Skip to main content
Mathematics LibreTexts

5.5: Improper Integrals

  • Page ID
    139457
  • \( \newcommand{\vecs}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } \)

    \( \newcommand{\vecd}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash {#1}}} \)

    \( \newcommand{\id}{\mathrm{id}}\) \( \newcommand{\Span}{\mathrm{span}}\)

    ( \newcommand{\kernel}{\mathrm{null}\,}\) \( \newcommand{\range}{\mathrm{range}\,}\)

    \( \newcommand{\RealPart}{\mathrm{Re}}\) \( \newcommand{\ImaginaryPart}{\mathrm{Im}}\)

    \( \newcommand{\Argument}{\mathrm{Arg}}\) \( \newcommand{\norm}[1]{\| #1 \|}\)

    \( \newcommand{\inner}[2]{\langle #1, #2 \rangle}\)

    \( \newcommand{\Span}{\mathrm{span}}\)

    \( \newcommand{\id}{\mathrm{id}}\)

    \( \newcommand{\Span}{\mathrm{span}}\)

    \( \newcommand{\kernel}{\mathrm{null}\,}\)

    \( \newcommand{\range}{\mathrm{range}\,}\)

    \( \newcommand{\RealPart}{\mathrm{Re}}\)

    \( \newcommand{\ImaginaryPart}{\mathrm{Im}}\)

    \( \newcommand{\Argument}{\mathrm{Arg}}\)

    \( \newcommand{\norm}[1]{\| #1 \|}\)

    \( \newcommand{\inner}[2]{\langle #1, #2 \rangle}\)

    \( \newcommand{\Span}{\mathrm{span}}\) \( \newcommand{\AA}{\unicode[.8,0]{x212B}}\)

    \( \newcommand{\vectorA}[1]{\vec{#1}}      % arrow\)

    \( \newcommand{\vectorAt}[1]{\vec{\text{#1}}}      % arrow\)

    \( \newcommand{\vectorB}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } \)

    \( \newcommand{\vectorC}[1]{\textbf{#1}} \)

    \( \newcommand{\vectorD}[1]{\overrightarrow{#1}} \)

    \( \newcommand{\vectorDt}[1]{\overrightarrow{\text{#1}}} \)

    \( \newcommand{\vectE}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash{\mathbf {#1}}}} \)

    \( \newcommand{\vecs}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } \)

    \( \newcommand{\vecd}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash {#1}}} \)

    Definite integrals so far have been defined only for continuous functions over finite closed intervals. There are times when you will need to perform integration despite those conditions not being met. For example, in quantum mechanics the Dirac delta function4 \(\delta\) is defined on \(\Reals\) by four properties:

    1. \(\delta(x) ~=~ 0\;\) for all \(x \ne 0\)
    2. \(\delta(0) ~=~ \infty\)
    3. \(\displaystyle\int_{-\infty}^{\infty} \delta(x)~\dx ~=~ 1\)
    4. For any continuous function \(f\) on \(\Reals\),

      \(\displaystyle\int_{-\infty}^{\infty} f(x)\;\delta(x)~\dx ~=~ f(0)\).

    Properties (3) and (4) provide examples of one type of improper integral: an integral over an infinite interval (in this case the entire real line \(\Reals = (-\infty,\infty)\)). Define this type of improper integral as follows:

    The limits in the above definitions are always taken after evaluating the integral inside the limit. Just as for “proper” definite integrals, improper integrals can be interpreted as representing the area under a curve.

    Example \(\PageIndex{1}\): improper1

    Evaluate \(~\displaystyle\int_1^{\infty}\,\dfrac{\dx}{x}~\).

