert\le L\,\lVert y-y_{0}\rVert,\qquad L=c^{-1}+c^{-1}\beta_{0}>0 . d E ( g ( y ) , g ( y 0 )) = ∥ g ( y ) − g ( y 0 )∥ ≤ L ∥ y − y 0 ∥ , L = c − 1 + c − 1 β 0 > 0.
Given a positive real ε \varepsilon ε , choose δ \delta δ to be the least of δ 0 \delta_{0} δ 0 and ε L − 1 \varepsilon L^{-1} ε L − 1 , which is positive. Every y ∈ U y\in U y ∈ U with d U ( y , y 0 ) = d E ( y , y 0 ) < δ d_{U}(y,y_{0})=d_{E}(y,y_{0})<\delta d U ( y , y 0 ) = d E ( y , y 0 ) < δ then satisfies d E ( g ( y ) , g ( y 0 ) ) ≤ L ∥ y − y 0 ∥ < L ε L − 1 = ε d_{E}(g(y),g(y_{0}))\le L\lVert y-y_{0}\rVert<L\,\varepsilon L^{-1}=\varepsilon d E ( g ( y ) , g ( y 0 )) ≤ L ∥ y − y 0 ∥ < L ε L − 1 = ε . By Continuous Map Between Metric Spaces , g g g is continuous at y 0 y_{0} y 0 relative to U U U , and as y 0 ∈ U y_{0}\in U y 0 ∈ U was arbitrary, g g g is continuous on U U U as a map from ( U , d U ) (U,d_{U}) ( U , d U ) into ( R m , d E ) (\mathbb{R}^{m},d_{E}) ( R m , d E ) .
Claim 4. The map N N N is continuous on R m \mathbb{R}^{m} R m by claim 1, so its restriction to K K K is continuous on K K K by claim 4 of Semicontinuity and Continuity Under Composition with a Continuous Map . Since K K K is nonempty and compact in ( R m , T ) (\mathbb{R}^{m},\mathcal{T}) ( R m , T ) , Extreme Value Theorem on a Compact Subset of a Metric Space provides x 0 ∈ K x_{0}\in K x 0 ∈ K with ∥ x ∥ ≤ ∥ x 0 ∥ \lVert x\rVert\le\lVert x_{0}\rVert ∥ x ∥ ≤ ∥ x 0 ∥ for every x ∈ K x\in K x ∈ K . Put ρ = ∥ h ( x 0 ) ∥ \rho=\lVert h(x_{0})\rVert ρ = ∥ h ( x 0 )∥ , so that 0 ≤ ρ < 1 0\le\rho<1 0 ≤ ρ < 1 by claim 3.
Let x ∈ K x\in K x ∈ K and write s = ∥ x ∥ s=\lVert x\rVert s = ∥ x ∥ , t = ∥ x 0 ∥ t=\lVert x_{0}\rVert t = ∥ x 0 ∥ , so that 0 ≤ s ≤ t 0\le s\le t 0 ≤ s ≤ t . Then
( 1 + t ) − 1 t − ( 1 + s ) − 1 s = ( t ( 1 + s ) − s ( 1 + t ) ) ( 1 + s ) − 1 ( 1 + t ) − 1 = ( t − s ) ( 1 + s ) − 1 ( 1 + t ) − 1 ≥ 0 , (1+t)^{-1}t-(1+s)^{-1}s=\bigl(t(1+s)-s(1+t)\bigr)(1+s)^{-1}(1+t)^{-1}=(t-s)(1+s)^{-1}(1+t)^{-1}\ge 0 , ( 1 + t ) − 1 t − ( 1 + s ) − 1 s = ( t ( 1 + s ) − s ( 1 + t ) ) ( 1 + s ) − 1 ( 1 + t ) − 1 = ( t − s ) ( 1 + s ) − 1 ( 1 + t ) − 1 ≥ 0 ,
because t − s ≥ 0 t-s\ge 0 t − s ≥ 0 and the two factors are positive. Hence ∥ h ( x ) ∥ = ( 1 + s ) − 1 s ≤ ( 1 + t ) − 1 t = ρ \lVert h(x)\rVert=(1+s)^{-1}s\le(1+t)^{-1}t=\rho ∥ h ( x )∥ = ( 1 + s ) − 1 s ≤ ( 1 + t ) − 1 t = ρ .
