TheoremBase

The positive part is defined on nonnegative functions as a supremum, shown to be finite, additive and positively homogeneous there, and extended linearly through the positive and negative parts of a function; the negative part is the positive part minus the functional, and both are bounded by the dual norm of the functional.

Proof

Throughout, the functions 00, f+gf+g, min⁡(g,f)\min(g,f), f−gf-g, cfcf, f+f^{+} and f−f^{-} formed below from elements of Cc(X)C_{c}(X) lie in Cc(X)C_{c}(X) by Continuous Functions with Compact Support: the Zero Function, Closure under Operations, Vanishing off the Support, Existence of the Maximum Modulus, and Uniform Continuity (i) and Continuous Functions with Compact Support: the Zero Function, Closure under Operations, Vanishing off the Support, Existence of the Maximum Modulus, and Uniform Continuity (ii). Write M=∥Λ∥∗M=\|\Lambda\|_{*}; the dual norm is a bound for Λ\Lambda,

∣Λg∣≤M ∥g∥u(g∈Cc(X)).|\Lambda g|\le M\,\|g\|_{u}\qquad(g\in C_{c}(X)).

Let Cc+C_{c}^{+} be the set of f∈Cc(X)f\in C_{c}(X) with f≥0f\ge0.

The supremum. Let f∈Cc+f\in C_{c}^{+} and let S(f)={Λg:g∈Cc(X), 0≤g≤f}S(f)=\{\Lambda g:g\in C_{c}(X),\ 0\le g\le f\}. It contains Λ0=0\Lambda0=0. If 0≤g≤f0\le g\le f, then ∣g(x)∣≤f(x)≤∥f∥u|g(x)|\le f(x)\le\|f\|_{u} for every x∈Xx\in X, so ∥g∥u≤∥f∥u\|g\|_{u}\le\|f\|_{u}, and Λg≤M∥g∥u≤M∥f∥u\Lambda g\le M\|g\|_{u}\le M\|f\|_{u}. Thus S(f)S(f) is nonempty and bounded above, and has a supremum p(f)=sup⁡S(f)p(f)=\sup S(f). Since 0∈S(f)0\in S(f) and Λf∈S(f)\Lambda f\in S(f),

0≤p(f)≤M ∥f∥u,Λf≤p(f)(f∈Cc+).(1)0\le p(f)\le M\,\|f\|_{u},\qquad\Lambda f\le p(f)\qquad(f\in C_{c}^{+}).\tag{1}

Moreover S(0)={Λ0}S(0)=\{\Lambda0\}, so p(0)=0p(0)=0.

Additivity on nonnegative functions. Let f1,f2∈Cc+f_{1},f_{2}\in C_{c}^{+}. If 0≤g1≤f10\le g_{1}\le f_{1} and 0≤g2≤f20\le g_{2}\le f_{2}, then 0≤g1+g2≤f1+f20\le g_{1}+g_{2}\le f_{1}+f_{2}, so Λg1+Λg2=Λ(g1+g2)≤p(f1+f2)\Lambda g_{1}+\Lambda g_{2}=\Lambda(g_{1}+g_{2})\le p(f_{1}+f_{2}); taking suprema over g1g_{1} and then over g2g_{2} gives p(f1)+p(f2)≤p(f1+f2)p(f_{1})+p(f_{2})\le p(f_{1}+f_{2}). Conversely, let 0≤g≤f1+f20\le g\le f_{1}+f_{2}, and put

g1=min⁡(g,f1),g2=g−g1.g_{1}=\min(g,f_{1}),\qquad g_{2}=g-g_{1}.

Then 0≤g1≤f10\le g_{1}\le f_{1} and g2≥0g_{2}\ge0. At a point xx with g(x)≤f1(x)g(x)\le f_{1}(x) we have g2(x)=0≤f2(x)g_{2}(x)=0\le f_{2}(x), and at any other point g2(x)=g(x)−f1(x)≤f2(x)g_{2}(x)=g(x)-f_{1}(x)\le f_{2}(x); so 0≤g2≤f20\le g_{2}\le f_{2}. Hence Λg=Λg1+Λg2≤p(f1)+p(f2)\Lambda g=\Lambda g_{1}+\Lambda g_{2}\le p(f_{1})+p(f_{2}), and taking the supremum over gg,

p(f1+f2)=p(f1)+p(f2)(f1,f2∈Cc+).(2)p(f_{1}+f_{2})=p(f_{1})+p(f_{2})\qquad(f_{1},f_{2}\in C_{c}^{+}).\tag{2}

