# Chapter ᚎ(Z): The roundabout *Teclam Áirme, the Mathematical Compendium — a copy for AI readers of https://www.teclamairme.com/#ch-T16, as of 2026-09-30. Links are relative to https://www.teclamairme.com/fortheaireaders/.* [Guide for AI readers](index.md) · [Its glossary cards](ch16-cards.md) · [Its references](ch16-refs.md) · [Previous chapter](ch15.md) · [Next chapter](ch17.md) *A footnote such as [r-16.3] is reference entry 16.3 and links to it. Section 16 of 28; chapters are named by their Ogham letter, never by number.* ## The open road ᚎ – [Straif](ch16-cards.md#g-og-straif): associated with the blackthorn tree, Section 16 ## Chapter ᚎ(Z): The roundabout > “It has been written that the shortest and best way between two truths of the real domain often passes through the imaginary one.” > — Jacques Hadamard, after Paul Painlevé · *An Essay on the Psychology of Invention in the Mathematical Field*, 1945 [r-16.15](ch16-refs.md#r-16.15) Formula: ([∮_|z|=1](ch16-cards.md#g-contour) [e^z dz](ch16-cards.md#g-cauchy) + [∂(e^z)/∂z̄](ch16-cards.md#g-wirtinger) + [Res₀ e^z](ch16-cards.md#g-resentire))/([∮_|z|=1 dz/z](ch16-cards.md#g-dzz) + [Res₀ e^1/z](ch16-cards.md#g-res) + [n(γ, 0)](ch16-cards.md#g-winding) + [#{z : z² + 1 = 0, |z| < 2}](ch16-cards.md#g-argprin)) #### Type of number Non-real object · Key concepts: Complex analysis: contour integrals, Cauchy’s theorem, the Cauchy–Riemann equations, and residues. It includes symbols you may not have seen before, such as: [∮](ch16-cards.md#g-contour), [∂/∂z̄](ch16-cards.md#g-wirtinger), and [Res](ch16-cards.md#g-res). Zero here is a round trip: carry e^z once around a circle and you come home with nothing. I checked the map and know exactly where to go: at the roundabout ahead, we go north. But let’s take a few spins around it first, just for fun! The central island gives us something to go around, and a complete circuit brings us back to where we joined. In [the complex plane](ch21-cards.md#g-cplane), the route alone will not tell us what we bring back: that also depends on the function we take with us and what lies inside the circle. Field notes A complex number z = x + iy is a point with horizontal coordinate x and vertical coordinate y, where i² = −1; Chapter ᚔ(I) returns to how its arithmetic is built. Complex analysis studies functions whose inputs and outputs live in this plane. On the real line a point can be approached from only two sides; in the plane it can be approached from every direction, and asking the derivative to come out the same from all of them is a much stronger demand. A function that meets it on an open region is called [holomorphic](ch16-cards.md#g-holomorphic). [r-16.8](ch16-refs.md#r-16.8) The demand makes such functions rigid: [their values inside a loop are fixed by their values on the loop](ch16-cards.md#g-cauchyformula), which is why a trip around a loop can report on what lies inside it. [r-16.16](ch16-refs.md#r-16.16) Most terms here are a trip around a circle, or a count of what such a trip detects; the one that is not, ∂/∂z̄, is the test for being holomorphic: passed everywhere inside a loop, it sends a trip around that loop home with nothing. #### Numerator [∮_|z|=1](ch16-cards.md#g-contour) [e^z dz](ch16-cards.md#g-cauchy) = 0; [Cauchy’s theorem](ch16-cards.md#g-cauchy), the founding result of the subject, on the friendliest function there is. The [contour integral](ch16-cards.md#g-contour) follows the unit circle once counterclockwise. The derivative of eᶻ is eᶻ, so eᶻ is its own antiderivative: integrating it along a path just measures how much eᶻ changes between the start and the end. A closed path ends where it began, so the change, and the integral, is zero. More generally, [Cauchy’s theorem](ch16-cards.md#g-cauchy) gives a zero integral around a simple closed curve whenever the function is [holomorphic](ch16-cards.md#g-holomorphic) on a region containing the curve and its inside. [r-16.2](ch16-refs.md#r-16.2) [∂(e^z)/∂z̄](ch16-cards.md#g-wirtinger) = 0; [the Cauchy–Riemann equations](ch16-cards.md#g-wirtinger) packed into one line: the defining test for being [holomorphic](ch16-cards.md#g-holomorphic). Here z̄ = x − iy is z reflected in the horizontal axis, called its conjugate, and ∂/∂z̄ is shorthand for ½(∂/∂x + [i](ch21-cards.md#g-i)∂/∂y), a blend of the rates of change in the horizontal and vertical directions. For a function whose ordinary real rates of change exist and vary continuously, this one derivative is zero exactly when [the Cauchy–Riemann equations](ch16-cards.md#g-wirtinger) hold: the conditions that make the derivative the same from every direction. e^z is [holomorphic](ch16-cards.md#g-holomorphic) everywhere, so for it the term is zero: the same function that walked the circle above. [r-16.5](ch16-refs.md#r-16.5) [Res₀ e^z](ch16-cards.md#g-resentire) = 0; a residue that is not there, chosen to set up [the residue](ch16-cards.md#g-res) that is. A *residue* is read off a function’s [Laurent expansion](ch16-cards.md#g-laurent) about a point, a power series that is also allowed negative powers: about the origin, it is the coefficient of 1/z, the one coefficient that can survive integration around that point. Expanding eᶻ gives 1 + z + z²/2! + ⋯, with no negative powers at all, so the coefficient of 1/z is 0. [r-16.4](ch16-refs.md#r-16.4) #### Denominator [∮_|z|=1 dz/z](ch16-cards.md#g-dzz) = 2πi; the integral that refuses to vanish: same path, same direction, a different function. 