Additional Methods for Calculating Integrals

Additional Methods of Integration

Motivation: Beyond Simple Fractions

Not all integrals are reducible to trivial fractions. Integrals of the form $\int_0^\infty \ln(x)\cdot f(x) dx$, $\int_0^\infty x^a\cdot f(x) dx$, or sums $\sum f(n)$ require special contours and tricks. The two main methods are: the “keyhole" contour for power integrals and summation via residues of the cotangent.

Logarithmic Integrals: The “Keyhole" Contour

For $\int_0^\infty x^a f(x) dx$ ($0 < a < 1$), we make a cut along $[0,+\infty)$ and integrate $f(z)\cdot z^a$ over the "keyhole" contour: a large circle $|z|=R \to$ disappears, a small $|z|=\varepsilon \to$ disappears, only the banks of the cut remain. On the upper bank $z = x$ $(x > 0)$, on the lower bank $z = x e^{2\pi i} \to z^a = x^a \cdot e^{2\pi i a}$.

$\oint = \int_0^\infty x^a f(x) dx - e^{2\pi i a}\int_0^\infty x^a f(x) dx = (1 - e^{2\pi i a}) \int_0^\infty x^a f(x) dx = 2\pi i \sum \text{Res}$.

Summation of Series via Residues

Key technique: $\pi\cot(\pi z)$ has simple poles at $z = n \in \mathbb{Z}$ with residue $1$. Then:

$ \sum_{n=-\infty}^\infty f(n) = -\sum_{\text{special points of }f} \text{Res}[f(z)\cdot \pi\cot(\pi z)]. $

Example — Euler’s Identity: $\sum_{n=1}^\infty \frac{1}{n^2} = \frac{\pi^2}{6}.$

Via the sine series: $\frac{\sin(\pi z)}{\pi z} = \prod_{n=1}^\infty \left(1 - \frac{z^2}{n^2}\right) \to \ln \sin(\pi z) - \ln(\pi z) = \sum_{n=1}^\infty \ln\left(1 - \frac{z^2}{n^2}\right)$. Differentiate twice and take $z \to 0$.

Integral Transforms — The Mellin Transform

$\mathcal{M}f = \int_0^\infty x^{s-1} f(x) dx$ — Mellin transform.

$\mathcal{M}e^{-x} = \Gamma(s)$ — the gamma function.

Inversion formula: $f(x) = \frac{1}{2\pi i} \int_{c-i\infty}^{c+i\infty} x^{-s} \mathcal{M}f ds$.

Connection to number theory: $\mathcal{M}\psi(x) = -\frac{\zeta'(s)}{s\zeta(s)}$, where $\psi$ is the Chebyshev function $\to$ distribution of prime numbers via zeros of $\zeta(s)$.

Special Points on the Contour

If a pole lies on the real axis, we bypass with a small semicircle. Contribution = $\pi i \cdot$ Res (half-residue):

$\int_{-\infty}^\infty \frac{\sin(x)}{x} dx = \operatorname{Im}\int \frac{e^{ix}}{x} dx = \operatorname{Im}(2\pi i \cdot 0 + \pi i \cdot \text{Res}_{z=0}) = \operatorname{Im}(\pi i \cdot 1) = \mathbf{\pi}$.

Numerical Example

Problem: Calculate $\int_0^\infty \frac{x^{-1/2}}{1+x} dx$ using the "keyhole" contour.

Step 1. The function $f(z) = \frac{z^{-1/2}}{1+z}$ with $z^{-1/2} = e^{(-1/2)\ln z}$, $\ln z$ has a cut along $[0,+\infty)$.

Step 2. Contour: the upper bank $[\varepsilon,R]$ $(\operatorname{arg} z = 0)$, the large arc $|z|=R$ $(\to 0$ as $R \to \infty)$, the lower bank $[R,\varepsilon]$ $(\operatorname{arg} z = 2\pi, z^{-1/2} = e^{-\pi i} x^{-1/2} = -x^{-1/2})$, the small arc $|z|=\varepsilon$ $(\to 0$ as $\varepsilon \to 0)$.