Claim 5. Since 0 ≤ ρ < 1 0\le\rho<1 0 ≤ ρ < 1 , we have σ − ρ = 2 − 1 ( 1 − ρ ) > 0 \sigma-\rho=2^{-1}(1-\rho)>0 σ − ρ = 2 − 1 ( 1 − ρ ) > 0 and 1 − σ = 2 − 1 ( 1 − ρ ) > 0 1-\sigma=2^{-1}(1-\rho)>0 1 − σ = 2 − 1 ( 1 − ρ ) > 0 , so ρ < σ < 1 \rho<\sigma<1 ρ < σ < 1 . Put ψ ( y ) = ( σ − ρ ) − 1 ( σ − N ( y ) ) \psi(y)=(\sigma-\rho)^{-1}(\sigma-N(y)) ψ ( y ) = ( σ − ρ ) − 1 ( σ − N ( y )) for y ∈ R m y\in\mathbb{R}^{m} y ∈ R m . By claim 1 the map N N N is continuous on R m \mathbb{R}^{m} R m , and by claims 1, 2, 4 and 5 of Continuity of Sums and Products of Real-Valued Functions on a Metric Space the constant maps are continuous and sums and scalar multiples of continuous real-valued maps are continuous; hence ψ \psi ψ is continuous on R m \mathbb{R}^{m} R m . By claim 4 of Continuity of the Absolute Value, Maximum and Minimum of Real-Valued Functions on a Metric Space , first max { 0 , ψ } \max\{0,\psi\} max { 0 , ψ } and then χ = min { 1 , max { 0 , ψ } } \chi=\min\{1,\max\{0,\psi\}\} χ = min { 1 , max { 0 , ψ }} are continuous on R m \mathbb{R}^{m} R m , the constant maps 0 0 0 and 1 1 1 being continuous.
By Minimum of Two Elements of a Totally Ordered Set and Elementary Properties of the Minimum of Two Elements , max { 0 , ψ ( y ) } ≥ 0 \max\{0,\psi(y)\}\ge 0 max { 0 , ψ ( y )} ≥ 0 and χ ( y ) ≤ 1 \chi(y)\le 1 χ ( y ) ≤ 1 , while χ ( y ) \chi(y) χ ( y ) is one of the two numbers 1 1 1 and max { 0 , ψ ( y ) } \max\{0,\psi(y)\} max { 0 , ψ ( y )} , both nonnegative; so 0 ≤ χ ( y ) ≤ 1 0\le\chi(y)\le 1 0 ≤ χ ( y ) ≤ 1 . If ∥ y ∥ ≤ ρ \lVert y\rVert\le\rho ∥ y ∥ ≤ ρ then σ − ∥ y ∥ ≥ σ − ρ > 0 \sigma-\lVert y\rVert\ge\sigma-\rho>0 σ − ∥ y ∥ ≥ σ − ρ > 0 , so ψ ( y ) ≥ 1 \psi(y)\ge 1 ψ ( y ) ≥ 1 on multiplying by the positive number ( σ − ρ ) − 1 (\sigma-\rho)^{-1} ( σ − ρ ) − 1 , whence max { 0 , ψ ( y ) } = ψ ( y ) ≥ 1 \max\{0,\psi(y)\}=\psi(y)\ge 1 max { 0 , ψ ( y )} = ψ ( y ) ≥ 1 and χ ( y ) = 1 \chi(y)=1 χ ( y ) = 1 . If σ ≤ ∥ y ∥ \sigma\le\lVert y\rVert σ ≤ ∥ y ∥ then σ − ∥ y ∥ ≤ 0 \sigma-\lVert y\rVert\le 0 σ − ∥ y ∥ ≤ 0 , so ψ ( y ) ≤ 0 \psi(y)\le 0 ψ ( y ) ≤ 0 , whence max { 0 , ψ ( y ) } = 0 \max\{0,\psi(y)\}=0 max { 0 , ψ ( y )} = 0 and χ ( y ) = min { 1 , 0 } = 0 \chi(y)=\min\{1,0\}=0 χ ( y ) = min { 1 , 0 } = 0 .
Claim 6. Vanishing near the unit sphere. Let y ∈ B ˉ y\in\bar{B} y ∈ B ˉ satisfy σ ≤ ∥ y ∥ \sigma\le\lVert y\rVert σ ≤ ∥ y ∥ . If y ∈ U y\in U y ∈ U then F ( y ) = χ ( y ) f ( g ( y ) ) = 0 F(y)=\chi(y)f(g(y))=0 F ( y ) = χ ( y ) f ( g ( y )) = 0 because χ ( y ) = 0 \chi(y)=0 χ ( y ) = 0 by claim 5; if y ∉ U y\notin U y ∈ / U then F ( y ) = 0 F(y)=0 F ( y ) = 0 by definition.
