Chapter X · The general theory of the logarithmic, exponential and circular functions

Integrals along a curve

§229 30 min

How to integrate a polynomial or rational function of z along a path in the Argand diagram, and when the answer depends on the path.

Why this matters

In Chapter IX the logarithm was defined as an integral, log⁡x=∫1xdt/t\log x = \int_1^x dt/t. To define the logarithm of a complex number ζ\zeta in the same way, we have to integrate 1/z1/z from 11 to ζ\zeta. But between two points of the plane there is no single interval to integrate over: there are infinitely many curves. §229 says what an integral along a curve means. Whether its value depends on the curve is the question the rest of the chapter turns on.

The idea

An ordinary integral ∫abf(x) dx\int_a^b f(x)\,dx is the limit of sums ∑f(xj)(xj+1−xj)\sum f(x_j)(x_{j+1} - x_j): cut [a,b][a, b] into small pieces, multiply each length by a value of ff, and add. Along a curve CC from AA to BB in the Argand diagram we do the same thing with complex numbers. Mark points z0=A,z1,…,zn=Bz_0 = A, z_1, \dots, z_n = B along CC; each step zj+1−zjz_{j+1} - z_j is now a small complex number (a displacement with a direction), and we form

∑j=0n−1f(zj) (zj+1−zj).\sum_{j=0}^{n-1} f(z_j)\,(z_{j+1} - z_j).

Every product f(zj)(zj+1−zj)f(z_j)(z_{j+1} - z_j) is a complex number, so the sum can be drawn as a chain of arrows laid tip to tail. As the steps get shorter, the end of the chain settles down: that end point is the integral of ff along CC.

Explore Integrate along a path you can bend

Try this. With f(z)=z2f(z) = z^2, increase the number of chords nn and watch the end of the chain approach the hollow dot. Then drag the purple control point PP to bend the path, keeping AA and BB fixed. Repeat with f(z)=x−iyf(z) = x - iy, and with f(z)=1/zf(z) = 1/z, bending the path so that it passes on the other side of the origin.

∑f(zj) Δzj→∫Cf(z) dz\textstyle\sum f(z_j)\,\Delta z_j \to \int_C f(z)\,dz, f(z)=z2f(z) = z^2
z-plane: the path C from A to B
xy
the products f(z_j)Δz_j, tip to tail
ReIm
  • path C (drag A and B)
  • bend it with P
  • chords and the chain of products
  • the integral along C
  • the integral along the segment AB
f(z)
Sum with n=6n = 6 chords: 0.6411−0.6495i0.6411 - 0.6495i. Integral along this path: 0.4587−0.144i0.4587 - 0.144i. Along the straight segment from A to B: 0.4587−0.144i0.4587 - 0.144i. The two integrals agree.
What to notice

For z2z^2 the integral does not move when you bend the path: it is always 13(B3−A3)\tfrac13(B^3 - A^3). For x−iyx - iy it moves with every change of shape. For 1/z1/z it stays put while you bend the path without crossing the origin, but jumps (by 2πi2\pi i) when the path is pulled across 00. That last behaviour is the logarithm, the subject of the next two lessons.

The explorer computes the sums; a definition must avoid limits of sums over arbitrary subdivisions and instead reduce everything to ordinary integrals of real functions, which we already have. That is what Hardy does: he describes the curve by a parameter.

Precise statements

Definition Real curvilinear integral Hardy: §229 (1)

Let the arc CC from AA to BB be given by x=ϕ(t)x = \phi(t), y=ψ(t)y = \psi(t) for t0≤t≤t1t_0 \le t \le t_1, where ϕ\phi and ψ\psi have continuous derivatives and the point moves along CC from AA to BB as tt increases. If g(x,y)g(x, y) and h(x,y)h(x, y) are continuous, then

∫C{g(x,y) dx+h(x,y) dy}=∫t0t1{g(ϕ,ψ) ϕ′(t)+h(ϕ,ψ) ψ′(t)} dt.\int_C \{g(x,y)\,dx + h(x,y)\,dy\} = \int_{t_0}^{t_1} \bigl\{ g(\phi, \psi)\,\phi'(t) + h(\phi, \psi)\,\psi'(t) \bigr\}\,dt.

