logo

Properties of Measurable Functions with Real Values 📂Measure Theory

Properties of Measurable Functions with Real Values

Theorem1

If two functions defined in a measurable space (X,E)(X,\mathcal{E}) are measurable functions, then the following functions are also measurable:

cf,f2,f+g,fg,f cf,\quad f^2,\quad f+g,\quad fg,\quad |f|

Proof

Measurable Function

A function f:XRf : X \to \overline{\mathbb{R}} is called a measurable function if it satisfies the following equation for all αR\alpha \in \mathbb{R}:

Sf(α):={xX  f(x)>α}E,αR S_{f}(\alpha):=\left\{ x\in X\ |\ f(x) >\alpha \right\} \in \mathcal{E},\quad \forall \alpha \in \mathbb{R}

cfcf

  • Case 1. c=0c=0

    {α>0{xX  cf(x)=0>α}=Eα0{xX  cf(x)=0>α}=XE \begin{cases} \alpha >0 & \left\{ x \in X\ |\ cf(x)=0 > \alpha \right\} =\emptyset \in \mathcal{E} \\ \alpha \le 0 & \left\{ x \in X\ |\ cf(x)=0 > \alpha \right\}=X \in \mathcal{E} \end{cases}

  • Case 2. c>0c>0

    {xX  cf(x)>α}={xX  f(x)>αc}E \left\{ x \in X\ |\ cf(x) > \alpha \right\}= \left\{ x \in X\ |\ f(x) > \frac{\alpha}{c} \right\} \in \mathcal{E}

  • Case 3. c<0c <0

    {xX  cf(x)>α}={xX  f(x)<αc}E \left\{ x \in X\ |\ cf(x) > \alpha \right\}= \left\{ x \in X\ |\ f(x) < \frac{\alpha}{c} \right\} \in \mathcal{E}

f2f^2

  • Case 1. α<0\alpha < 0

    {xX  f2(x)>α}=XE \left\{ x \in X\ |\ f^2(x) > \alpha \right\}=X \in \mathcal{E}

  • Case 2. α0\alpha \ge 0

    {xX  f2(x)>α}={xX  f(x)>α}{xX  f(x)<α}E \left\{ x \in X\ |\ f^2(x) > \alpha \right\}=\left\{ x \in X\ |\ f(x) > \sqrt{\alpha} \right\} \cup \left\{ x \in X\ |\ f(x) < -\sqrt{\alpha} \right\} \in \mathcal{E}

    Both sets on the right side are elements of E\mathcal{E}. Therefore, by the properties of the σ\sigma-algebra, the union of the two sets is also an element of the σ\sigma-algebra.

f+gf+g

Let’s say rQr\in \mathbb{Q} and define the following set.

Sr:={xX  f(x)>r}{xX  g(x)>αr} S_{r} :=\left\{ x \in X\ |\ f(x) > r \right\} \cap \left\{ x \in X\ |\ g(x) > \alpha -r \right\}

Then, since the two sets on the right are elements of E\mathcal{E}, their intersection SrS_{r} is also an element of E\mathcal{E} by the definition of the σ\sigma-algebra. Therefore, the countable union of SrS_{r} is also an element of E\mathcal{E}. That is, the proof is complete if the following equation is shown.

{xX  (f+g)(x)>α}=rQSr \left\{ x \in X\ |\ (f+g)(x) > \alpha \right\} = \bigcup_{r\in\mathbb{Q}}S_{r}

To simplify, let’s call the equation {xX  (f+g)(x)>α}={f+g>α}\left\{ x \in X\ |\ (f+g)(x) > \alpha \right\}=\left\{ f+g>\alpha \right\}.

  • Part 1. {xX  (f+g)(x)>α}rQSr\left\{ x \in X\ |\ (f+g)(x) > \alpha \right\} \supset \bigcup_{r\in\mathbb{Q}}S_{r}

    Since SrS_{r} is a subset of {f+g>α}\left\{ f+g>\alpha \right\} for all rr, it is obvious that the equation holds.

  • Part 2. {xX  (f+g)(x)>α}rQSr\left\{ x \in X\ |\ (f+g)(x) > \alpha \right\} \subset \bigcup_{r \in \mathbb{Q} } S_{r}

    Assuming x{f+g>α}x \in \left\{ f+g>\alpha \right\}, we can find a sufficiently small ϵ\epsilon that satisfies the following equation.

    f(x)+g(x)>α+ϵ \begin{equation} f(x) + g(x) > \alpha + \epsilon \end{equation}

    And let’s choose a rational number r<f(x)r<f(x) so that it becomes f(x)r<ϵ|f(x)-r| < \epsilon. Then, the following holds.

    r<f(x)<r+ϵ    f(x)ϵ<r r < f(x) < r +\epsilon \implies f(x)-\epsilon < r

    Substituting this into (1)(1) gives the following.

    g(x)>α(f(x)ϵ)>αr g(x) > \alpha -\big( f(x) -\epsilon \big) > \alpha-r

    Therefore, for such rr, since xSrx \in S_{r}, for the union of all rr, xrQSrx \in \bigcup_{r\in \mathbb{Q}}S_{r} also holds.

By Part 1 and Part 2, the following holds.

{xX  (f+g)(x)>α}=rQSr \left\{ x \in X\ |\ (f+g)(x) > \alpha \right\} = \bigcup_{r\in\mathbb{Q}}S_{r}

fgfg

Since fg=14[(f+g)2(fg)2]fg=\dfrac{1}{4} \big[ (f+g)^2 - (f-g)^2 \big], it holds by the three results above.

f|f|

  • Case 1. α0\alpha \ge 0

    Since {xX  f(x)>α}={xX  f(x)>α}{f(x)<α}\left\{ x\in X\ |\ |f(x)| > \alpha \right\}=\left\{x\in X\ |\ f(x)>\alpha \right\} \cup \left\{ f(x)<-\alpha \right\} and both sets on the right are elements of E\mathcal{E}, their intersection is also an element of E\mathcal{E}.

  • Case 2. α<0\alpha < 0

    It is obvious since {xX  f(x)>α}=XE\left\{x\in X\ |\ |f(x)| > \alpha \right\}=X \in \mathcal{E}.


  1. Robert G. Bartle, The Elements of Integration and Lebesgue Measure (1995), p9-10 ↩︎