§ 2.4. Contour Integrals

Referenced by:: §2.10(iii), §2.10(iv)
Permalink:: http://dlmf.nist.gov/2.4

§2.4(i)Watson's Lemma
§2.4(ii)Inverse Laplace Transforms
§2.4(iii)Laplace's Method
§2.4(iv)Saddle Points
§2.4(v)Coalescing Saddle Points: Chester, Friedman, and Ursell's Method
§2.4(vi)Other Coalescing Critical Points

§ 2.4(i). Watson's Lemma

Notes:: See Olver (1997b, pp. 112–118).
Referenced by:: §2.11(ii)
Permalink:: http://dlmf.nist.gov/2.4.SS1

The result in §2.3(ii) carries over to a complex parameter $z$ . Except that $\lambda$ is now permitted to be complex, with $\realpart{\lambda}>0$ , we assume the same conditions on $q(t)$ and also that the Laplace transform in (2.3.8) converges for all sufficiently large values of $\realpart{z}$ . Then

2.4.1 $\int _{{0}}^{{\infty}}e^{{-zt}}q(t)dt\sim\sum _{{s=0}}^{{\infty}}\Gamma\!\left(\frac{s+\lambda}{\mu}\right)\frac{a_{s}}{z^{{(s+\lambda)/\mu}}}$

Defines:: $\lambda$ : constant and $\mu$ : positive constant
Symbols:: $\Gamma\!\left(z\right)$ : Gamma function, $\sim$ : asymptotically equal, $a$ : left endpoint and $q(t)$ : analytic function
Referenced by:: §2.4(i)
Permalink:: http://dlmf.nist.gov/2.4.E1
Encodings:: TeX, pMathML, png

as $z\to\infty$ in the sector $|\mathrm{ph}z|\leq\frac{1}{2}\pi-\delta$ ( $<\frac{1}{2}\pi$ ), with $z^{{(s+\lambda)/\mu}}$ assigned its principal value.

If $q(t)$ is analytic in a sector $\alpha _{1}<\mathrm{ph}t<\alpha _{2}$ containing $\mathrm{ph}t=0$ , then the region of validity may be increased by rotation of the integration paths. We assume that in any closed sector with vertex $t=0$ and properly interior to $\alpha _{1}<\mathrm{ph}t<\alpha _{2}$ , the expansion (2.3.7) holds as $t\to 0$ , and $q(t)=O\!\left(e^{{\sigma|t|}}\right)$ as $t\to\infty$ , where $\sigma$ is a constant. Then (2.4.1) is valid in any closed sector with vertex $z=0$ and properly interior to $-\alpha _{2}-\frac{1}{2}\pi<\mathrm{ph}z<-\alpha _{1}+\frac{1}{2}\pi$ . (The branches of $t^{{(s+\lambda-\mu)/\mu}}$ and $z^{{(s+\lambda)/\mu}}$ are extended by continuity.)

For examples and extensions (including uniformity and loop integrals) see Olver (1997b, Chapter 4), Wong (1989, Chapter 1), and Temme (1985).

§ 2.4(ii). Inverse Laplace Transforms

Notes:: See Wong (1989, p. 31) and Olver (1997b, pp. 315–320).
Permalink:: http://dlmf.nist.gov/2.4.SS2

On the interval $0<t<\infty$ let $q(t)$ be differentiable and $e^{{-ct}}q(t)$ be absolutely integrable, where $c$ is a real constant. Then the Laplace transform

2.4.2 $Q(z)=\int _{{0}}^{{\infty}}e^{{-zt}}q(t)dt$

Defines:: $Q(z)$ : Laplace transform of $q(t)$
Symbols:: $q(t)$ : analytic function
Permalink:: http://dlmf.nist.gov/2.4.E2
Encodings:: TeX, pMathML, png

is continuous in $\realpart{z}\geq c$ and analytic in $\realpart{z}>c$ , and by inversion (§Ch.1)

2.4.3 $q(t)=\frac{1}{2\pi i}\lim\limits _{{\eta\to\infty}}\int _{{\sigma-i\eta}}^{{\sigma+i\eta}}e^{{tz}}Q(z)dz,$ $0<t<\infty$ ,

Defines:: $Q(z)$ : Laplace transform of $q(t)$ and $\sigma$ : constant
Symbols:: $q(t)$ : analytic function
Referenced by:: §2.4(ii)
Permalink:: http://dlmf.nist.gov/2.4.E3
Encodings:: TeX, pMathML, png

where $\sigma$ ( $\ge c$ ) is a constant.