CC is called the path of integration.

Definition Complex curvilinear integral Hardy: §229 (2)

With CC as above, put z=x+iy=ϕ(t)+iψ(t)z = x + iy = \phi(t) + i\psi(t), and let f(z)=u+ivf(z) = u + iv be a polynomial or rational function of zz (with no pole on CC). Then

∫Cf(z) dz=∫C(u+iv)(dx+i dy)=∫C(u dx−v dy)+i∫C(v dx+u dy),\int_C f(z)\,dz = \int_C (u + iv)(dx + i\,dy) = \int_C (u\,dx - v\,dy) + i\int_C (v\,dx + u\,dy),

the two integrals on the right being real curvilinear integrals as above.

Multiplying out under the integral sign, the definition says simply

∫Cf(z) dz=∫t0t1f(z(t)) z′(t) dt,z′(t)=ϕ′(t)+iψ′(t),\int_C f(z)\,dz = \int_{t_0}^{t_1} f\bigl(z(t)\bigr)\, z'(t)\,dt, \qquad z'(t) = \phi'(t) + i\psi'(t),

where the integral of a complex function of the real variable tt means the integral of its real part plus ii times the integral of its imaginary part. A path made of several smooth pieces (a polygon, say) is handled by adding the integrals over the pieces.

Two properties are used again and again.

Theorem Integrals of derivatives depend only on the end-points

Let FF be a polynomial or rational function of zz, with derivative F′F', and let CC be a path from AA to BB on which FF has no pole. Then

∫CF′(z) dz=F(B)−F(A),\int_C F'(z)\,dz = F(B) - F(A),

whatever the shape of CC. In particular ∫Czn dz=Bn+1−An+1n+1\int_C z^n\,dz = \dfrac{B^{n+1} - A^{n+1}}{n+1} for every integer n≥0n \ge 0, and for every integer n≤−2n \le -2 if CC avoids 00.

Lemma The size of an integral Hardy: cf. §170 (1)

If ∣f(z)∣≤M|f(z)| \le M on CC and CC has length L=∫t0t1∣z′(t)∣ dtL = \int_{t_0}^{t_1} |z'(t)|\,dt, then ∣∫Cf(z) dz∣≤ML\bigl|\int_C f(z)\,dz\bigr| \le ML.

The theorem leaves out exactly one power: n=−1n = -1. There is no rational function FF with F′(z)=1/zF'(z) = 1/z (exercise below), and that is why ∫dz/z\int dz/z can, and does, depend on the path.

Proofs

Proof Integrals of derivatives depend only on the end-points
Strategy

Compose FF with the path to get the complex function t↦F(z(t))t \mapsto F(z(t)) of the real variable tt. Show that its derivative is F′(z(t)) z′(t)F'(z(t))\,z'(t), which is exactly the integrand in the parametric form. Then the fundamental theorem of the integral calculus (§165), applied to the real and imaginary parts separately, does the rest.

Differentiate z(t)nz(t)^n

For a complex function w(t)=p(t)+iq(t)w(t) = p(t) + iq(t) of a real variable, define w′(t)=p′(t)+iq′(t)w'(t) = p'(t) + iq'(t). The product rule (w1w2)′=w1′w2+w1w2′(w_1 w_2)' = w_1' w_2 + w_1 w_2' holds, because it holds for the real and imaginary parts, which are sums of products of real functions. By induction, ddtz(t)n=nz(t)n−1z′(t)\frac{d}{dt} z(t)^n = n z(t)^{n-1} z'(t) for n≥1n \ge 1.