    Solution

    For all real numbers \(b > 1\),

    \[\begin{aligned} \int_1^{\infty}\,\dfrac{\dx}{x} ~&=~ \lim_{b \to \infty}~\int_{1}^{b} \dfrac{\dx}{x} ~=~ \lim_{b \to \infty}~\left(\ln\,x~\Biggr|_1^{b}\right)\

    \[6pt] &=~ \lim_{b \to \infty}~ (\ln\,b ~-~ \ln\,1) ~=~ \lim_{b \to \infty}~ b ~=~ \infty\end{aligned} \nonumber \]

    and so the integral is divergent. This means that the area under the curve \(y=1/x\) over the interval \(\lival{1}{\infty}\)—as shown in the graph above—is infinite.

    Example \(\PageIndex{2}\): improper2

    Evaluate \(~\displaystyle\int_1^{\infty}\,\dfrac{\dx}{x^2}~\).

    Solution

    For all real numbers \(b > 1\),

    \[\begin{aligned} \int_1^{\infty}\,\dfrac{\dx}{x^2} ~&=~ \lim_{b \to \infty}~\int_{1}^{b} \dfrac{\dx}{x^2} ~=~ \lim_{b \to \infty}~\left(-\frac{1}{x}~\Biggr|_1^{b}\right)\

    \[6pt] &=~ \lim_{b \to \infty}~ \left(-\frac{1}{b} ~-~ \left(-\frac{1}{1}\right)\right) ~=~ \lim_{b \to \infty}~ \left(1 ~-~ \frac{1}{b}\right) ~=~ 1 ~-~ 0 ~=~ 1 ~.\end{aligned} \nonumber \]

    This means that the area under the curve \(y=1/x^2\) over the interval \(\lival{1}{\infty}\)—as shown in the graph above—equals 1. Thus, an infinite region can have a finite area. Length and area are different and not necessarily related concepts, as this example illustrates. Notice that \(y=1/x^2\) approaches the \(x\)-axis asymptote much faster than \(y=1/x\) does—fast enough to make the integral convergent.

    Example \(\PageIndex{3}\): improper3

    Evaluate \(~\displaystyle\int_{-\infty}^0\,e^x~\dx~\).

    Solution

    For all real numbers \(b < 0\),

    \[\begin{aligned} \int_{-\infty}^0\,e^x~\dx ~&=~ \lim_{b \to -\infty}~\int_{b}^0\,e^x~\dx ~=~ \lim_{b \to -\infty}~\left(e^x~\Biggr|_b^{0}\right)\

    \[6pt] &=~ \lim_{b \to -\infty}~ (1 ~-~ e^b) ~=~ 1 ~-~ 0 ~=~ 1 ~.\end{aligned} \nonumber \]

    This means that the area under the curve \(y=e^x\) over the interval \(\rival{-\infty}{0}\)—as shown in the graph above—equals 1.

    Example \(\PageIndex{4}\): improper4

    Evaluate \(~\displaystyle\int_{0}^{\infty}\,\sin\,x~\dx~\).

    Solution

    Since

    \[\begin{aligned} \int_{0}^{\infty}\,\sin\,x~\dx ~&=~ \lim_{b \to \infty}~\int_{0}^{b}\,\sin\,x~\dx\

    \[6pt] &=~ \lim_{b \to \infty}~\left(-\cos\,x~\Biggr|_0^{b}\right) ~=~ \lim_{b \to \infty}~(-\cos\,b ~+~ 1)\end{aligned} \nonumber \]

    then the integral is divergent, since \(\lim_{b \to \infty}~\cos\,b\) does not exist (\(\cos b\) oscillates between 1 and -1). This means that the net area over \(\lival{0}{\infty}\)—counted as positive above the \(x\)-axis and negative below—is indeterminate.

    Example \(\PageIndex{5}\): improper5

    Add text here.

    Solution

    Evaluate \(~\displaystyle\int_{-\infty}^{\infty}\,\frac{\dx}{1 + x^2}~\).

