Computing Lesbegue Integral Using the definition of super level set

239 Views Asked by At

Use the definition of the Lebesgue integral in terms of level sets to compute: $$\int_{\mathbb{R^2}} e^{-\mid x\mid^2} dx$$

Also, I am required to verify my answer using polar coordinates.

My Approach Defining a super level set for $f(x)$ $$s_{f}(t):=\{x \in \mathbb{R^2}:f(x) \gt t\}$$ After solving I got these cases for $s_{f}(t)$ $$s_{f}(t)=\begin{cases} \mathbb{R^2}, t \leq 0\\ 0 \lt \mid x \mid \lt \sqrt{ln\mid 1/t \mid}, 0 \lt t \leq 1\\ \phi,t \gt 1 \end{cases}$$

My question: How to proceed forward and evaluate the Lebesgue Integral?

I know that the Lebesgue Integral formula that is : $$\int_{\mathbb{R^2}} f d\mu = \int_{0}^{\infty} \mu(s_f(t)) dt$$

$\mu$ is the measure.

1

There are 1 best solutions below

10
On BEST ANSWER

For $t> 0$ we have that $$ e^{-|x|^2}>t\iff -|x|^2>\ln t\iff |x|<\sqrt{-\ln t}\tag1 $$

Also we knows that (as is shown in the book)

$$ \int_{\Bbb R ^2}e^{-|x|^2}\mu_2(dx)=\int_{[0,\infty)}\mu_2(S_f(t))\; \mu_1(dt)=\int_0^\infty\mu_2(S_f(t))\; dt\tag2 $$

where $\mu_n$ is the Lebesgue measure in $\Bbb R^n$ and the RHS above is an improper integral of Riemann. By last note that

$$ \mu_2(S_f(t))=\mu_2\left(\left\{x\in \Bbb R ^2:e^{-|x|^2}>t\right\}\right) =\begin{cases} \infty ,& t=0\\ -\pi\ln t,& t\in(0,1)\\ 0, &\text{ otherwise } \end{cases}\tag3 $$ because the Lebesgue measure in $\Bbb R^2$ is the notion of area, in this case the area of circles of infinite radius (if $t=0$), finite radius (if $t\in(0,1)$), or the area of the empty set when $t\geqslant 1$, that in measure theory is set to zero. And so $$ \begin{align*} \int_{\Bbb R ^2}e^{-|x|^2}\mu_2(dx)&=\int_{(0,1)}-\pi\ln t\,\mu_1(dt)\\ &=-\pi\lim_{s\to 0^+}\int_s^1\ln t\; dt\\ &=-\pi\lim_{s\to 0^+}[t\ln t-t]^{1}_{s}\\ &=\pi \end{align*}\tag4 $$

Of course using polar coordinates we get the same answer, that is

$$ \int_{\Bbb R^2}e^{-|x|^2}\mu_2(dx)=\int_0^{2\pi}\int_0^\infty re^{-r^2}\;dr\; d\alpha=-\pi\int_0^\infty d(e^{-r^2})=\pi\tag5 $$

because $[e^{-r^2}]'=-2re^{-r^2}$. Also we can use Fubini's theorem to evaluate this integral, that is, if $x:=(x_1,x_2)$ then

$$ \begin{align*} \int_{\Bbb R^2}e^{-|x|^2}\mu_2(dx)&=\int_{\Bbb R^2}e^{-x_1^2-x_2^2}\mu_2(d(x_1,x_2))\\ &=\left(\int_{\Bbb R}e^{-x_1^2}\mu_1(dx_1)\right)\left(\int_{\Bbb R}e^{-x_2^2}\mu_1(dx_2)\right)\\ &=\left(\int_{\Bbb R}e^{-s^2}\mu_1(ds)\right)^2\\ &=(\sqrt\pi)^2\\ &=\pi \end{align*}\tag6 $$

provided that $\int_{\Bbb R}e^{-s^2}\, \mu_1(ds)=\sqrt\pi$, as can be seen here.