819
When you finish this section, you should be able to:
In Chapters 1 and 2, we saw the importance of the limit of a function of one variable in determining continuity and defining and finding the derivative. Limits of functions of several variables play a similar role.
We begin by reviewing the definition of a limit (Chapter 1, pp. 76 and 132).
Let \(f\,\) be a function of one variable defined everywhere in an open interval containing \(c\), except possibly at \(c\). Then the limit of \( f(x) \) as \(x\) approaches \(c\) is \(L\), written \[\bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt] { \lim\limits_{x\rightarrow c}f(x)=L }} \]
if for any number \(\varepsilon >0\), there is a number \(\delta > 0\), so that \[\bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt] { \hbox{whenever }\ 0<\left\vert x-c\right\vert <\delta \qquad \hbox{then}\qquad \vert f(x)-L\vert <\varepsilon }} \]
In this definition, the phrase “whenever \(0<\left\vert x-c\right\vert <\delta \)” means “for all \( x, x \ne c\), within a distance \(\delta \) of \(c\),” and “\(\left\vert f(x) -L\right\vert <\varepsilon \)” means “the corresponding value of \(f\) is within a distance \(\varepsilon \) of \(L.\)” See Figure 18.
The definition of a limit for functions of two or more variables is similar, but requires additional terminology.
Let \(P_{0}\) be a point in the plane or in space. A \(\delta\) -neighborhood of \(P_{0}\) consists of all the points \(P\) that lie within a distance \(\delta \) of \(P_{0}\). That is, a \(\delta \)-neighborhood consists of all points \(P\) for which \(d(P, P_{0})<\delta \), where \(d( P,P_{0}) \) is the distance from \(P\) to \(P_{0}.\)
See Figure 19.
With these definitions, we are ready to define the limit of a function of several variables.
On a real line, a \(\delta\)-neighborhood of a number \(P_{0}\) is an open interval with \(P_{0}\) as midpoint.
Let \(f\) be a function of two variables defined at each point in some disk centered at \((x_{0},\,y_{0})\), except possibly at \((x_{0},y_{0})\). Then the limit of \(f(x,y)\) as \((x,y)\) approaches \( (x_{0},y_{0})\) is \(L\) if, for any given number \(\varepsilon >0\), there is a number \(\delta >0\), so that \[\bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt] { \hbox{whenever }\ 0<\sqrt{(x-x_{0})^{2}+(y-y_{0})^{2}}<\delta \qquad \hbox{then} \qquad \vert f(x,y) -L\vert <\varepsilon }} \]
Then we say, “The limit exists”, and write \[\bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt] { \lim\limits_{(x, y)\rightarrow (x_{0}, y_{0})}f(x,y)=L }} \]
820
Figure 20 illustrates the definition.
Let \(f\) be a function of three variables defined at each point in some ball centered at \((x_{0},y_{0},z_{0})\), except possibly at \( (x_{0},y_{0},z_{0})\). Then the limit of \(f(x,y,z)\) as \( (x,y,z)\) approaches \((x_{0},y_{0},z_{0})\) is \(L\) if, for any given number \(\varepsilon >0\), there is a number \(\delta >0\), so that \[\bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt]{ \hbox{whenever }\ 0<\sqrt{(x-x_{0})^{2}+(y-y_{0})^{2}+(z-z_{0})^{2}} <\delta \quad \hbox{then}\ \vert f(x,y,z) -L\vert <\varepsilon }} \]
Then we say, “The limit exists”, and write \[\bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt]{ \lim\limits_{(x, y, z)\rightarrow (x_{0}, y_{0}, z_{0})}f(x,y,z)=L }} \]
The definition states that if you decide how close you want \(f\) to be to \(L,\) that is, if you choose \(\varepsilon ,\) you can find a number \(\delta \) so that whenever the point \((x,y,z) \) is within a distance \(\delta \) of \( ( x_{0},y_{0},z_{0}) \), then \(f\) is within a distance \( \varepsilon \) of \(L.\) See Figure 21.