Step 3. Sum around the contour: $ \int_0^\infty \frac{x^{-1/2}}{1+x} dx + \int_0^\infty \frac{-x^{-1/2}}{1+x} dx = (1 - (-1)) \int_0^\infty \frac{x^{-1/2}}{1+x} dx = 2I. $

Step 4. Residue at $z = -1 = e^{i\pi}$: $z^{-1/2} = e^{-i\pi/2} = -i$. $\operatorname{Res}_{z=-1} f = \frac{-i}{1} = -i$.

Step 5. $2I = 2\pi i\cdot(-i) = 2\pi \to \mathbf{I = \pi}$.

Check: Substitute $x = t^2 \to 2\int_0^\infty \frac{dt}{1 + t^2} = 2\cdot(\pi/2) = \pi$ ✓.

Real Application

Probability theory: the Mellin transform is connected with the distribution of products of random variables. If $X, Y$ are independent, then the distribution of $X \cdot Y$ is calculated through the Mellin convolution — an analogue of Fourier for multiplicative structure.

Additional Aspects

The special methods section includes asymptotic expansions (Laplace method and stationary phase method), estimates for the growth of entire functions (order and type by Hadamard), factorization representations via infinite products. These tools are especially important in combinatorial asymptotics (the Stirling formula $n! \approx \sqrt{2\pi n}\cdot (n/e)^n$ is obtained by the Laplace method) and in number theory (precise estimations of the number of primes via products over zeros of the $\zeta$-function). In modern numerical mathematics, the stationary phase method underlies fast integral algorithms for wave problems, where usual quadrature diverges due to rapid oscillations; identifying stationary points reduces computational cost from $O(N)$ to $O(\log N)$ operations.

Connection with Other Branches of Mathematics

Many additional integration techniques naturally appear when solving differential equations. Contour methods are applied, for instance, in the theory of linear operators for determination of the resolvent and functional calculus: Duamel’s formula and spectral decomposition via complex integral along a contour that encloses the spectrum of the operator. The Cauchy–Gilbert approach to boundary value problems for the Laplace equation in the plane is also based on contour integrals and residues.

The Mellin transform is closely connected to Fourier theory: substitution $t = \ln x$ turns a multiplicative convolution into an additive one, and the Mellin integral — into the Fourier transform over the variable $\ln x$. In number theory, this provides analytical apparatus for $\zeta$-functions and $L$-series. Results such as Perron’s formula for the sums of arithmetic functions are derived via integrals along vertical lines and contour shifting taking into account zeros and poles.

In probability problems, integral methods manifest when studying stable distributions and large deviations. Classic Chernov–Cramer bounds are derived through analysis of the complex logarithmic moment. The stationary phase and steepest descent methods are used in statistical mechanics: asymptotics of the canonical distribution, analysis of action functionals in quantum field theory.

Numerical integration methods also use the analytic structure of the integrand. Gauss and Jacobi–Poncelet’s ideas on selecting nodes as roots of orthogonal polynomials are linked to the analysis of measures and their moments, formulated via Mellin and Stieltjes integrals. Modern algorithms such as Talbot’s numerical method for the inverse Laplace transform are based on contour deformation and control of residues, directly continuing the line of complex analysis.

Historical Note and Development of the Idea

Contour integration methods took shape in the nineteenth century in the works of Cauchy (Cours d’Analyse, 1821) and Riemann. Cauchy formulated the integral theorem and formula, from which the concept of residue naturally emerged. Later, Jordan and Poinso developed the technique of bypassing poles and cuts, including small arcs and sectors. The Mellin transform was systematically studied by Hermann Mellin at the end of the nineteenth century; his 1895 monograph links the Mellin integral with Euler’s gamma function and the Riemann $\zeta$-function. Already in the works of Hadamard and de la Vallée-Poussin, integral representations of the $\zeta$-function were used for rigorous proofs of the asymptotic formula for the distribution of primes. The method of summing series via the cotangent function appeared with Euler, but rigorous justification via complex analysis is tied to Weierstrass and his theorem on the factorization of entire functions (1876). The representation of $\sin(\pi z)$ as an infinite product became the prototype for many subsequent infinite products and Bloch–Wigner products.