7

We are asked to evaluate $\displaystyle \int_A x^{-1}dV(A)$, with $A=\{ (x,y):2<x+y<4,y>0,x-y>0\}$. From the solutions we know that$\displaystyle \int_A x^{-1}dV(A)=2Log(2)$. The point is to use the change of coordinates theorem to evaluate the integral.

After proving that $\mathbf{g}(s,t)=(s^2+t^2,s^2-t^2)$ is regular (it's univalent only in each quadrant), I've tried computing $\displaystyle \int_B (s^2+t^2)^{-1}\cdot |det(D\mathbf{g}(s,t))|\text{ }dV(A)$, with $B=\{ (s,t):2<s^2<4,\text{ }s^2>t^2,\text{ and } |s|>|t| \}$.

Well, when trying to compute $\displaystyle \int_B (s^2+t^2)^{-1}\cdot |det(D\mathbf{g}(s,t))|\text{ }dV(A)$,

$\displaystyle \int_B (s^2+t^2)^{-1}\cdot |det(D\mathbf{g}(s,t))|\text{ }dV(A)= \int^2_\sqrt{2}\int^s_0\frac{8st}{s^2+t^2}+\int^2_\sqrt{2}\int^0_{-s}\frac{-8st}{s^2+t^2}+\int^{-\sqrt{2}}_{-2}\int^0_s\frac{8st}{s^2+t^2}+\int^{-\sqrt{2}}_{-2}\int^{-s}_0\frac{-8st}{s^2+t^2}=4Log(16)$

Where did I go wrong? Any help will be appreciated.

2 Answers2

9

Method A: With Change of Variable as chosen by the OP

The OP changed the region A into B with the following variables: $$s^2+t^2=x\ \text{and} \ s^2-t^2=y$$ These equations define a transformation $T^{-1}$ from the $x\text{-}y$ plane to the $s\text{-}t$ plane. The Jacobian of T is:

$$\begin{vmatrix} 2s & 2t \\[0.3em] 2s & -2t \end{vmatrix}=-8st$$ To find the region A in the $s\text{-}t$ plane we look at the sides of the image in Method B.

$$y=0,\ \ x=y,\ \ x+y=4,\ \ x+y=2$$ This transforms to the image lines of B $$s^2=t^2, \ \ t=0, \ \ s^2=1, \ \ s^2=\sqrt{2} $$ Notice that this mapping is not $1\text{-}1$, hence its not possible to directly use the change of variable theorem. If we limit the region to only $s,t>0$ then we would have a one-to-one mapping. So an added constraint is $s,t>0$

The region B should be divided into symmetric parts $B1$ and $B2$ as shown in the image below. However, to insure $1\text{-}1$ mapping the relevant region is only $B1$

enter image description here

Below is the integral to find $B1$

$$\int \limits_{B1} \frac{|-8st|}{s^2+t^2} dtds=\int_{1}^{\sqrt{2}} \int_0^{s} \frac{|-8st|}{s^2+t^2} dtds=\ln(4)$$

Method B: Without Change of Variable

The first thing to do in this exercise is draw the region A. Its a very easy region to draw. I graphed it and the image is shown below, enter image description here

We are trying to find the double integral over the enclosed region. It is clear that this region needs to be divided into two parts, $A_1$ and $A_2$ as shown in the image below with the red region and green region representing $A_1$ and $A_2$ respectively. enter image description here

The two integrals are then supposed to be expressed as such:

$$\underbrace{\int_1^2\int_{2-x}^{x}x^{-1}dydx}_{A_1}+\underbrace{\int_2^4\int_{0}^{4-x} x^{-1} dydx}_{A_2}$$ $$\left[2-\ln(4) \right] +\left[\ln(16)-2\right]=\ln(4)=2\ln(2)$$

John
  • 1,149
  • @John, Thanks for your answer, but the point is to use the change of coordinates theorem to evaluate the integral. And, if not to much trouble, to understand where I got it wrong. – An old man in the sea. Jan 02 '14 at 18:31
  • @Anoldmaninthesea. If the problem can be solved in an easier method, then it's most logically solved in that method.There is no reason why the region should be parametrized with your given change of variables. – John Jan 02 '14 at 18:41
  • @John, some exercises have a pedagogical point. Your resolution is fine, from what I could understand, but it's not what I asked at the beginning. That's why I cannot accept it as an answer to my question. However, it doesn't mean that I'm not grateful to you, because I am. Thanks for your interest in this question. – An old man in the sea. Jan 02 '14 at 18:59
  • @Anoldmaninthesea. After you did your change of variables, did you draw your new region. And another thing, you jacobian determinant is $-8st$ – John Jan 02 '14 at 19:02
  • @Anoldmaninthesea. I will add a new answer that reflects the method you like – John Jan 02 '14 at 19:11
  • ,thanks. But when I integrate you area of Method A,

    www.wolframalpha.com/input/?i=Integrate[%28s^2%2Bt^2%29^-18s*t%2C{s%2C2^%281%2F2%29%2C2}%2C{t%2C0%2Cs}]%2BIntegrate[%28s^2%2Bt^2%29^-1-8s*t%2C{s%2C-2%2C-2^%281%2F2%29}%2C{t%2C0%2C-s}]

    (in the theorem we use the absolute value of Jacobian hence the minus in the second integral) we do not get the right solution. Why is that?

    – An old man in the sea. Jan 02 '14 at 20:13
  • @Anoldmaninthesea. Can you please relink me to the WA, this one is not working properly – John Jan 02 '14 at 20:18
  • I don't seem to be able to...