Notice the similarity of the definitions for the limit of a function of one variable, two variables, and three variables. In fact, limits of functions of more than three variables are defined similarly, as follows. Suppose \( f(P)\) is a function of one or more variables defined at each point \(P\) in some neighborhood of \(P_{0}\), except possibly at \(P_{0}\). Then \[\bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt]{ \lim\limits_{P\rightarrow P_{0}}f(P)=L }} \]
if, for any given number \(\varepsilon >0\), there is a number \(\delta >0\), so that \[\bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt]{ \hbox{whenever}\ 0<d(P, P_{0})<\delta \qquad \hbox{then}\qquad \left\vert f(P)-L\right\vert <\varepsilon }} \]
821
The algebra of limits for functions of one variable extends to that for functions of several variables. We state the algebraic properties of limits of functions of two variables without proof. These properties are also true for functions of three or more variables.
Let \(f\) and \(g\) be functions of two variables for which \[ \lim_{(x, y)\rightarrow (x_{0}, y_{0})}f(x,y)=L\qquad \hbox{and}\qquad \lim_{(x, y)\rightarrow (x_{0}, y_{0})}g(x,y)=M \]
where \(L\) and \(M\) are two real numbers.
822
Find each limit:
Solution (a) \[ \begin{eqnarray*} \lim_{(x, y)\rightarrow (1, 2)}(3x^{2}+2xy+y^{2})& =&\lim_{(x, y)\rightarrow (1, 2)}( 3x^{2}) +\lim_{(x, y)\rightarrow (1, 2)}( 2xy) +\lim_{(x, y)\rightarrow (1, 2)}y^{2} \\[8pt] & =&3\lim_{x\rightarrow 1}x^{2}+\Big( \lim_{(x, y)\rightarrow (1, 2)}( 2x) \Big) \Big( \lim_{(x, y)\rightarrow (1, 2)}y\Big) +\lim_{y\rightarrow 2}y^{2} \\[8pt] &=& 3+2\Big( \lim_{x\rightarrow 1}x\Big) \Big( \lim_{y\rightarrow 2}y\Big) +4=3+2\cdot 1\cdot 2+4=11 \end{eqnarray*} \]
(b) Since \(\lim\limits_{(x, y)\rightarrow (1, 2)} (x^2+y^2) = \lim\limits_{x\rightarrow 1} x^2 + \lim\limits_{y\rightarrow 2} y^2 =5\neq 0\), we use the limit of a quotient: \[ \begin{eqnarray*} \lim_{(x, y)\rightarrow (1, 2)}\frac{xy}{x^{2}+y^{2}} &=&\frac{ \lim\limits_{(x, y)\rightarrow (1, 2)}(xy) }{\lim\limits_{(x, y) \rightarrow (1, 2)}(x^{2}+y^{2})} =\frac{\Big( \lim\limits_{x\rightarrow 1}x\Big) \Big( \lim\limits_{y\rightarrow 2}y\Big) }{\lim\limits_{x\rightarrow 1}x^{2}+\lim\limits_{y\rightarrow 2}y^{2}}=\frac{1\cdot 2}{1+4}=\frac{2}{5} \end{eqnarray*} \]
(c) Look at the denominator. Since \(\lim\limits_{(x, y) \rightarrow (3,-1)}( x^{2}+9y^{2}) =18,\) we can use the limit of a quotient property. Then \[ \begin{eqnarray*} \lim\limits_{(x, y)\rightarrow (3,-1)}\dfrac{x^{2}+2xy-3y^{2}}{x^{2}+9y^{2}} &=&\dfrac{\lim\limits_{(x, y)\rightarrow (3,-1)}(x^{2} + 2xy - 3y^{2}) }{\lim\limits_{(x, y)\rightarrow (3,-1)}(x^{2} + 9y^{2}) }\\[8pt] &=&\frac{\lim\limits_{x\rightarrow 3}x^{2} + 2\Big( \lim\limits_{x\rightarrow 3}x\Big) \Big( \lim\limits_{y\rightarrow -1}y\Big) - 3 \lim\limits_{y\rightarrow -1}y^{2}}{18}\\[8pt] &=&\dfrac{9 + 2(3)(-1) - 3(1)}{18}=0 \end{eqnarray*} \]
Problems 9 and 21.
