FIND ME ON

GitHub

LinkedIn

Lebesgue-Stieltjes Measure

🌱

Definition
MeasureTheoryStochasticDiffs

Let F:RRF:\mathbb{R}\to \mathbb{R} be non-decreasing and right continuous. Let (aj,bj],1jn(a_{j},b_{j}],1\le j\le n be disjoint h-intervals. (At most one of the aja_{j}’s are allowed to be -\infty and at most one of the bjb_{j}’s are allowed to be ++\infty, in which case we think of the corresponding interval as (aj,+)(a_{j},+\infty)). We define λ:A[0,+]\lambda:\mathscr{A}\to[0,+\infty] (where A\mathscr{A} is the h-interval) s.t. 1. λ()=0\lambda(\emptyset)=0 2. λ(j=1n(aj,bj]):=j=1n(F(bj)F(aj))\lambda\left(\bigsqcup^{n}_{j=1}(a_{j},b_{j}]\right):=\sum_{j=1}^{n}\left(F(b_{j})-F(a_{j})\right)

Then λ\lambda is a pre-measure on A\mathscr{A}.

Let F:RRF:\mathbb{R}\to \mathbb{R} be non-decreasing and right continuous. Then there exists a unique Borel measure μF\mu_{F} such that μF((a,b])=F(b)F(a)a<b\mu_{F}((a,b])=F(b)-F(a)\quad\forall a<bConversely, if μ\mu is a Borel measure on R\mathbb{R} that is finite on all bounded Borel subsets, then defining F(x)={μ((0,x])x>00x=0μ((x,0])x<0F(x)=\begin{cases} \mu((0,x]) & x>0 \\ 0 & x=0 \\ -\mu((x,0]) & x<0 \end{cases} we have that FF is increasing and right continuous and μ=μF\mu=\mu_{F}.

For some non-decreasing and right continuous F:RRF:\mathbb{R}\to \mathbb{R}, let λ\lambda be the pre-measure defined in . Define μF(E):=inf{j=1(F(bj)F(aj)):j=1(aj,bj]E}ER\mu^{*}_{F}(E):=\inf\left\{ \sum_{j=1}^{\infty}(F(b_{j})-F(a_{j})):\bigcup_{j=1}^{\infty}(a_{j},b_{j}]\supseteq E \right\}\quad\forall E\subseteq \mathbb{R}(which is an outer measure) and let MF={AR:μF(E)=μF(EA)+μF(EAc) ER}\mathscr{M}_{F}=\{ A\subseteq \mathbb{R}:\mu^{*}_{F}(E)=\mu^{*}_{F}(E\cap A)+\mu_{F}^{*}(E\cap A^{c})\ \forall E\subseteq \mathbb{R} \}By Carathéodory Theorem, we have that MF\mathscr{M}_{F} is a σ-algebra and setting μFMF=μˉF\left. \mu_{F}^{*}\right|_{\mathscr{M}_{F}}=\bar{\mu}_{F} we have that (R,MF,μˉF)(\mathbb{R},\mathscr{M}_{F},\bar{\mu}_{F}) is a complete measure space. We call μˉF\bar{\mu}_{F} the Lebesgue-Stieltjes measure. Also, noting that BMF\mathcal{B}\subset \mathscr{M}_{F} We have that μF=μˉFB\mu_{F}=\left. \bar{\mu}_{F}\right|_{\mathcal{B}} is a Borel measure.

For M2R\mathscr{M}\subset 2^{\mathbb{R}}:EM:μF(E)=inf{j=1μ((aj,bj)):j=1(aj,bj)E}=:ν(E)\forall E\in\mathscr{M}: \mu_{F}(E)=\inf\left\{ \sum_{j=1}^{\infty}\mu((a_{j},b_{j})):\bigcup_{j=1}^{\infty}(a_{j},b_{j})\supseteq E \right\}=:\nu(E)

Let μˉF\bar{\mu}_{F} be the complete Lebesgue-Stieltjes Measure on MF\mathscr{M}_{F}. Then EM\forall E\in\mathscr{M} we have μˉF=inf{μˉF(U):U open, UE}μˉF(E)=sup{μˉF(K):K compact, KE}\begin{align*} \bar{\mu}_{F}&=\inf\{ \bar{\mu}_{F}(U):U\text{ open, }U\supseteq E \}\\ \bar{\mu}_{F}(E)&=\sup \{ \bar{\mu}_{F}(K):K\text{ compact, }K\subseteq E \} \end{align*}

Let μF:M[0,+]\overline{\mu}_{F}:\mathscr{M}\to[0,+\infty] be a complete Lebesgue-Stieltjes Measure, where the σ-algebra M=MF\mathscr{M}=\mathscr{M}_{F}. Then the following are equivalent: 1. EME\in\mathscr{M} 2. E=VN1E=V\setminus N_{1}, where VV is a Gδ set and μ(N1)=0\mu(N_{1})=0 3. E=HN2E=H\cup N_{2}, where HH is an Fσ set and μ(N2)=0\mu(N_{2})=0 # 437 [!def] 437 definition Let g:RR+g:\mathbb{R}\to \mathbb{R}^{+} be an increasing function. Define μg\mu_{g} on the set of all bounded intervals of R+\mathbb{R}^{+} by μg([a,b])=g(b+)g(a)μg(]a,b])=g(b+)g(a+)μg([a,b[)=g(b)g(a)μg(]a,b[)=g(b)g(a+)  where 0ab\begin{align*} \mu_{g}([a,b])=g(b^{+})-g(a^{-})\\ \mu_{g}(]a,b])=g(b^{+})-g(a^{+})\\ \mu_{g}([a,b[)=g(b^{-})-g(a^{-})\\ \mu_{g}(]a,b[)=g(b^{-})-g(a^{+})\\ \end{align*} \ \text{ where }0\le a\le band g(c+)=limxc x>cg(x)g(c)=limxc,x<cg(x)\begin{align*} g(c^{+})=\lim_{ x \to c \ x>c }g(x)\\ g(c^{-})=\lim_{ x \to c, x<c } g(x) \end{align*}where g increasing    g(c+)  &  g(c) exist cR+g\text{ increasing}\implies g(c^{+}) \ \ \& \ \ g(c^{-})\text{ exist }\forall c\in\mathbb{R}^{+}

The Lebesgue-Stieltjes Measure, μg\mu_{g} extends to a σ-finite measure on the Borel σ-algebra, B(R+)\mathcal{B}(\mathbb{R}^{+}).

Let g:R+Rg:\mathbb{R}^{+}\to \mathbb{R}. Then: g=g1g2 where g1,g2:R+R increasing     g is of finite variation on R+g=g_{1}-g_{2}\text{ where }g_{1},g_{2}:\mathbb{R}^{+}\to \mathbb{R}\text{ increasing }\iff g\text{ is of finite variation on }\mathbb{R}^{+}or \begin{array} -g\text{ is the difference of two increasing functions}\\ \iff \\ g\text{ is of finite variation on }\mathbb{R}^{+} \end{array}

Linked from