Now assume that $c>0$ and we are given a function $Q(z)$ that is both analytic and has the expansion

2.4.4 $Q(z)\sim\sum _{{s=0}}^{{\infty}}\Gamma\!\left(\frac{s+\lambda}{\mu}\right)\frac{a_{s}}{z^{{(s+\lambda)/\mu}}},$ $z\to\infty$ ,

Symbols:: $\Gamma\!\left(z\right)$ : Gamma function, $\sim$ : asymptotically equal, $\lambda$ : constant, $a$ : left endpoint, $\mu$ : positive constant and $Q(z)$ : Laplace transform of $q(t)$
Referenced by:: §2.4(ii)
Permalink:: http://dlmf.nist.gov/2.4.E4
Encodings:: TeX, pMathML, png

in the half-plane $\realpart{z}\geq c$ . Here $\realpart{\lambda}>0$ , $\mu>0$ , and $z^{{(s+\lambda)/\mu}}$ has its principal value. Assume also (2.4.4) is differentiable. Then by integration by parts the integral

2.4.5 $q(t)=\frac{1}{2\pi i}\int _{{\sigma-i\infty}}^{{\sigma+i\infty}}e^{{tz}}Q(z)dz,$ $0<t<\infty$ ,

Symbols:: $q(t)$ : analytic function, $Q(z)$ : Laplace transform of $q(t)$ and $\sigma$ : constant
Permalink:: http://dlmf.nist.gov/2.4.E5
Encodings:: TeX, pMathML, png

is seen to converge absolutely at each limit, and be independent of $\sigma\in[c,\infty)$ . Furthermore, as $t\to 0+$ , $q(t)$ has the expansion (2.3.7).

For large $t$ , the asymptotic expansion of $q(t)$ may be obtained from (2.4.3) by Haar's method. This depends on the availability of a comparison function $F(z)$ for $Q(z)$ that has an inverse transform

2.4.6 $f(t)=\frac{1}{2\pi i}\lim\limits _{{\eta\to\infty}}\int _{{\sigma-i\eta}}^{{\sigma+i\eta}}e^{{tz}}F(z)dz$

Defines:: $F(z)$ : comparison function and $f(x)$ : inverse transform of comparison function
Symbols:: $\sigma$ : constant
Permalink:: http://dlmf.nist.gov/2.4.E6
Encodings:: TeX, pMathML, png

with known asymptotic behavior as $t\to+\infty$ . By subtraction from (2.4.3)

2.4.7 $q(t)-f(t)=\frac{e^{{\sigma t}}}{2\pi}\lim\limits _{{\eta\to\infty}}\int _{{-\eta}}^{{\eta}}e^{{it\tau}}(Q(\sigma+i\tau)-F(\sigma+i\tau))d\tau.$

Defines:: $F(z)$ : comparison function and $f(x)$ : inverse transform of comparison function
Symbols:: $q(t)$ : analytic function, $Q(z)$ : Laplace transform of $q(t)$ and $\sigma$ : constant
Permalink:: http://dlmf.nist.gov/2.4.E7
Encodings:: TeX, pMathML, png

If this integral converges uniformly at each limit for all sufficiently large $t$ , then by the Riemann-Lebesgue lemma (§Ch.1)

2.4.8 $q(t)=f(t)+o\!\left(e^{{ct}}\right),$ $t\to+\infty$ .