The limit of a function \(f\) of single variable exists only if the left-hand limit equals the right-hand limit. That is, \(\lim\limits_{x\rightarrow c}f(x) \) exists if the limit is the same no matter how the number \(c\) is approached. For a single variable, there are only two ways to approach \(c\): from the left or from the right. See Figure 22(a). When \( z=f(x,y) \) is a function of two variables, however, there are infinitely many ways to approach a point \(P_{0},\) namely along any curve that passes through \(P_{0}\). See Figure 22(b). So, for \(\lim\limits_{(x,y) \rightarrow ( x_{0},y_{0}) }f(x,y) \) to exist, it is necessary that the same limit be obtained no matter what curve \( C\) containing \(P_{0}\) is used. Put another way, if two different curves that contain \((x_{0},y_{0}) \) result in different limits, then \( \lim\limits_{(x,y) \rightarrow (x_{0},y_{0}) }f(x,y) \) does not exist.
823
If \(\lim\limits_{P\rightarrow P_{0}}f(P)\) is computed along two different curves containing \(P_{0}\) and if two different answers are obtained, then \(\lim\limits_{P\rightarrow P_{0}}f(P)\) does not exist.
Show that \(\lim\limits_{(x, y)\rightarrow (0, 0)}\dfrac{xy}{ x^{2}+y^{2}}\) does not exist.
Solution Look at the denominator. Since \(\lim\limits_{(x, y) \rightarrow (0, 0)}( x^{2}+y^{2}) =0\), we cannot use the limit of a quotient to investigate the limit. We investigate the limit using two curves that contain \((0,0)\), say, \(y=2x\) and \(y=-x.\)
First we investigate the limit using the line \(y=2x\). Then \[ \begin{equation*} {\rm Using}\; y=2x: \quad \lim\limits_{(x, y)\rightarrow (0, 0)}\dfrac{x\left( 2x\right) }{x^{2}+\left( 2x\right) ^{2}}=\lim\limits_{x\rightarrow 0}\dfrac{ 2x^{2}}{x^{2}+4x^{2}}=\lim\limits_{x\rightarrow 0}\dfrac{2x^{2}}{5x^{2}}= \dfrac{2}{5} \end{equation*} \]
Now we investigate the limit using the line \(y=-x\). \[ \begin{equation*} {\rm Using}\; y=-x: \quad \lim\limits_{(x, y)\rightarrow (0, 0)}\dfrac{x\left( -x\right) }{x^{2}+\left( -x\right) ^{2}}=\lim\limits_{x\rightarrow 0}\dfrac{ -x^{2}}{x^{2}+x^{2}}=\lim\limits_{x\rightarrow 0}\dfrac{-x^{2}}{2x^{2}}=- \dfrac{1}{2} \end{equation*} \]
Since two different answers are obtained, the limit does not exist.
Figure 23 shows the graph of \(f(x,y) =\dfrac{xy}{x^{2}+y^{2}}\).
Problem 25.
Show that \(\lim\limits_{(x, y)\rightarrow (0, 0)}\dfrac{x^{2}y}{x^{4}+y^{2}}\) does not exist.
Solution We investigate the limit using curves that contain \( (0,0).\) Based on our experience with the function in Example 2, we can simultaneously test all nonvertical lines that contain \((0,0)\) by using \( y=mx \). Then \[ \begin{equation*} {\rm Using}\; y=mx: \quad \lim\limits_{(x, y)\rightarrow (0, 0)}\dfrac{x^{2}\left( mx\right) }{x^{4}+\left( mx\right) ^{2}}=\lim\limits_{x\rightarrow 0}\dfrac{ mx^{3}}{x^{4}+m^{2}x^{2}}=\lim\limits_{x\rightarrow 0}\dfrac{mx}{x^{2}+m^{2}} =0 \end{equation*} \]
Even if the same limit is obtained along two curves that contain \(P_{0}\) (or three, or even several hundred curves), we cannot conclude that the limit of the function exists. For instance, in Example 2, using the curves \(y=2x\) and \(y=\dfrac{1}{2}x\) would both result in the same value for the limit, but we have shown the limit does not exist.
