Chapter 42
Recall that, under a change of variable $x = x(u)$, a definite integral transforms as \[ \int_a^bf(x)~dx = \int_c^d f\left(x(u)\right)\frac{dx}{du}~du, \] where $a = x(c)$ and $b = x(d)$.
We have seen how this generalises to iterated integrals, when changing from cartesian coordinates to polar, cylindrical or spherical coordinates.
We have seen how this generalises to iterated integrals, when changing from cartesian coordinates to polar, cylindrical or spherical coordinates.
Here, we discuss how integrals transform under more general coordinate changes.
For simplicity, emphasis will be on coordinate changes in two dimensions and on the corresponding iterated (double) integrals.
In two dimensions, a change of coordinates is conveniently described by a surjective transformation \[ T:S\to R,\qquad (u,v)\mapsto (x,y) \] where \[ x=x(u,v), \qquad y=(u,v). \]
We shall assume that $T$ and its first-order partial derivatives are continuous on $S$.
A key property of $T$ is that it maps any boundary of the region $S$ in the $u$-$v$ plane to a boundary of $R$ in the $x$-$y$ plane. Such a transformation is particularly useful if we can restrict the coordinates $u,v$ to take values on a rectangle. After possibly applying a second transformation between this rectangle and the unit square \[ \left\{ (u,v)\in \R^2~|~ 0\leq u,v\leq 1 \right\}, \] this prompts us to focus on the case where $S$ itself is the unit square.
If $T(u,v) = (x,y)$, then the point $(x,y)$ is called the image of the point $ (u,v)$. We extend this definition to regions.
|
Images of the boundary: \[ I=\left\{(u,v)~|~ 0\leq u\leq 1, v=0 \right\}. \]
\[ II=\left\{(u,v)~|~ u = 1, 0\leq v\leq 1 \right\}. \]
|
|
Images of the boundary: \[ III=\left\{(u,v)~|~ 0\leq u\leq 1, v=1 \right\}. \]
\[ IV=\left\{(u,v)~|~ u = 0, 0\leq v\leq 1 \right\}. \]
|
We can view the change of variables in two dimensions as a parameterisation \[ \r(u,v) = x(u,v)~\i+y(u,v)~\j, \qquad (u,v)\in S, \] of the region $R$ in the $x$-$y$ plane. This can then be used to analyse how the change of variables affects a double integral over $R$.
To this end, we recall that a double integral over $R$ arises as the limit of a sum over 'larger and larger' families of 'smaller and smaller' patches making up $R$: \[ \sum_{\text{patches}} f\left(x^*, y^*\right)\Delta S^* \to \iint_R f(x,y) ~dS. \] Here $\Delta S^* $ is the area of the patch containing the point $\left(x^*, y^*\right)$.
To this end, we recall that a double integral over $R$ arises as the limit of a sum over 'larger and larger' families of 'smaller and smaller' patches making up $R$: \[ \sum_{\text{patches}} f\left(x^*, y^*\right)\Delta S^* \to \iint_R f(x,y) ~dS. \] Here $\Delta S^* $ is the area of the patch containing the point $\left(x^*, y^*\right)$.
|
\(\Delta S^* \approx\) area of parallelogram defined by two vectors $\mathbf a$ and $\mathbf b$ $= \norm{\mathbf a \times \mathbf b}$.
|
|
|
By working out an approximation of the patch area $\Delta S^*,$ and expressing it in terms of the $u,$ $v$ variables, we thus arrive at the formula \[ \iint_R f(x,y)~dx~dy =\iint_S f\big(x(u,v), y(u,v)\big)\left|\frac{\partial (x,y)}{\partial (u,v)}\right|~du~dv \] where \[ \frac{\partial (x,y)}{\partial (u,v)} =\text{det} \left( \begin{array}{cc} \dfrac{\partial x}{\partial u} & \dfrac{\partial x}{\partial v}\\ \dfrac{\partial y}{\partial u} & \dfrac{\partial y}{\partial v}\\ \end{array} \right) = \dfrac{\partial x}{\partial u} \dfrac{\partial y}{\partial v} - \dfrac{\partial x}{\partial v} \dfrac{\partial y}{\partial u} . \] is called the Jacobian of the transformation $T.$
Question: Can one always find a variable transformation that maps from a rectangular or a unit square?