    Solution: Split the integral at \(x=0\):

    \[\begin{aligned} \int_{-\infty}^{\infty}\,\frac{\dx}{1 + x^2} ~&=~ \int_{-\infty}^{0}\,\frac{\dx}{1 + x^2} ~+~ \int_{0}^{\infty}\,\frac{\dx}{1 + x^2}\

    \[6pt] &=~ \left(\lim_{b \to -\infty}~\int_{b}^{0}\,\frac{\dx}{1 + x^2}\right) ~+~ \left(\lim_{b \to \infty}~\int_{0}^{b}\,\frac{\dx}{1 + x^2}\right)\

    \[6pt] &=~ \left(\lim_{b \to -\infty}~\tan^{-1} x~\Biggr|_b^{0}\right) ~+~ \left(\lim_{b \to \infty}~\tan^{-1} x~\Biggr|_0^{b}\right)\

    \[6pt] &=~ \lim_{b \to -\infty}~ (\tan^{-1} 0 ~-~ \tan^{-1} b) ~+~ \lim_{b \to \infty}~ (\tan^{-1} b ~-~ \tan^{-1} 0)\

    \[4pt] &=~ (0 - (-\pi/2)) ~+~ (\pi/2 - 0) ~=~ \pi\end{aligned} \nonumber \]

    This means that the area under the curve \(y=\frac{1}{1+x^2}\) over the entire real line \((-\infty,\infty)\)—as shown in the graph above—equals \(\pi\). Note that if the integral were split at any number \(c\) then the answer would be the same. Another way to evaluate the integral would have been to use the symmetry around the \(y\)-axis—as \(f(x)=\frac{1}{1+x^2}\) is an even function—so that

    \[\int_{-\infty}^{\infty}\,\frac{\dx}{1 + x^2} ~=~ 2\,\int_{0}^{\infty}\,\frac{\dx}{1 + x^2} ~=~ \cdots ~=~ 2 (\pi/2 - 0) ~=~ \pi ~. \nonumber \]

    Since the integrand is continuous over \(\Reals\), a common way of evaluating the integral—especially among students—is to simply use \(\pm\infty\) as actual limits of integration, thus avoiding the need to take a limit:

    \[\int_{-\infty}^{\infty}\,\frac{\dx}{1 + x^2} ~=~ \tan^{-1} x~\Biggr|_{-\infty}^{\infty} ~=~ \tan^{-1} (\infty) ~-~ \tan^{-1} (-\infty) ~=~ \frac{\pi}{2} ~-~ \frac{-\pi}{2} ~=~ \pi \nonumber \]

    This type of shortcut is fine as long as you are aware of what plugging \(x=\pm\infty\) into \(\tan^{-1} x\) actually means, and that there are no numbers for which the integrand is undefined (which would yield an improper integral of a different type, to be discussed shortly).

    The second type of improper integral is of a function not continuous or not bounded over its interval of integration. For example, the integral in property (3) of the Dirac delta function is of that type, since \(\delta\) is discontinuous at \(x=0\). Define this type of improper integral as follows:

    Example \(\PageIndex{6}\): improper6

    Evaluate \(~\displaystyle\int_{0}^{1}\,\frac{\dx}{x}~\).

    Solution

    Since \(x=0\) is a vertical asymptote for \(y = \frac{1}{x}\),

    \[\begin{aligned} \int_{0}^{1}\,\frac{\dx}{x} ~&=~ \lim_{c \to 0+}~\int_{c}^{1}\,\frac{\dx}{x} ~=~ \lim_{c \to 0+}~\left(\ln\,x~\Biggr|_c^{1}\right)\

    \[6pt] &=~ \lim_{c \to 0+}~(\ln\,1 ~-~ \ln\,c) ~=~ 0 ~-~ (-\infty) ~=~ \infty\end{aligned} \nonumber \]

    and so the integral is divergent. This means that the area under the curve \(y=1/x\) over the interval \(\rival{0}{1}\)—as shown in the graph above—is infinite. The region is infinite in the \(y\) direction.