Behaviour along h h h , and the bound. For x ∈ R m x\in\mathbb{R}^{m} x ∈ R m we have h ( x ) ∈ U h(x)\in U h ( x ) ∈ U by claim 3, so F ( h ( x ) ) = χ ( h ( x ) ) f ( g ( h ( x ) ) ) = χ ( h ( x ) ) f ( x ) F(h(x))=\chi(h(x))f(g(h(x)))=\chi(h(x))f(x) F ( h ( x )) = χ ( h ( x )) f ( g ( h ( x ))) = χ ( h ( x )) f ( x ) , again by claim 3. If ∣ f ( x ) ∣ ≤ M |f(x)|\le M ∣ f ( x ) ∣ ≤ M for every x ∈ R m x\in\mathbb{R}^{m} x ∈ R m , then M ≥ 0 M\ge 0 M ≥ 0 ; for y ∈ U y\in U y ∈ U , claim 4 of Properties of the Absolute Value in an Ordered Field gives ∣ F ( y ) ∣ = ∣ χ ( y ) ∣ ∣ f ( g ( y ) ) ∣ |F(y)|=|\chi(y)|\,|f(g(y))| ∣ F ( y ) ∣ = ∣ χ ( y ) ∣ ∣ f ( g ( y )) ∣ ; from 0 ≤ χ ( y ) ≤ 1 0\le\chi(y)\le 1 0 ≤ χ ( y ) ≤ 1 by claim 5 of the present lemma, together with − 1 ≤ 0 -1\le 0 − 1 ≤ 0 , which follows from 0 < 1 0<1 0 < 1 (claim 6 of Elementary Order Arithmetic in an Ordered Field ) by the sign reversal of claim 4 there, claim 6 of Properties of the Absolute Value in an Ordered Field gives ∣ χ ( y ) ∣ ≤ 1 |\chi(y)|\le 1 ∣ χ ( y ) ∣ ≤ 1 ; so claim 5 of Elementary Arithmetic in an Ordered Field , applied to ∣ χ ( y ) ∣ ≤ 1 |\chi(y)|\le 1 ∣ χ ( y ) ∣ ≤ 1 with the nonnegative factor ∣ f ( g ( y ) ) ∣ |f(g(y))| ∣ f ( g ( y )) ∣ , gives ∣ F ( y ) ∣ ≤ ∣ f ( g ( y ) ) ∣ ≤ M |F(y)|\le|f(g(y))|\le M ∣ F ( y ) ∣ ≤ ∣ f ( g ( y )) ∣ ≤ M . For y ∈ B ˉ ∖ U y\in\bar{B}\setminus U y ∈ B ˉ ∖ U , ∣ F ( y ) ∣ = 0 ≤ M |F(y)|=0\le M ∣ F ( y ) ∣ = 0 ≤ M .
Continuity. Let y 0 ∈ B ˉ y_{0}\in\bar{B} y 0 ∈ B ˉ and let ε \varepsilon ε be a positive real number.