Along every nonvertical line containing the point \((0,0)\), the limit is \(0\). But this does not mean that \(\lim\limits_{(x, y)\rightarrow (0, 0)}\dfrac{ x^{2}y}{x^{4}+y^{2}}\) is \(0\). Suppose we use the curve \(y=x^{2}\). Then \[ \begin{equation*} {\rm Using}\; y=x^{2}: \quad \lim\limits_{(x, y)\rightarrow (0, 0)}\dfrac{x^{4}}{ x^{4}+x^{4}}=\lim\limits_{x\rightarrow 0}\dfrac{x^{4}}{2x^{4}}=\dfrac{1 }{2} \end{equation*} \]
Since two different curves that contain \((0,0) \) result in different values for the limit, the limit does not exist.
The graph of \(f(x,y) =\dfrac{x^{2}y}{x^{4}+y^{2}}\) is shown in Figure 24.
In Example 3, the limit failed to exist even though along all nonvertical lines the limit equals \(0.\) We will see that this kind of behavior makes multivariable calculus different from single-variable calculus.
Problem 33.
How do we know a limit exists? The next example shows how to use the \( \varepsilon \)-\(\delta \) definition of a limit to prove that a limit exists.
824
Use the \(\varepsilon \)-\(\delta \) definition of a limit to show that \( \lim\limits_{(x, y)\rightarrow (0, 0)}\dfrac{2xy^{2}}{x^{2}+y^{2}}=0.\)
Solution Given a number \(\varepsilon >0\), we seek a number \( \delta >0\), so that \[ \begin{equation*} \hbox{whenever}\quad 0<\sqrt{(x-0) ^{2}+( y-0) ^{2}} <\delta \qquad \hbox{then}\quad \left\vert \dfrac{2xy^{2}}{x^{2}+y^{2}} -0\right\vert <\varepsilon \end{equation*} \]
That is, \[ \begin{equation*} \hbox{whenever }\quad 0<\sqrt{x^{2}+y^{2}}<\delta \quad \hbox{then }\quad \left\vert \dfrac{2xy^{2}}{x^{2}+y^{2}}\right\vert <\varepsilon \end{equation*} \]
We need to find a connection between \(\dfrac{2xy^{2}}{x^{2}+y^{2}}\) and \( \sqrt{x^{2}+y^{2}}.\) We begin with the observation that since \(x^{2}\geq 0,\) \[ \begin{equation*} y^{2}\leq x^{2}+y^{2}\hbox{ } \end{equation*} \]
So, for all points \((x,y)\) not equal to \((0,0),\) we have \[ \dfrac{y^{2}}{x^{2}+y^{2}}\leq 1\qquad {\color{#0066A7}{\hbox{Divide both sides by \(x^2+y^2\).}}} \]
Now \[ \begin{equation*} \left\vert \frac{2xy^{2}}{x^{2}+y^{2}}\right\vert =\frac{2\vert x\vert y^{2}}{x^{2}+y^{2}}=2\vert x\vert \cdot \dfrac{ y^{2}}{x^{2}+y^{2}}\leq 2\left\vert x\right\vert \cdot 1=2\sqrt{x^{2}}\leq 2 \sqrt{x^{2}+y^{2}} \end{equation*} \]
Given \(\varepsilon >0\), we choose \(\delta \leq \dfrac{\varepsilon }{2}\). Then whenever \(0<\sqrt{x^{2}+y^{2}}<\delta \), we have \[ \begin{equation*} \left\vert \frac{2xy^{2}}{x^{2}+y^{2}}\right\vert \leq 2\sqrt{x^{2}+y^{2}} <2\delta \leq \varepsilon \end{equation*} \]
That is, \[ \lim_{(x, y)\rightarrow (0, 0)}\left\vert \frac{2xy^{2}}{x^{2}+y^{2}} \right\vert =0 %%\tag{$\blacksquare \(} \]
Figure 25 shows the graph of \(f(x,y) =\dfrac{2xy^{2}}{x^{2}+y^{2}}.\)
Problem 63.