    Example \(\PageIndex{7}\): improper7

    Evaluate \(~\displaystyle\int_{0}^{1}\,\frac{\dx}{\sqrt{x}}~\).

    Solution

    Since \(x=0\) is a vertical asymptote for \(y = \frac{1}{\sqrt{x}}\),

    \[\begin{aligned} \int_{0}^{1}\,\frac{\dx}{\sqrt{x}} ~&=~ \lim_{c \to 0+}~\int_{c}^{1}\,\frac{\dx}{\sqrt{x}} ~=~ \lim_{c \to 0+}~\left(2\,\sqrt{x}~\Biggr|_c^{1}\right)\

    \[6pt] &=~ \lim_{c \to 0+}~(2 ~-~ 2\,\sqrt{c}\,) ~=~ 2 ~-~ 0 ~=~ 2 ~.\end{aligned} \nonumber \]

    This means that the area under the curve \(y=1/\sqrt{x}\) over the interval \(\rival{0}{1}\)—as shown in the graph above—equals 2. The region is infinite in the \(y\) direction.

    Example \(\PageIndex{8}\): improper8

    Evaluate \(~\displaystyle\int_{1}^{3}\,\lfloor x \rfloor~\dx~\).

    Solution

    Recall from Example

    \[\begin{aligned} \int_{1}^{3}\,\lfloor x \rfloor~\dx ~&=~ \int_{1}^{2}\,\lfloor x \rfloor~\dx ~+~ \int_{2}^{3}\,\lfloor x \rfloor~\dx\

    \[6pt] &=~ \left(\lim_{b \to 2-}~\int_{1}^{b}\,\lfloor x \rfloor~\dx\right) ~+~ \left(\lim_{c \to 3-}~\int_{2}^{c}\,\lfloor x \rfloor~\dx\right)\

    \[6pt] &=~ \left(\lim_{b \to 2-}~\int_{1}^{b}\,1~\dx\right) ~+~ \left(\lim_{c \to 3-}~\int_{2}^{c}\,2~\dx\right)\

    \[6pt] &=~ \lim_{b \to 2-}~\left(x~\Biggr|_1^{b}\right) ~+~ \lim_{c \to 3-}~\left(2x~\Biggr|_2^{c}\right)\

    \[6pt] &=~ \lim_{b \to 2-}~(b - 1) ~+~ \lim_{c \to 3-}~(2c - 4) ~=~ (2-1) ~+~ (6-4) ~=~ 3\end{aligned} \nonumber \]

    Example \(\PageIndex{1}\): floorceil

    in Section 3.3 that the floor function \(y=\lfloor x \rfloor\) has jump discontinuities at each integer value of \(x\), as shown in the graph on the right. The integral \(\int_{1}^{3}\,\lfloor x \rfloor\;\dx\) is thus an improper integral over the interval \(\lival{1}{3}\), which needs to be split at the point of discontinuity \(x=2\) within that interval:

    Similar to some of the above examples, the following result is easy to prove (see the exercises):

    The following test for convergence or divergence is sometimes helpful:

    The idea behind part (a) is that if \(-g(x) \le f(x) \le g(x)\) over \(\lival{a}{\infty}\), then—thinking of improper integrals as areas—the integral of \(f\) is “squeezed” between the two finite integrals for \(\pm g\). There are, however, some subtle issues to prove about the limit in the integral of \(f\)—finite bounds might not necessarily mean the limit exists.5

    Example \(\PageIndex{1}\): improper9

    Add text here.

    Solution

    Show that \(~\displaystyle\int_{1}^{\infty}\,\dfrac{\sin\,x}{x^2}~\dx~\) is convergent.

    Solution: By Example

    Example \(\PageIndex{1}\): improper2

    Add text here.