Defines:: $f(x)$ : inverse transform of comparison function
Symbols:: $o\!\left(x\right)$ : order symbol, $q(t)$ : analytic function and $c$ : real constant
Permalink:: http://dlmf.nist.gov/2.4.E8
Encodings:: TeX, pMathML, png

If, in addition, the corresponding integrals with $Q$ and $F$ replaced by their derivatives $Q^{{(j)}}$ and $F^{{(j)}}$ , $j=1,2,\dots,m$ , converge uniformly, then by repeated integrations by parts

2.4.9 $q(t)=f(t)+o\!\left(t^{{-m}}e^{{ct}}\right),$ $t\to+\infty$ .

Defines:: $f(x)$ : inverse transform of comparison function
Symbols:: $o\!\left(x\right)$ : order symbol, $q(t)$ : analytic function and $c$ : real constant
Permalink:: http://dlmf.nist.gov/2.4.E9
Encodings:: TeX, pMathML, png

The most successful results are obtained on moving the integration contour as far to the left as possible. For examples see Olver (1997b, pp. 315–320).

§ 2.4(iii). Laplace's Method

Notes:: See Olver (1997b, pp. 121–125).
Referenced by:: §2.11(iii), §2.4(iv)
Permalink:: http://dlmf.nist.gov/2.4.SS3

Let $\mathscr{P}$ denote the path for the contour integral

2.4.10 $I(z)=\int _{a}^{b}e^{{-zp(t)}}q(t)dt,$

Defines:: $I(z)$ : contour integral, $a$ : left endpoint, $b$ : right endpoint, $p(t)$ : analytic function and $q(t)$ : analytic function
Referenced by:: §2.4(iv)
Permalink:: http://dlmf.nist.gov/2.4.E10
Encodings:: TeX, pMathML, png

in which $a$ is finite, $b$ is finite or infinite, and $\omega$ is the angle of slope of $\mathscr{P}$ at $a$ , that is, $\lim(\mathrm{ph}\!\left(t-a\right))$ as $t\to a$ along $\mathscr{P}$ . Assume that $p(t)$ and $q(t)$ are analytic on an open domain $\mathbf{T}$ that contains $\mathscr{P}$ , with the possible exceptions of $t=a$ and $t=b$ . Other assumptions are:

(a)
In a neighborhood of $a$

2.4.11
$p(t)=p(a)+\sum _{{s=0}}^{{\infty}}p_{s}(t-a)^{{s+\mu}},$

$q(t)=\sum _{{s=0}}^{{\infty}}q_{s}(t-a)^{{s+\lambda-1}},$

Defines:

$a$ : left endpoint, $p(t)$ : analytic function, $q(t)$ : analytic function, $p_{s}$ : coefficients and $q_{s}$ : coefficients

Symbols:

$\lambda$ : constant and $\mu$ : positive constant

Referenced by:

§2.4(iv)

Permalink:

http://dlmf.nist.gov/2.4.E11

Encodings:

TeX, TeX, pMathML, pMathML, png, png

with $\realpart{\lambda}>0$ , $\mu>0$ , $p_{0}\neq 0$ , and the branches of $(t-a)^{{\lambda}}$ and $(t-a)^{{\mu}}$ continuous and constructed with $\mathrm{ph}\!\left(t-a\right)\to\omega$ as $t\to a$ along $\mathscr{P}$ .
(b)
$z$ ranges along a ray or over an annular sector $\theta _{1}\leq\theta\leq\theta _{2}$ , $|z|\geq Z$ , where $\theta=\mathrm{ph}z$ , $\theta _{2}-\theta _{1}<\pi$ , and $Z>0$ . $I(z)$ converges at $b$ absolutely and uniformly with respect to $z$ .
(c)
Excluding $t=a$ , $\realpart{\left(e^{{i\theta}}p(t)-e^{{i\theta}}p(a)\right)}$ is positive when $t\in\mathscr{P}$ , and is bounded away from zero uniformly with respect to $\theta\in[\theta _{1},\theta _{2}]$ as $t\to b$ along $\mathscr{P}$ .