As Example 4 illustrates, showing a limit exists using the \(\varepsilon \)-\(\delta \) definition can be challenging. For most of our work, we will not be required to show that limits exist in this way. We will instead find limits using various properties of functions, such as continuity.
When investigating continuity for a function of one variable in Chapter 1, we began by defining what it means for a function \(f\) to be continuous at a number \(c\), where \(f\) is defined on an open interval containing \(c\). Then we extended the definition to include continuity on intervals: open, closed, or neither, and finally continuity on the domain of \(f\).
Continuity for functions of one variable is discussed in Section 1.3, pp. 93-100.
For functions of two or three variables, we begin by defining continuity at a point in the plane or in space. But first, we need some preliminary definitions.
825
Let \(S\) be a set of points in the plane or in space.
Figure 26(a) shows an interior point \(P_{0}\) in the set \(S.\) In Figure 26(b), the point \(P_{0}\) is a boundary point of \(S.\) Figure 27 shows an open set (a), a closed set (b), and a set that is neither open nor closed (c).
You may find it helpful to compare the definition given here with the one on p. 93 in Chapter 1.
Let \(f\) be a function whose domain \(D\) is some subset of points in the plane, and let \(P_{0}\) be an interior point of \(D.\) Then \(f\) is continuous at \(P_{0}\) if the following two conditions are met:
826
If either one of the two conditions is not met, then we say that \(f\) is discontinuous at the point \(P_{0}\).
If \(P_{0}\) is a boundary point of the domain \(D\) of \(f\), then \(f\) is continuous at \(P_{0}\), provided \(\lim\limits_{P\rightarrow P_{0}}f(P)=f(P_{0})\) is taken to mean that given any \(\varepsilon >0\), there is a \(\delta >0\), so that whenever \(P\) is in the domain of \(f\) and \( d(P,P_{0})<\delta \), then \(\left\vert f(P)-f(P_{0})\right\vert <\varepsilon \).
A function \(f\) is continuous on its domain if it is continuous at every point in its domain. Such functions are referred to as continuous functions.
We use limit properties and information about continuous functions of one variable to obtain information about functions of two variables. For example, two important properties of continuous functions (of one or more variables) are:
Let’s see how we use these properties. The function \(f(x,y) =ax^{n}y^{m}\), where \(n\) and \(m\) are nonnegative integers and \(a\) is a constant, is continuous at every point \((x,y) \) in the plane. This follows since \[ \begin{eqnarray*} \lim\limits_{(x,y) \rightarrow (x_{0},y_{0}) }f(x,y) &=&\lim\limits_{(x,y) \rightarrow (x_{0},y_{0}) }( ax^{n}y^{m}) =\left[ \lim\limits_{(x,y) \rightarrow (x_{0},y_{0}) } ( ax^{n}) \right]\; \left[ \lim\limits_{(x,y) \rightarrow (x_{0},y_{0}) }y^{m}\right] \\[6pt] &=&a\left[ \lim\limits_{x\rightarrow x_{0}}\,x^{n}\right]\; \left[ \lim\limits_{y\rightarrow y_{0}}y^{m}\right] =ax_{0}^{n} y_{0}^{m} \end{eqnarray*} \]
Because a polynomial in two variables is a sum of functions of the form \( ax^{n}y^{m},\) where \(n\) and \(m\) are nonnegative integers and \(a\) is a constant, it follows that polynomials in two variables are continuous at every point \((x,y) \) in the plane. Further, rational functions, quotients of polynomials, are continuous on their domains, all points in the plane except those at which the denominator is 0.
Determine where the following functions are continuous:
Solution (a) Since \(f\) is a polynomial function, it is continuous on its domain, every point \((x,y) \) in the plane, as shown in Figure 28(a).
(b) \(R\) is a rational function and is continuous on its domain, all points \((x,y) \) in the plane, except those for which \(xy=1.\) The graph of \(R\) will have no points on the cylinder \(xy=1\), as shown in Figure 28(b).
Problems 37 and 39.
Another property of a continuous function is:
827
Problem 43.
Finding limits of continuous functions can be done by substitution:
Problem 53.