    Solution

    , the integral \(\int_{1}^{\infty} \frac{1}{x^2}\,\dx\) is convergent. So since \(\abs{\sin\,x} \le 1\) for all \(x\), then

    \[\ABS{\frac{\sin\,x}{x^2}} ~\le~ \frac{1}{x^2} \nonumber \]

    for all \(x\) in \(\lival{1}{\infty}\). Thus, by the Comparison Test, \(\int_{1}^{\infty} \frac{\sin\,x}{x^2}\,\dx\) is convergent. The graph on the right shows how the curve \(y=\tfrac{\sin\,x}{x^2}\) is bounded between the curves \(y=\pm\tfrac{1}{x^2}\).

    The rules and properties from Section 5.3 concerning definite integrals still apply to improper integrals, provided the improper integrals are convergent. For example, suppose a function \(f\) has a discontinuity or vertical asymptote at \(x=c\). If both improper integrals \(\int_a^{c} f(x)\,\dx\) and \(\int_c^{b} f(x)\,\dx\) are convergent, then the improper integral \(\int_a^{b} f(x)\,\dx\) is convergent and

    \[\int_a^{b} f(x)~\dx ~=~ \int_a^{c} f(x)~\dx ~+~ \int_c^{b} f(x)~\dx ~. \nonumber \]

    Likewise, if \(\int_a^{c} f(x)\,\dx\) and \(\int_c^{\infty} f(x)\,\dx\) are convergent, then so is \(\int_a^{\infty} f(x)\,\dx\), with

    \[\int_a^{\infty} f(x)~\dx ~=~ \int_a^{c} f(x)~\dx ~+~ \int_c^{\infty} f(x)~\dx ~. \nonumber \]

    [sec5dot5]

    For Exercises 1-15, evaluate the given improper integral.

    5

    \(\displaystyle\int_{1}^{\infty} \frac{\dx}{x^3}\vphantom{\displaystyle\int_{0}^{1}\frac{\dx}{\sqrt[3]{x}}}\)

    \(\displaystyle\int_{0}^{1} \frac{\dx}{\sqrt[3]{x}}\)

    \(\displaystyle\int_{0}^{\infty} e^{-x} ~\dx\vphantom{\displaystyle\int_{0}^{1}\frac{\dx}{\sqrt[3]{x}}}\)

    \(\displaystyle\int_{0}^{\infty} e^{-2x} ~\dx\vphantom{\displaystyle\int_{0}^{1}\frac{\dx}{\sqrt[3]{x}}}\)

    \(\displaystyle\int_{-1}^{1} \frac{\dx}{x}\vphantom{\displaystyle\int_{0}^{1}\frac{\dx}{\sqrt[3]{x}}}\)

    5

    \(\displaystyle\int_{0}^{\infty} x e^{-x^2}~\dx\vphantom{\displaystyle\int_{0}^{\pi/2}}\)

    \(\displaystyle\int_{-\infty}^{0} 2^x ~\dx\vphantom{\displaystyle\int_{0}^{\pi/2}}\)

    \(\displaystyle\int_{0}^{\pi/2} \tan x ~\dx\)

    \(\displaystyle\int_{0}^{1} \frac{\ln\,x}{x}~\dx\vphantom{\displaystyle\int_{0}^{\pi/2}}\)

    \(\displaystyle\int_{-1}^{1} \frac{\dx}{\sqrt{1 - x^2}}\vphantom{\displaystyle\int_{0}^{\pi/2}}\)

    5

    \(\displaystyle\int_{0}^{3} \lceil x \rceil ~\dx\vphantom{\displaystyle\int_{0}^{1} \frac{\dx}{(x - 1)^3}}\)

    \(\displaystyle\int_{-\infty}^{\infty} \frac{\dx}{x^2 ~+~ 4}\vphantom{\displaystyle\int_{0}^{1} \frac{\dx}{(x - 1)^3}}\)

    \(\displaystyle\int_{0}^{1} \frac{\dx}{(x - 1)^3}\)

