The Point at Infinity and the Riemann Sphere

The Riemann sphere is the coordinate physicists reach for whenever a two-sphere needs a single complex label. The most literal case is the qubit: the pure states of a two-level system are the Bloch sphere, and the stereographic coordinate $z=\tan(\theta/2)\,e^{i\phi}$ makes it the Riemann sphere exactly, with $|0\rangle$ and $|1\rangle$ at $0$ and $\infty$; unitary gates act as $SU(2)$ rotations, which in the $z$-chart are Möbius transformations. In relativity, an observer's sky is a celestial sphere, and a Lorentz boost acts on incoming ray directions as a Möbius transformation of it — relativistic aberration, worked out by Penrose and central to the twistor programme, where $\mathbb{P}^1(\mathbb{C})$ is the basic building block. Conformal field theory and string theory routinely trade the plane for the sphere via this conformal map, with correlators fixed by their values at the three marked points $0,1,\infty$. And numerically, the Majorana stellar representation of high-spin states as constellations on $S^2$, spin-coherent states, and stereographic grids for fluid and gravitational-wave data all rest on $\hat{\mathbb{C}}\cong S^2$. Wherever "the value at infinity" must be an honest data point rather than a limit, this is the structure doing the work.

Take the infinite flat plane and pull its entire far boundary together with a drawstring until it closes at a single point. Nothing finite is disturbed — every complex number keeps its place — but "off to infinity" now has a definite endpoint, and the surface has become a sphere. That endpoint is $\infty$, and the closing-up is what compact means: there is nowhere left for a sequence to run off to. The physical realisation is a lamp atop a globe. Shine a point source from the north pole $N$ of a sphere touching the plane, and the shadow of each sphere point lands at a unique plane point; run the rays backward and each plane point picks out one sphere point. Points near the bottom map near the origin, the equator maps to the unit circle, and points crowding toward $N$ fly off to ever-larger radius — so $N$ is the shadow of "infinitely far away". The first figure shows this cross-section with a real point carried through the map.

Because the projection is conformal, a tiny right angle inked on the sphere is still a right angle in its shadow: local shapes survive even though sizes do not. This is why analytic functions — conformal wherever their derivative is nonzero — feel at home on the sphere; the two structures speak the same angular language. The projection also turns the plane's lines and circles into a single family, the circles of $S^2$, with lines distinguished only as the circles that pass through $N$ — a straight line is just a circle through the point at infinity. The second figure makes compactness quantitative, plotting how far each finite point sits from $\infty$ in the chordal metric, the straight-line distance between sphere images.

Everything begins with a set-level definition: we take the plane and throw in one new element. The point $\infty$ is not a number to be added or multiplied freely — it is a topological limit point that closes up the space.

where: $\mathbb{C}$ — the ordinary complex plane; $\infty$ — the single adjoined ideal point; $\hat{\mathbb{C}}$ — the extended complex plane, compact and homeomorphic to $S^2$. Reading: one point, not two and not a circle of directions, closes the plane into a sphere. Stereographic projection makes the identification explicit. Projecting from the north pole $N=(0,0,1)$ of the unit sphere onto the equatorial plane $Z=0\cong\mathbb{C}$ gives every plane point $z=x+iy$ a home on $S^2$.

where: $z=x+iy$ — the plane point; $|z|^2=x^2+y^2$ — its squared modulus; $(X,Y,Z)$ — the corresponding sphere point, satisfying $X^2+Y^2+Z^2=1$ identically. Reading: small $|z|$ lands near the south pole $Z=-1$, $|z|=1$ lands on the equator $Z=0$, and $|z|\to\infty$ drives $Z\to+1$, the north pole. The inverse recovers $z$ from a sphere point, and shows directly why $N$ is the sole obstruction — the one place the map blows up.

where: $(X,Y,Z)$ — a sphere point with $Z\neq1$; $z$ — its plane image; $1-Z\to0$ — the north-pole limit, which sends $|z|\to\infty$. Reading: only $Z=1$ (the point $N$) has no finite image, so $N\leftrightarrow\infty$ is forced. To do analysis at infinity we change coordinates. The chart $w=1/z$ is centred on the north pole and turns the mysterious "large $z$" into the ordinary "small $w$".