Then

2.4.12 $I(z)\sim e^{{-zp(a)}}\sum _{{s=0}}^{{\infty}}\Gamma\!\left(\frac{s+\lambda}{\mu}\right)\frac{b_{s}}{z^{{(s+\lambda)/\mu}}}$

Symbols:: $\Gamma\!\left(z\right)$ : Gamma function, $\sim$ : asymptotically equal, $\lambda$ : constant, $I(z)$ : contour integral, $a$ : left endpoint, $b$ : right endpoint, $p(t)$ : analytic function and $\mu$ : positive constant
Permalink:: http://dlmf.nist.gov/2.4.E12
Encodings:: TeX, pMathML, png

as $z\to\infty$ in the sector $\theta _{1}\leq\mathrm{ph}z\leq\theta _{2}$ . The coefficients $b_{s}$ are determined as in §2.3(iii), the branch of $\mathrm{ph}p_{0}$ being chosen to satisfy

2.4.13 $|\theta+\mu\omega+\mathrm{ph}p_{0}|\leq\tfrac{1}{2}\pi.$

Symbols:: $\omega$ : angle, $p_{s}$ : coefficients, $\theta _{j}$ : angles and $\mu$ : positive constant
Permalink:: http://dlmf.nist.gov/2.4.E13
Encodings:: TeX, pMathML, png

For examples see Olver (1997b, Chapter 4). For error bounds see Boyd (1993).

§ 2.4(iv). Saddle Points

Notes:: See Olver (1997b, pp. 125–127).
Referenced by:: §2.11(iii), §9.17(iii)
Permalink:: http://dlmf.nist.gov/2.4.SS4

Now suppose that in (2.4.10) the minimum of $\realpart{(zp(t))}$ on $\mathscr{P}$ occurs at an interior point $t_{0}$ . Temporarily assume that $\theta$ $(=\mathrm{ph}z)$ is fixed, so that $t_{0}$ is independent of $z$ . We may subdivide

2.4.14 $I(z)=\int _{{t_{0}}}^{{b}}e^{{-zp(t)}}q(t)dt-\int _{{t_{0}}}^{{a}}e^{{-zp(t)}}q(t)dt,$

Symbols:: $I(z)$ : contour integral, $a$ : left endpoint, $b$ : right endpoint, $p(t)$ : analytic function and $q(t)$ : analytic function
Referenced by:: §2.4(iv)
Permalink:: http://dlmf.nist.gov/2.4.E14
Encodings:: TeX, pMathML, png

and apply the result of §2.4(iii) to each integral on the right-hand side, the role of the series (2.4.11) being played by the Taylor series of $p(t)$ and $q(t)$ at $t=t_{0}$ . If $p^{{\prime}}(t_{0})\neq 0$ , then $\mu=1$ , $\lambda$ is a positive integer, and the two resulting asymptotic expansions are identical. Thus the right-hand side of (2.4.14) reduces to the error terms. However, if $p^{{\prime}}(t_{0})=0$ , then $\mu\geq 2$ and different branches of some of the fractional powers of $p_{0}$ are used for the coefficients $b_{s}$ ; again see §2.3(iii). In consequence, the asymptotic expansion obtained from (2.4.14) is no longer null.

Zeros of $p^{{\prime}}(t)$ are called saddle points (or cols) owing to the shape of the surface $|p(t)|$ , $t\in\Complex$ , in their vicinity. Cases in which $p^{{\prime}}(t_{0})\neq 0$ are usually handled by deforming the integration path in such a way that the minimum of $\realpart{(zp(t))}$ is attained at a saddle point or at an endpoint. Additionally, it may be advantageous to arrange that $\imagpart{(zp(t))}$ is constant on the path: this will usually lead to greater regions of validity and sharper error bounds. Paths on which $\imagpart{(zp(t))}$ is constant are also the ones on which $|\exp\!\left(-zp(t)\right)|$ decreases most rapidly. For this reason the name method of steepest descents is often used. However, for the purpose of simply deriving the asymptotic expansions the use of steepest descent paths is not essential.