Suppose first that y 0 ∈ U y_{0}\in U y 0 ∈ U , and put δ 1 = 1 − ∥ y 0 ∥ > 0 \delta_{1}=1-\lVert y_{0}\rVert>0 δ 1 = 1 − ∥ y 0 ∥ > 0 . If y ∈ B ˉ y\in\bar{B} y ∈ B ˉ satisfies d E ( y , y 0 ) < δ 1 d_{E}(y,y_{0})<\delta_{1} d E ( y , y 0 ) < δ 1 , then ∥ y ∥ ≤ ∥ y 0 ∥ + ∥ y − y 0 ∥ < 1 \lVert y\rVert\le\lVert y_{0}\rVert+\lVert y-y_{0}\rVert<1 ∥ y ∥ ≤ ∥ y 0 ∥ + ∥ y − y 0 ∥ < 1 by claim 1, so y ∈ U y\in U y ∈ U . On U U U the map y ↦ χ ( y ) f ( g ( y ) ) y\mapsto\chi(y)f(g(y)) y ↦ χ ( y ) f ( g ( y )) is continuous relative to U U U : the map g g g is continuous on U U U as a map into ( R m , d E ) (\mathbb{R}^{m},d_{E}) ( R m , d E ) by claim 3 and f f f is continuous on R m \mathbb{R}^{m} R m , so f ∘ g f\circ g f ∘ g is continuous on U U U by claim 3 of Semicontinuity and Continuity Under Composition with a Continuous Map ; the restriction of χ \chi χ to U U U is continuous on U U U by claim 4 of that lemma; and the pointwise product of two maps continuous on U U U is continuous on U U U by claim 5 of Continuity of Sums and Products of Real-Valued Functions on a Metric Space . Hence there is a positive real δ 2 \delta_{2} δ 2 such that every y ∈ U y\in U y ∈ U with d E ( y , y 0 ) < δ 2 d_{E}(y,y_{0})<\delta_{2} d E ( y , y 0 ) < δ 2 satisfies ∣ χ ( y ) f ( g ( y ) ) − χ ( y 0 ) f ( g ( y 0 ) ) ∣ < ε |\chi(y)f(g(y))-\chi(y_{0})f(g(y_{0}))|<\varepsilon ∣ χ ( y ) f ( g ( y )) − χ ( y 0 ) f ( g ( y 0 )) ∣ < ε . Let δ \delta δ be the least of δ 1 \delta_{1} δ 1 and δ 2 \delta_{2} δ 2 , which is positive. Every y ∈ B ˉ y\in\bar{B} y ∈ B ˉ with d B ˉ ( y , y 0 ) = d E ( y , y 0 ) < δ d_{\bar{B}}(y,y_{0})=d_{E}(y,y_{0})<\delta d B ˉ ( y , y 0 ) = d E ( y , y 0 ) < δ then lies in U U U , so F ( y ) = χ ( y ) f ( g ( y ) ) F(y)=\chi(y)f(g(y)) F ( y ) = χ ( y ) f ( g ( y )) and F ( y 0 ) = χ ( y 0 ) f ( g ( y 0 ) ) F(y_{0})=\chi(y_{0})f(g(y_{0})) F ( y 0 ) = χ ( y 0 ) f ( g ( y 0 )) , and ∣ F ( y ) − F ( y 0 ) ∣ < ε |F(y)-F(y_{0})|<\varepsilon ∣ F ( y ) − F ( y 0 ) ∣ < ε .
Suppose next that y 0 ∈ B ˉ ∖ U y_{0}\in\bar{B}\setminus U y 0 ∈ B ˉ ∖ U , so that ∥ y 0 ∥ = 1 \lVert y_{0}\rVert=1 ∥ y 0 ∥ = 1 , since ∥ y 0 ∥ ≤ 1 \lVert y_{0}\rVert\le 1 ∥ y 0 ∥ ≤ 1 and not ∥ y 0 ∥ < 1 \lVert y_{0}\rVert<1 ∥ y 0 ∥ < 1 . Then F ( y 0 ) = 0 F(y_{0})=0 F ( y 0 ) = 0 . Put δ = 1 − σ > 0 \delta=1-\sigma>0 δ = 1 − σ > 0 . If y ∈ B ˉ y\in\bar{B} y ∈ B ˉ satisfies d B ˉ ( y , y 0 ) = d E ( y , y 0 ) < δ d_{\bar{B}}(y,y_{0})=d_{E}(y,y_{0})<\delta d B ˉ ( y , y 0 ) = d E ( y , y 0 ) < δ , then by claim 1
∥ y ∥ ≥ ∥ y 0 ∥ − ∥ y 0 − y ∥ > 1 − ( 1 − σ ) = σ , \lVert y\rVert\ge\lVert y_{0}\rVert-\lVert y_{0}-y\rVert>1-(1-\sigma)=\sigma , ∥ y ∥ ≥ ∥ y 0 ∥ − ∥ y 0 − y ∥ > 1 − ( 1 − σ ) = σ ,
so F ( y ) = 0 F(y)=0 F ( y ) = 0 by the vanishing established above, and ∣ F ( y ) − F ( y 0 ) ∣ = 0 < ε |F(y)-F(y_{0})|=0<\varepsilon ∣ F ( y ) − F ( y 0 ) ∣ = 0 < ε .
In both cases F F F is continuous at y 0 y_{0} y 0 relative to B ˉ \bar{B} B ˉ in the sense of Continuous Map Between Metric Spaces , read for the metric space ( B ˉ , d B ˉ ) (\bar{B},d_{\bar{B}}) ( B ˉ , d B ˉ ) ; as y 0 ∈ B ˉ y_{0}\in\bar{B} y 0 ∈ B ˉ was arbitrary, F F F is continuous on B ˉ \bar{B} B ˉ .