    \(\displaystyle\int_{2}^{\infty} \frac{\dx}{x\, \ln x}\vphantom{\displaystyle\int_{0}^{1} \frac{\dx}{(x - 1)^3}}\)

    \(\displaystyle\int_{1}^{\infty} \frac{\dx}{x\,\sqrt{x^2 - 1}}\vphantom{\displaystyle\int_{0}^{1} \frac{\dx}{(x - 1)^3}}\)

    In a standby system of two non-identical components, the normal operating component A has a failure rate of \(\lambda_A > 0\) failures per unit time, while the standby component B—which takes over when A fails—has a failure rate \(\lambda_B > 0\) (with \(\lambda_A \ne \lambda_B\)).

    1. Find the standby system’s reliability \(R(t)\) beyond time \(t \ge 0\), where

      \[R(t) ~=~ \int_t^{\infty} \frac{\lambda_A \lambda_B}{\lambda_A - \lambda_B} \left(e^{-\lambda_B x} ~-~ e^{-\lambda_A x}\right)~\dx ~. \nonumber \]

    2. Show that the system’s mean time to failure (MTTF) \(m\), where \(m = \int_0^{\infty} R(t)\,\dt~\), is \(m = \frac{1}{\lambda_A} + \frac{1}{\lambda_B}\).

    Show that for all \(a > 0\), \(~\displaystyle\int_{a}^{\infty}\,\frac{\dx}{x^p}~\) is convergent if \(p > 1\), and divergent if \(0 < p \le 1\).

    Show that for all \(a > 0\), \(~\displaystyle\int_{0}^{a}\,\frac{\dx}{x^p}~\) is convergent if \(0 < p < 1\), and divergent if \(p \ge 1\).

    2

    Is \(~\displaystyle\int_{1}^{\infty} \frac{\dx}{x + x^4}~\) convergent? Explain.

    Is \(~\displaystyle\int_{2}^{\infty} \frac{\dx}{x - \sqrt{x}}~\) convergent? Explain.

    [[1.]]

    Example

    Example \(\PageIndex{1}\): improper4

    Add text here.

    Solution

    showed that \(\int_0^{\infty}\sin x\,\dx\) is divergent. What is the flaw in the argument that the integral must be 0 since each “hump” of \(\sin x\) above the \(x\)-axis is canceled by one below the \(x\)-axis?

    This exercise concerns the subtraction rule \(\int_a^{\infty} (f(x) - g(x))\,\dx = \int_a^{\infty} f(x)\,\dx \;-\; \int_a^{\infty} g(x)\,\dx\).

    1. Show that \(\frac{1}{x (x+1)} = \frac{1}{x} - \frac{1}{x+1}\) for all \(x\) except 0 and -1
    2. Show that \(\int_1^{\infty} \frac{\dx}{x (x+1)}\) is convergent.
    3. Show that both \(\int_1^{\infty} \frac{\dx}{x}\) and \(\int_1^{\infty} \frac{\dx}{x+1}\) are divergent.
    4. Does part (c) contradict parts(a)-(b) and the subtraction rule? Explain.

    The improper integral \(\int_{-\infty}^{\infty} \delta(x)\,\dx = 1\) is one of the notable “improprieties” of the Dirac delta function \(\delta\). One way to think of that integral is by approximating \(\delta\) by triangular “pulse” functions \(D_n\) (for \(n \ge 1\)), as in the picture on the right.

    1. Write a formula for each \(D_n(x)\) over all of \(\Reals\).
    2. Show that \(\int_{-\infty}^{\infty} D_n(x)\,\dx = 1\,\) for all integers \(n \ge 1\).
    3. Show that \(\lim_{n \to \infty} D_n(0) = \infty = \delta(0)\).
    4. Do the \(D_n\) functions begin to resemble \(\delta\) as \(n \to \infty\)?

    This page titled 5.5: Improper Integrals is shared under a GNU General Public License 3.0 license and was authored, remixed, and/or curated by Michael Corral.