Consider a type I region $R=\left\{ (x,y)~|~ a\leq x\leq b , g_1(x)\leq y\leq g_2(x)\right\}$
Set
$\displaystyle x(u, v) $ | $\displaystyle = \;(1 - u)~a + ub,$ | $0\leq u\leq 1$ |
$y$ | $\displaystyle = \;(1 - v)~g_1(x) + vg_2(x),$ | $0\leq v\leq 1$ |
$\Rightarrow y(u,v)$ | $\displaystyle = \;(1 - v) ~ g_1\big((1-u)~a +ub \big) + v g_2\big((1-u)~a+ ub \big).$ |
\begin{array}{rclc} x(u, v) & = &(1 - u)~a + ub, & 0\leq u \leq 1\\ y(u,v) & = &(1 - v) ~ g_1\big((1-u)~a +ub \big) + v g_2\big((1-u)~a+ ub \big),& 0\leq v \leq 1 \end{array}
\begin{array}{rclc} x(u, v) & = &(1 - u)~a + ub, & 0\leq u \leq 1\\ y(u,v) & = &(1 - v) ~ g_1\big((1-u)~a +ub \big) + v g_2\big((1-u)~a+ ub \big),& 0\leq v \leq 1 \end{array}
What is the Jacobian then? Because $\dfrac{\partial x}{\partial v}=0$ the corresponding Jacobian is simply given by
$\displaystyle \frac{\partial (x,y)}{\partial (u,v)}$ | $\displaystyle = \left| \begin{array}{cc} \dfrac{\partial x}{\partial u} & \dfrac{\partial x}{\partial v}\\ \dfrac{\partial y}{\partial u} & \dfrac{\partial y}{\partial v}\\ \end{array} \right|$ $\displaystyle = \left| \begin{array}{cc} b-a & 0\\ \star & g_2\big((1-u)~a + ub\big) - g_1\big((1-u)~a + ub\big)\\ \end{array} \right|$ |
$\displaystyle =(b-a) \big[ g_2\big((1-u)~a + ub\big) - g_1\big((1-u)~a + ub\big) \big]$ |
Of course, a similar change of variables applies if $R$ is of type II.
Consider the unit square
$S=\left\{(u,v)~|~ 0\leq u\leq 1, 0\leq v\leq 1\right\}$.
$S=\left\{(u,v)~|~ 0\leq u\leq 1, 0\leq v\leq 1\right\}$.
|
So we map the unit square \[ S=\left\{(u,v)~|~ 0\leq u\leq 1, 0\leq v\leq 1\right\} \] to $R$ considering the following \[ \begin{array}{rclcrcl} g_1(x) & = & x^2,& & g_2(x)&=&x\\ a &= &0, & & b &=&1. \end{array} \] Set
|
|
Simplifying we get
Then $x=x(u,v)$, $y = y(u,v)$ map $S$ to $R$.
|
Therefore
$ \displaystyle \iint_R f(x,y)~dx~dy $ $= \displaystyle \int_0^1\int_0^1 f\left(u, u^2+uv-u^2v\right)$$\left(u-u^2\right)$$du~dv.$
to write $\iint_R f(x,y)~dA$ as iterated integral, where $R$ is the region in the $x$-$y$ plane bounded by $x^2 +2xy+y^2 = 2(x-y)$ and $y = 0$.
to write $\iint_R f(x,y)~dA$ as iterated integral, where $R$ is the region in the $x$-$y$ plane bounded by $x^2 +2xy+y^2 = 2(x-y)$ and $y = 0$.
Using the change of vaiable $x = u + v,$ $y = u - v$, we have \[ \iint_R f\left(x,y\right)dA=\iint_S f\left(u+v,u-v\right)\left|\frac{\partial(x,y)}{\partial(u,v)}\right|du~dv. \]
Here the Jacobian is
$\dfrac{\partial(x,y)}{\partial(u,v)}$ $ = \text{det}\left( \begin{array}{cc} \dfrac{\partial x}{\partial u} & \dfrac{\partial x}{\partial v}\\ \dfrac{\partial y}{\partial u} & \dfrac{\partial y}{\partial v} \end{array} \right)$ $ = \text{det}\left( \begin{array}{cc} 1 & 1\\ 1 & -1 \end{array} \right)$ $= -2$.
to write $\iint_R f(x,y)~dA$ as iterated integral, where $R$ is the region in the $x$-$y$ plane bounded by $x^2 +2xy+y^2 = 2(x-y)$ and $y = 0$.
Then
$ \displaystyle \iint_R f\left(x,y\right)dA=\iint_S f\left(u+v,u-v\right)\left|-2\right|du~dv. $
$ \displaystyle \iint_R f\left(x,y\right)dA=2\iint_S f\left(u+v,u-v\right)du~dv. $
Now, to identify the region $S$ we substitute $x = u + v,$ $y = u - v$ into $x^2 +2xy+y^2 = 2(x-y)\,$ and $y=0$. Then we obtain \[ v = u\quad \text{and} \quad v = u^2. \]
So the region $S$ in the $u$-$v$ plane is bounded by the parabola $v=u^2$ and a straightline $v=u$.
to write $\iint_R f(x,y)~dA$ as iterated integral, where $R$ is the region in the $x$-$y$ plane bounded by $x^2 +2xy+y^2 = 2(x-y)$ and $y = 0$.
Therefore, considering the region $S$ as type I we have that
$ \displaystyle \iint_R f\left(x,y\right)dA $ | $\displaystyle =2\iint_S f\left(u+v,u-v\right)du~dv$ |
$\displaystyle =2\int_{u=0}^{u=1}\int_{v=u^2}^{v=u} f\left(u+v,u-v\right)du~dv.$ |
to write $\iint_R f(x,y)~dA$ as iterated integral, where $R$ is the region in the $x$-$y$ plane bounded by $x^2 +2xy+y^2 = 2(x-y)$ and $y = 0$.
Therefore, considering the region $S$ as type I we have that
$ \displaystyle \iint_R f\left(x,y\right)dA $ | $\displaystyle =\;2\int_{u=0}^{u=1}\int_{v=u^2}^{v=u} f\left(u+v,u-v\right)du~dv.$ |
|
|