In the commonest case the interior minimum $t_{0}$ of $\realpart{(zp(t))}$ is a simple zero of $p^{{\prime}}(t)$ . The final expansion then has the form

2.4.15 $\int _{a}^{b}e^{{-zp(t)}}q(t)dt\sim 2e^{{-zp(t_{0})}}\sum _{{s=0}}^{{\infty}}\Gamma\!\left(s+\tfrac{1}{2}\right)\frac{b_{{2s}}}{z^{{s+(1/2)}}},$

Symbols:: $\Gamma\!\left(z\right)$ : Gamma function, $\sim$ : asymptotically equal, $a$ : left endpoint, $b$ : right endpoint, $p(t)$ : analytic function and $q(t)$ : analytic function
Referenced by:: §2.4(iv)
Permalink:: http://dlmf.nist.gov/2.4.E15
Encodings:: TeX, pMathML, png

in which

2.4.16

$b_{0}=\dfrac{q}{(2p^{{\prime\prime}})^{{1/2}}},$

$b_{2}=\left(2q^{{\prime\prime}}-\frac{2p^{{\prime\prime\prime}}q^{{\prime}}}{p^{{\prime\prime}}}+\left(\frac{5(p^{{\prime\prime\prime}})^{2}}{6(p^{{\prime\prime}})^{2}}-\frac{p^{{\mathrm{iv}}}}{2p^{{\prime\prime}}}\right)q\right)\frac{1}{(2p^{{\prime\prime}})^{{3/2}}},$

Symbols:: $b$ : right endpoint, $p(t)$ : analytic function and $q(t)$ : analytic function
Permalink:: http://dlmf.nist.gov/2.4.E16
Encodings:: TeX, TeX, pMathML, pMathML, png, png

with $p,q$ and their derivatives evaluated at $t_{0}$ . The branch of $\omega _{0}=\mathrm{ph}\!\left(p^{{\prime\prime}}(t_{0})\right)$ is the one satisfying $|\theta+2\omega+\omega _{0}|\leq\frac{1}{2}\pi$ , where $\omega$ is the limiting value of $\mathrm{ph}\!\left(t-t_{0}\right)$ as $t\to t_{0}$ from $b$ .

Higher coefficients $b_{{2s}}$ in (2.4.15) can be found from (2.3.18) with $\lambda=1$ , $\mu=2$ , and $s$ replaced by $2s$ . For integral representations of the $b_{{2s}}$ and their asymptotic behavior as $s\to\infty$ see Boyd (1995). The last reference also includes examples, as do Olver (1997b, Chapter 4), Wong (1989, Chapter 2), and Bleistein and Handelsman (1975, Chapter 7).

§ 2.4(v). Coalescing Saddle Points: Chester, Friedman, and Ursell's Method

Referenced by:: §2.4(vi), §2.8(vi), §9.15
Permalink:: http://dlmf.nist.gov/2.4.SS5

Consider the integral

2.4.17 $I(\alpha,z)=\int _{{\mathscr{P}}}e^{{-zp(\alpha,t)}}q(\alpha,t)dt$

Defines:: $I(\alpha,z)$ : integral
Symbols:: $\mathscr{P}$ : contour, $p(t)$ : analytic function and $q(t)$ : analytic function
Permalink:: http://dlmf.nist.gov/2.4.E17
Encodings:: TeX, pMathML, png

in which $z$ is a large real or complex parameter, $p(\alpha,t)$ and $q(\alpha,t)$ are analytic functions of $t$ and continuous in $t$ and a second parameter $\alpha$ . Suppose that on the integration path $\mathscr{P}$ there are two simple zeros of $\ifrac{\partial p(\alpha,t)}{\partial t}$ that coincide for a certain value $\widehat{\alpha}$ of $\alpha$ . The problem of obtaining an asymptotic approximation to $I(\alpha,z)$ that is uniform with respect to $\alpha$ in a region containing $\widehat{\alpha}$ is similar to the problem of a coalescing endpoint and saddle point outlined in §2.3(v).