Extend to quotients

In the same way (1/w)′=−w′/w2(1/w)' = -w'/w^2 where w≠0w \ne 0 (differentiate w⋅(1/w)=1w \cdot (1/w) = 1). Hence for any rational function, ddtF(z(t))=F′(z(t)) z′(t)\frac{d}{dt} F(z(t)) = F'(z(t))\,z'(t), with F′F' given by the usual formal rules.

Integrate

Write F(z(t))=P(t)+iQ(t)F(z(t)) = P(t) + iQ(t). Then F′(z(t)) z′(t)=P′(t)+iQ′(t)F'(z(t))\,z'(t) = P'(t) + iQ'(t), and so ∫CF′(z) dz=∫t0t1P′ dt+i∫t0t1Q′ dt={P(t1)−P(t0)}+i{Q(t1)−Q(t0)}=F(B)−F(A).\int_C F'(z)\,dz = \int_{t_0}^{t_1} P'\,dt + i \int_{t_0}^{t_1} Q'\,dt = \{P(t_1) - P(t_0)\} + i\{Q(t_1) - Q(t_0)\} = F(B) - F(A).

Nothing depended on the shape of CC

Only the values of FF at the two ends appear. For several smooth pieces, the contributions telescope: {F(z1)−F(A)}+{F(z2)−F(z1)}+⋯=F(B)−F(A)\{F(z_1) - F(A)\} + \{F(z_2) - F(z_1)\} + \dots = F(B) - F(A).

Proof The size of an integral
Strategy

Reduce to the real inequality ∣∫g∣≤∫∣g∣|\int g| \le \int |g| for a complex function gg of a real variable, then bound ∣f(z(t)) z′(t)∣|f(z(t))\,z'(t)|.

∣∫g∣≤∫∣g∣\lvert\int g\rvert \le \int\lvert g\rvert

Let I=∫t0t1g(t) dtI = \int_{t_0}^{t_1} g(t)\,dt. If I=0I = 0 there is nothing to prove; otherwise choose c=Iˉ/∣I∣c = \bar I/|I|, of modulus 11, so that cI=∣I∣cI = |I| is real. Then ∣I∣=∫Re⁡{c g(t)} dt≤∫∣c g(t)∣ dt=∫∣g(t)∣ dt|I| = \int \operatorname{Re}\{c\,g(t)\}\,dt \le \int |c\,g(t)|\,dt = \int |g(t)|\,dt.

Bound the integrand

With g(t)=f(z(t)) z′(t)g(t) = f(z(t))\,z'(t) we have ∣g(t)∣≤M∣z′(t)∣|g(t)| \le M|z'(t)|, so ∣∫Cf dz∣≤M∫t0t1∣z′(t)∣ dt=ML\bigl|\int_C f\,dz\bigr| \le M\int_{t_0}^{t_1}|z'(t)|\,dt = ML.

Worked examples and exercises

Exercise One path, computed by hand ●○○ Hardy: §229

Let CC be the parabolic arc z=t+it2z = t + it^2, 0≤t≤10 \le t \le 1, from 00 to 1+i1 + i. Compute ∫Cz dz\int_C z\,dz from the definition and check it against the theorem.

Hint
z′(t)=1+2itz'(t) = 1 + 2it. Multiply out (t+it2)(1+2it)(t + it^2)(1 + 2it) and separate real and imaginary parts.

Solution

(t+it2)(1+2it)=t+2it2+it2−2t3=(t−2t3)+3it2(t + it^2)(1 + 2it) = t + 2it^2 + it^2 - 2t^3 = (t - 2t^3) + 3it^2. So ∫Cz dz=∫01(t−2t3) dt+i∫013t2 dt=(12−12)+i=i\int_C z\,dz = \int_0^1 (t - 2t^3)\,dt + i\int_0^1 3t^2\,dt = (\tfrac12 - \tfrac12) + i = i. The theorem gives 12{(1+i)2−0}=12⋅2i=i\tfrac12\{(1+i)^2 - 0\} = \tfrac12 \cdot 2i = i. ✓

Exercise x−iyx - iy depends on the path ●●○ Hardy: §229

Compute ∫zˉ dz\int \bar z\,dz from 11 to ii (a) along the straight segment, (b) along the quarter of the unit circle. Then compute it once round the whole unit circle, counter-clockwise.