1/z fails at the origin, inside our path, so [Cauchy’s theorem](ch16-cards.md#g-cauchy) does not apply. Put z = eⁱᵗ for 0 ≤ t ≤ 2π; then dz/z = [i](ch21-cards.md#g-i) dt, and the full turn gives 2πi. The missing point changes the answer. Along any short stretch of the path, 1/z does have an antiderivative, a logarithm; but each counterclockwise circuit of the origin adds 2πi to the logarithm, so no [single-valued logarithm](ch16-cards.md#g-riemannsurf) covers the whole loop. The path ends where it began, the logarithm does not: it ends 2πi higher, and that difference is the integral. [r-16.3](ch16-refs.md#r-16.3) [Res₀ e^1/z](ch16-cards.md#g-res) = 1; a residue at [an essential singularity](ch16-cards.md#g-picard), the wildest kind of failure, which still leaves a simple trace. e^1/z = 1 + 1/z + 1/(2z²) + ⋯. Infinitely many negative powers make the origin an [essential singularity](ch16-cards.md#g-picard), but its residue is simply the coefficient of 1/z: 1. The [residue theorem](ch16-cards.md#g-residuethm) turns the residues inside a loop into its integral: 2πi times their sum. [Great Picard’s theorem](ch16-cards.md#g-picard) describes the stranger behavior nearby: in every punctured neighborhood of the origin, e^1/z takes every nonzero complex value infinitely often. [r-16.4](ch16-refs.md#r-16.4) [r-16.10](ch16-refs.md#r-16.10) [n(γ, 0)](ch16-cards.md#g-winding) = 1; the winding number, our roundabout in one number. It counts how many times a closed curve goes around the origin, with orientation. Our counterclockwise unit circle goes around once; a second circuit would make it 2, and reversing the arrow would give −1. The normalized integral (1/2πi)[∮ dz/z](ch16-cards.md#g-dzz) measures exactly this count, which is why [∮ dz/z](ch16-cards.md#g-dzz) is 2πi and not 0. [r-16.6](ch16-refs.md#r-16.6) [#{z : z² + 1 = 0, |z| < 2}](ch16-cards.md#g-argprin) = 2; the argument principle, which counts roots without solving for them. The zeros of z² + 1 inside |z| < 2 are i and −i. [The argument principle](ch16-cards.md#g-argprin) counts them another way: for a function f, (1/2πi)∮ f′/f around a contour gives the number of zeros minus the number of poles inside, a double root counted twice, provided no zero or pole lies on the path itself. A polynomial has no poles, so around |z| = 2 the count is just the two roots. Such counts matter most when we cannot find the roots individually. The same two roots return at the first landing of the launch tower, counted there by a different theorem. [r-16.7](ch16-refs.md#r-16.7) The only fraction is 0/(4 + 2πi): 2πi + 1 + 1 + 2 = 4 + 2πi, and its real part is 4, so it is not zero. **Fine print.** ∂/∂z̄ is a convenient differential operator, not permission to treat z and its conjugate as independent coordinates. [The residue theorem](ch16-cards.md#g-residuethm) needs finitely many isolated singularities inside the contour, poles or essential ones like that of e^1/z, and none on the contour itself. The quotient is complex zero, read as the ordinary scalar 0. What this chapter adds to the formula: 0 / (4 + 2πi) = 0 The circuit with eᶻ returned zero, while the circuit with 1/z kept an exact record of going around the missing point. A boundary has told us about what lies inside it. Up ahead I see a mechanic’s workshop, and I’d like a little more horsepower before the road gets harder. At least one tool on those benches is made the way we just saw: a function defined by what an integral around a loop brings back. Let’s pull in and get some upgrades. Watch and read - [Complex integration, Cauchy and residue theorems | Essence of Complex Analysis #6](https://www.youtube.com/watch?v=EyBDtUtyshk) Mathemaniac: why the integral of 1/z is the only one that survives, drawn rather than derived - [Complex Analysis L10: Cauchy Integral Formula](https://www.youtube.com/watch?v=4epESGF3qcU) a university lecture, but a clear one, on what the formula lets you do - [Cauchy’s integral theorem](https://en.wikipedia.org/wiki/Cauchy%27s_integral_theorem) Wikipedia - [Residue theorem](https://en.wikipedia.org/wiki/Residue_theorem) Wikipedia