The change of integration variable is given by

2.4.18 $p(\alpha,t)=\tfrac{1}{3}w^{3}+aw^{2}+bw+c,$

Defines:: $w$ : change of variable, $a$ and $b$
Symbols:: $p(t)$ : analytic function and $c$ : real constant
Permalink:: http://dlmf.nist.gov/2.4.E18
Encodings:: TeX, pMathML, png

with $a$ and $b$ chosen so that the zeros of $\ifrac{\partial p(\alpha,t)}{\partial t}$ correspond to the zeros $w_{1}(\alpha),w_{2}(\alpha)$ , say, of the quadratic $w^{2}+2aw+b$ . Then

2.4.19 $I(\alpha,z)=e^{{-cz}}\int _{{\mathscr{Q}}}\exp\!\left(-z\left(\tfrac{1}{3}w^{3}+aw^{2}+bw\right)\right)f(\alpha,w)dw,$

Defines:: $w$ : change of variable, $a$ , $b$ , $\mathscr{Q}$ : countour and $f(\alpha,w)$ : function
Symbols:: $I(\alpha,z)$ : integral and $c$ : real constant
Referenced by:: §2.4(v)
Permalink:: http://dlmf.nist.gov/2.4.E19
Encodings:: TeX, pMathML, png

where $\mathscr{Q}$ is the $w$ -map of $\mathscr{P}$ , and

2.4.20 $f(\alpha,w)=q(\alpha,t)\frac{dt}{dw}=q(\alpha,t)\frac{w^{2}+2aw+b}{\ifrac{\partial p(\alpha,t)}{\partial t}}.$

Defines:: $w$ : change of variable, $a$ , $b$ and $f(\alpha,w)$ : function
Symbols:: $p(t)$ : analytic function and $q(t)$ : analytic function
Permalink:: http://dlmf.nist.gov/2.4.E20
Encodings:: TeX, pMathML, png

The function $f(\alpha,w)$ is analytic at $w=w_{1}(\alpha)$ and $w=w_{2}(\alpha)$ when $\alpha\neq\widehat{\alpha}$ , and at the confluence of these points when $\alpha=\widehat{\alpha}$ . For large $|z|$ , $I(\alpha,z)$ is approximated uniformly by the integral that corresponds to (2.4.19) when $f(\alpha,w)$ is replaced by a constant. By making a further change of variable

2.4.21 $w=z^{{-1/3}}v-a,$

Defines:: $w$ : change of variable and $a$
Permalink:: http://dlmf.nist.gov/2.4.E21
Encodings:: TeX, pMathML, png

and assigning an appropriate value to $c$ to modify the contour, the approximating integral is reducible to an Airy function or a Scorer function (§§9.2, 9.12).

For examples, proofs, and extensions see Olver (1997b, Chapter 9), Wong (1989, Chapter 7), Olde Daalhuis and Temme (1994), Chester et al. (1957), and Bleistein and Handelsman (1975, Chapter 9).

For a symbolic method for evaluating the coefficients in the asymptotic expansions see Vidunas and Temme (2002).

§ 2.4(vi). Other Coalescing Critical Points

Referenced by:: §2.8(vi)
Permalink:: http://dlmf.nist.gov/2.4.SS6

The problems sketched in §§2.3(v) and 2.4(v) involve only two of many possibilities for the coalescence of endpoints, saddle points, and singularities in integrals associated with the special functions. For a coalescing saddle point and a pole see Wong (1989, Chapter 7) and van der Waerden (1951); in this case the uniform approximants are complementary error functions. For a coalescing saddle point and endpoint see Olver (1997b, Chapter 9) and Wong (1989, Chapter 7); if the endpoint is an algebraic singularity then the uniform approximants are parabolic cylinder functions with fixed parameter, and if the endpoint is not a singularity then the uniform approximants are complementary error functions.

For two coalescing saddle points and an endpoint see Leubner and Ritsch (1986). For two coalescing saddle points and an algebraic singularity see Temme (1986), Jin and Wong (1998). For a coalescing saddle point, a pole, and a branch point see Ciarkowski (1989). For many coalescing saddle points see §Ch.36. For double integrals with two coalescing stationary points see Qiu and Wong (2000).