Homogeneity on nonnegative functions. Let f∈Cc+f\in C_{c}^{+} and c≥0c\ge0. For c=0c=0, p(0f)=p(0)=0p(0f)=p(0)=0. For c>0c>0, g↦cgg\mapsto cg is a bijection from {g∈Cc(X):0≤g≤f}\{g\in C_{c}(X):0\le g\le f\} onto {g∈Cc(X):0≤g≤cf}\{g\in C_{c}(X):0\le g\le cf\}, and Λ(cg)=cΛg\Lambda(cg)=c\Lambda g; hence S(cf)={cs:s∈S(f)}S(cf)=\{cs:s\in S(f)\} and

p(cf)=c p(f)(f∈Cc+, c≥0).(3)p(cf)=c\,p(f)\qquad(f\in C_{c}^{+},\ c\ge0).\tag{3}

Extension to Cc(X)C_{c}(X). Every f∈Cc(X)f\in C_{c}(X) is f=f+−f−f=f^{+}-f^{-} with f+,f−∈Cc+f^{+},f^{-}\in C_{c}^{+}. If also f=u−vf=u-v with u,v∈Cc+u,v\in C_{c}^{+}, then f++v=u+f−f^{+}+v=u+f^{-}, and (2) gives p(f+)+p(v)=p(u)+p(f−)p(f^{+})+p(v)=p(u)+p(f^{-}), that is,

p(f+)−p(f−)=p(u)−p(v).(4)p(f^{+})-p(f^{-})=p(u)-p(v).\tag{4}

Define Λ+f=p(f+)−p(f−)\Lambda^{+}f=p(f^{+})-p(f^{-}) for f∈Cc(X)f\in C_{c}(X). For f∈Cc+f\in C_{c}^{+} we have f+=ff^{+}=f and f−=0f^{-}=0, so Λ+f=p(f)\Lambda^{+}f=p(f), which is the displayed supremum formula.

Linearity and positivity of Λ+\Lambda^{+}. Let f,g∈Cc(X)f,g\in C_{c}(X). Since f+g=(f++g+)−(f−+g−)f+g=(f^{+}+g^{+})-(f^{-}+g^{-}) with both terms in Cc+C_{c}^{+}, (4) and (2) give

Λ+(f+g)=p(f++g+)−p(f−+g−)=Λ+f+Λ+g.\Lambda^{+}(f+g)=p(f^{+}+g^{+})-p(f^{-}+g^{-})=\Lambda^{+}f+\Lambda^{+}g .

Let c∈Rc\in\mathbb{R}. If c≥0c\ge0, then cf=cf+−cf−cf=cf^{+}-cf^{-}, and (4) and (3) give Λ+(cf)=c p(f+)−c p(f−)=cΛ+f\Lambda^{+}(cf)=c\,p(f^{+})-c\,p(f^{-})=c\Lambda^{+}f. If c<0c<0, then cf=(−c)f−−(−c)f+cf=(-c)f^{-}-(-c)f^{+}, and likewise Λ+(cf)=(−c)p(f−)−(−c)p(f+)=cΛ+f\Lambda^{+}(cf)=(-c)p(f^{-})-(-c)p(f^{+})=c\Lambda^{+}f. So Λ+\Lambda^{+} is a linear functional on Cc(X)C_{c}(X), and it is positive by (1).

The negative part. Define Λ−f=Λ+f−Λf\Lambda^{-}f=\Lambda^{+}f-\Lambda f for f∈Cc(X)f\in C_{c}(X). As a difference of linear functionals, Λ−\Lambda^{-} is a linear functional, and Λf=Λ+f−Λ−f\Lambda f=\Lambda^{+}f-\Lambda^{-}f for every f∈Cc(X)f\in C_{c}(X). For f∈Cc+f\in C_{c}^{+}, (1) gives Λ−f=p(f)−Λf≥0\Lambda^{-}f=p(f)-\Lambda f\ge0, so Λ−\Lambda^{-} is positive. For f∈Cc+f\in C_{c}^{+} and 0≤g≤f0\le g\le f we have 0≤f−g≤f0\le f-g\le f, hence ∥f−g∥u≤∥f∥u\|f-g\|_{u}\le\|f\|_{u} as in the first step, and

Λg−Λf=−Λ(f−g)≤M ∥f∥u;\Lambda g-\Lambda f=-\Lambda(f-g)\le M\,\|f\|_{u};

taking the supremum over gg gives Λ−f=p(f)−Λf≤M∥f∥u\Lambda^{-}f=p(f)-\Lambda f\le M\|f\|_{u}.

Bounds. Let QQ be Λ+\Lambda^{+} or Λ−\Lambda^{-}. By (1) and the previous step, 0≤Qf≤M∥f∥u0\le Qf\le M\|f\|_{u} for every f∈Cc+f\in C_{c}^{+}. Let f∈Cc(X)f\in C_{c}(X). For every x∈Xx\in X, 0≤f±(x)≤∣f(x)∣≤∥f∥u0\le f^{\pm}(x)\le|f(x)|\le\|f\|_{u}, so ∥f±∥u≤∥f∥u\|f^{\pm}\|_{u}\le\|f\|_{u}. By linearity Qf=Qf+−Qf−Qf=Qf^{+}-Qf^{-}, a difference of two numbers in [0,M∥f∥u][0,M\|f\|_{u}], so

∣Qf∣≤M ∥f∥u(f∈Cc(X)).|Qf|\le M\,\|f\|_{u}\qquad(f\in C_{c}(X)).

Thus M≥0M\ge0 is a bound for QQ: QQ is bounded, so Q∈Cc(X)∗Q\in C_{c}(X)^{*}, and MM lies in the set of nonnegative bounds of QQ, whose infimum is ∥Q∥∗\|Q\|_{*} by The Dual Norm (i). Hence ∥Λ±∥∗≤M=∥Λ∥∗\|\Lambda^{\pm}\|_{*}\le M=\|\Lambda\|_{*}, which completes the proof of (i).

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…