Hint
(a) z=1+t(i−1)z = 1 + t(i - 1), 0≤t≤10 \le t \le 1. (b) z=cos⁡t+isin⁡tz = \cos t + i\sin t, 0≤t≤12π0 \le t \le \frac12\pi; on the circle zˉ=1/z\bar z = 1/z.

Solution

(a) z′(t)=i−1z'(t) = i - 1 and zˉ=1−t−it\bar z = 1 - t - it. The product is (1−t−it)(i−1)=(1−t)(i−1)−it(i−1)=−1+t+t+i(1−t+t)(1 - t - it)(i - 1) = (1 - t)(i - 1) - it(i - 1) = -1 + t + t + i(1 - t + t), i.e. (2t−1)+i(2t - 1) + i. Integrating over [0,1][0, 1] gives 0+i=i0 + i = i.

(b) On the circle zˉ z′(t)=(cos⁡t−isin⁡t)(−sin⁡t+icos⁡t)=i\bar z\,z'(t) = (\cos t - i\sin t)(-\sin t + i\cos t) = i, so the integral is ∫0π/2i dt=12πi\int_0^{\pi/2} i\,dt = \tfrac12\pi i.

The two answers differ, so zˉ\bar z has no “antiderivative” in the sense of the theorem. Once round the circle, the same computation gives ∫02πi dt=2πi\int_0^{2\pi} i\,dt = 2\pi i. In general ∮zˉ dz=2i×\oint \bar z\,dz = 2i \times (enclosed area), which is why every bend in the explorer changed the value.

Exercise Once round the origin with 1/z1/z ●○○ Hardy: §229

Let CC be the circle z=R(cos⁡t+isin⁡t)z = R(\cos t + i\sin t), 0≤t≤2π0 \le t \le 2\pi. Show that ∫Cdz/z=2πi\int_C dz/z = 2\pi i for every R>0R > 0, and that ∫Czn dz=0\int_C z^n\,dz = 0 for every other integer nn.

Hint
z′(t)=iz(t)z'(t) = iz(t).

Solution

Since z′=izz' = iz, ∫Czndz=∫02πizn+1 dt=iRn+1∫02π{cos⁡(n+1)t+isin⁡(n+1)t} dt\int_C z^n dz = \int_0^{2\pi} i z^{n+1}\,dt = iR^{n+1}\int_0^{2\pi}\{\cos(n+1)t + i\sin(n+1)t\}\,dt by De Moivre’s theorem. For n≠−1n \ne -1 both integrals of the trigonometric functions over a full period vanish. For n=−1n = -1 the integrand is ii and the integral is 2πi2\pi i. This agrees with the theorem: for n≠−1n \ne -1 the circle starts and ends at the same point, so F(B)−F(A)=0F(B) - F(A) = 0.

Exercise No rational antiderivative for 1/z1/z ●●● Hardy: §229

Show that there is no rational function FF with F′(z)=1/zF'(z) = 1/z.

Hint
If there were, the theorem would give ∫Cdz/z=0\int_C dz/z = 0 round any closed path avoiding the poles of FF. Which closed path is available?

Solution

Suppose F′=1/zF' = 1/z, with FF rational. FF has finitely many poles, so choose R>0R > 0 so small that no pole of FF other than possibly 00 lies in 0<∣z∣≤R0 < |z| \le R, and also so that no pole lies on the circle ∣z∣=R|z| = R. The circle is a closed path on which FF has no pole, so the theorem gives ∮dz/z=F(B)−F(A)=0\oint dz/z = F(B) - F(A) = 0. But the previous exercise gives 2πi2\pi i. Contradiction. So the integral of 1/z1/z is a genuinely new function, which is how the logarithm enters.