    Just type this in wolfram:

    Integrate[(s^2+t^2)^-18s*t,{s,2^(1/2),2},{t,0,s}]+Integrate[(s^2+t^2)^-1-8s*t,{s,-2,-2^(1/2)},{t,0,-s}]

    – An old man in the sea. Jan 02 '14 at 20:19
  • Your bounds are wrong its supposed to be from $1$ to $\sqrt{2}$ since $x+y=2s^2 \implies 1<s^2<2$. Please read above I have made changes from before. – John Jan 02 '14 at 20:22
  • @Anoldmaninthesea. Notice that $s>0$ which means that the second quadrant is ignored hence $B2$ is also ignored – John Jan 02 '14 at 20:29
  • I do not understand this «Notice that that $s>t>0$ comes from the fact that $s^2>t^2$» But isn't $s^2>t^2 \Leftrightarrow (|t|<s \vee s<-|t|) $ ? – An old man in the sea. Jan 02 '14 at 20:40
  • Yes that is true – John Jan 02 '14 at 20:46
  • Sorry, I misread your answer. Why can't $t<0$ ? – An old man in the sea. Jan 02 '14 at 20:47
  • Indeed, I am finding something weird in this question. We also have $2t^2>0$ which implies the whole set of R. – John Jan 02 '14 at 20:48
  • That's it! My whole problem lies there...

    And if you notice, g is not injective in the whole domain. Only when each quadrant is considered separately...

    – An old man in the sea. Jan 02 '14 at 20:50
  • @Anoldmaninthesea. There must be a one-to-one mapping in our transformation. I reckon that such is not the case. This needs a bit more research. – John Jan 02 '14 at 20:52
  • @Anoldmaninthesea. I think the solution would be to make the mapping one-to-one by only selecting the first quadrant. – John Jan 02 '14 at 21:04
  • You're probably right. It must be a typo in the exercise, and A should have some extra conditions like $s,t>0$ – An old man in the sea. Jan 02 '14 at 21:07
  • 1
    Jeez, this question was annoying. – John Jan 02 '14 at 21:09
  • Yes... and it is the first in a series of 8 exercises, of increasing difficulty =D – An old man in the sea. Jan 02 '14 at 21:11
  • Heheheh, good luck :\ – John Jan 02 '14 at 21:12
2

$\newcommand{\+}{^{\dagger}}% \newcommand{\angles}[1]{\left\langle #1 \right\rangle}% \newcommand{\braces}[1]{\left\lbrace #1 \right\rbrace}% \newcommand{\bracks}[1]{\left\lbrack #1 \right\rbrack}% \newcommand{\ceil}[1]{\,\left\lceil #1 \right\rceil\,}% \newcommand{\dd}{{\rm d}}% \newcommand{\ds}[1]{\displaystyle{#1}}% \newcommand{\equalby}[1]{{#1 \atop {= \atop \vphantom{\huge A}}}}% \newcommand{\expo}[1]{\,{\rm e}^{#1}\,}% \newcommand{\fermi}{\,{\rm f}}% \newcommand{\floor}[1]{\,\left\lfloor #1 \right\rfloor\,}% \newcommand{\half}{{1 \over 2}}% \newcommand{\ic}{{\rm i}}% \newcommand{\iff}{\Longleftrightarrow} \newcommand{\imp}{\Longrightarrow}% \newcommand{\isdiv}{\,\left.\right\vert\,}% \newcommand{\ket}[1]{\left\vert #1\right\rangle}% \newcommand{\ol}[1]{\overline{#1}}% \newcommand{\pars}[1]{\left( #1 \right)}% \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\pp}{{\cal P}}% \newcommand{\root}[2][]{\,\sqrt[#1]{\,#2\,}\,}% \newcommand{\sech}{\,{\rm sech}}% \newcommand{\sgn}{\,{\rm sgn}}% \newcommand{\totald}[3][]{\frac{{\rm d}^{#1} #2}{{\rm d} #3^{#1}}} \newcommand{\ul}[1]{\underline{#1}}% \newcommand{\verts}[1]{\left\vert\, #1 \,\right\vert}$ $\ds{? \equiv \int_{A}{\dd V \over x}\quad\mbox{where}\quad A \equiv \braces{\pars{x,y}\quad \ni\quad 2 < x + y < 4,\quad y>0,\quad x - y > 0}}$

The use of the $\it Heaviside\ Step$ function $\Theta:{\mathbb R}\verb*\*\braces{0} \to {\mathbb R}$: $$ \Theta\pars{x} \equiv \left\lbrace \begin{array}{lcl} 0 & \mbox{if} & x < 0 \\[1mm] 1 & \mbox{if} & x > 0 \end{array}\right. $$ avoids any pictures.

\begin{align} \color{#0000ff}{\Large ?}& = \int_{-\infty}^{\infty}\int_{-\infty}^{\infty} \Theta\pars{x + y - 2}\Theta\pars{4 - x - y}\Theta\pars{y}\Theta\pars{x - y}\, {\dd x\,\dd y \over x} \\[3mm]&= \int_{-\infty}^{\infty}\int_{-\infty}^{\infty} \Theta\pars{y - 2}\Theta\pars{4 - y}\Theta\pars{y - x}\Theta\pars{2x - y}\, {\dd x\,\dd y \over x} \\[3mm]&= \int_{-\infty}^{\infty}\dd y\,\Theta\pars{y - 2}\Theta\pars{4 - y} \int_{y/2}^{y}\Theta\pars{y}\,{\dd x \over x} = \int_{2}^{4}\dd y\,\ln\pars{y \over y/2} = \color{#0000ff}{\large 2\ln\pars{2}} \end{align}

Felix Marin
  • 89,464