Exercise An estimate ●●○ Hardy: §229, with the lemma

Let CC be the upper half of the circle ∣z∣=R|z| = R from RR to −R-R. Show that ∣∫Cdzz2+1∣≤πRR2−1\bigl|\int_C \dfrac{dz}{z^2 + 1}\bigr| \le \dfrac{\pi R}{R^2 - 1} when R>1R > 1, so that the integral tends to 00 as R→∞R \to \infty.

Hint
On CC, ∣z2+1∣≥∣z∣2−1|z^2 + 1| \ge |z|^2 - 1.

Solution

On CC, ∣z2+1∣≥∣z2∣−1=R2−1>0|z^2 + 1| \ge |z^2| - 1 = R^2 - 1 > 0, so ∣1/(z2+1)∣≤1/(R2−1)=M|1/(z^2+1)| \le 1/(R^2 - 1) = M. The length is L=πRL = \pi R. By the lemma the integral has modulus at most πR/(R2−1)\pi R/(R^2 - 1), which tends to 00.

Check yourself

Check yourself Choose one

For which ff does ∫Cf(z) dz\int_C f(z)\,dz from A=1A = 1 to B=iB = i have the same value for every path (avoiding poles)?

Why

z2+3zz^2 + 3z is the derivative of the polynomial 13z3+32z2\tfrac13 z^3 + \tfrac32 z^2, so the theorem applies. x−iyx - iy and ∣z∣|z| are not functions of zz in Hardy’s sense and their integrals depend on the shape. For 1/z1/z the value depends on which side of 00 the path passes; there is no rational antiderivative.

Check yourself Enter a number

What is ∫Cz2 dz\int_C z^2\,dz round the unit circle, once counter-clockwise?

Why

z2z^2 is the derivative of 13z3\tfrac13 z^3, and on a closed path the end-points coincide, so the integral is 13(1−1)=0\tfrac13(1 - 1) = 0. The exceptional power is z−1z^{-1}, whose integral round the circle is 2πi2\pi i.

Check yourself True or false?

If f(z)f(z) is real for every zz on CC, then ∫Cf(z) dz\int_C f(z)\,dz is real.

Why

False. The steps dzdz are complex. For example f=1f = 1 along the segment from 00 to ii gives ∫Cdz=i\int_C dz = i. The integral is real for every real ff only if the path runs parallel to the real axis (so that every dzdz is real).

Check yourself Choose one

Reversing the direction of CC (from BB to AA instead of AA to BB)

Why

Reparametrise by s=t0+t1−ts = t_0 + t_1 - t. Then z′z' changes sign and the limits are swapped back, so the integral changes sign, exactly as ∫ab=−∫ba\int_a^b = -\int_b^a for real integrals. The direction is part of the path; that is why Hardy insists the point moves “from AA to BB”.

Hardy’s language

Summary

Summary
  • ∫Cf(z) dz\int_C f(z)\,dz is the limit of sums of products f(zj) Δzjf(z_j)\,\Delta z_j; by definition it is ∫t0t1f(z(t)) z′(t) dt\int_{t_0}^{t_1} f(z(t))\,z'(t)\,dt, split into real and imaginary parts.
  • If f=F′f = F' with FF rational, the integral is F(B)−F(A)F(B) - F(A) for every path: the shape does not matter.
  • ∫zn dz\int z^n\,dz is path-independent for every integer nn except n=−1n = -1; once round the origin, ∮dz/z=2πi\oint dz/z = 2\pi i.
  • ∣∫Cf dz∣≤ML|\int_C f\,dz| \le ML. The direction of the path is part of the data.