where: $z$ — the chart valid near the south pole (finite points); $w$ — the chart valid near the north pole, with $w=0$ the point $\infty$; $g$ — the function $f$ expressed in the $w$-chart. Reading: $f$ is analytic, or has a pole / zero / essential singularity at $\infty$, exactly when $g$ does at $w=0$. The charts $z$ and $w=1/z$ overlap on $\mathbb{C}\setminus\{0\}$ with a holomorphic transition — precisely the statement that $\hat{\mathbb{C}}$ is a Riemann surface. Its natural distance is the chordal metric, the length of the straight chord between two sphere images.

where: $d$ — Euclidean chord length between images on $S^2$; $z_1,z_2$ — finite points; the second form is the $|z_2|\to\infty$ limit. Reading: $\infty$ is at finite distance ($\le2$) from every point, so $\hat{\mathbb{C}}$ is bounded and, being closed, compact. Pulling the round metric back to the plane exhibits the conformal factor and the whole point of the construction: locally the geometry is a positive multiple of the flat one.

where: $ds^2$ — the round metric of $S^2$ written in the $z$-chart; $\lambda(z)=2/(1+|z|^2)$ — the conformal factor. Reading: because $ds^2=\lambda^2(dx^2+dy^2)$ scales every direction equally, angles are preserved and the projection is conformal. Finally, viewing a rational map as a genuine self-map of $\hat{\mathbb{C}}$ enforces exact global accounting: what leaves as a zero must return as a pole somewhere on the sphere, including at $\infty$.

where: $\#$ — the count including the point $\infty$; the map is any nonconstant rational $R(z)=P(z)/Q(z)$. Reading: the common value is the degree of $R$ as a branched cover of the sphere.

The behaviour of standard functions at $\infty$ is read off from $g(w)=f(1/w)$ near $w=0$:

1 · Lifting points to the sphere and measuring chordal distances (fully worked)

Lift the plane points $z=0$, $z=1$, $z=i$, and $z=1+i$ to the Riemann sphere via equation (2); verify each lands on $S^2$; then compute the chordal distance of each to $\infty$ and the chordal distance between $z=1$ and $z=i$.

Step 1. Apply $(X,Y,Z)=(2x,2y,|z|^2-1)/(|z|^2+1)$. For $z=0$: $(0,0,-1)$, the south pole. For $z=1$ ($|z|^2=1$): $(1,0,0)$, on the equator. For $z=i$: $(0,1,0)$, also on the equator. For $z=1+i$ ($|z|^2=2$): $(2/3,\,2/3,\,1/3)$.

Step 2. Verify each is on $S^2$. The first three are immediate; for $z=1+i$, $X^2+Y^2+Z^2=\tfrac{4}{9}+\tfrac{4}{9}+\tfrac{1}{9}=\mathbf{1}$. Every image lies on the unit sphere, as the projection derivation guarantees.

Step 3. Chordal distances to infinity from $d(z,\infty)=2/\sqrt{1+|z|^2}$: $d(0,\infty)=2/\sqrt1=\mathbf{2}$ (the origin is antipodal to $N$, a full diameter away — the farthest any point can be), $d(1,\infty)=\mathbf{\sqrt2\approx1.414}$, and $d(1+i,\infty)=2/\sqrt3\approx\mathbf{1.155}$. The distances shrink monotonically as $|z|$ grows, matching the second figure.

Step 4. Chordal distance between $z=1$ and $z=i$ from equation (5): $d(1,i)=2|1-i|/\sqrt{(1+1)(1+1)}=2\sqrt2/2=\mathbf{\sqrt2}$. Cross-check on the sphere: images $(1,0,0)$ and $(0,1,0)$ have separation $\sqrt{1^2+1^2}=\sqrt2$. The chord formula and the raw 3-D distance agree exactly, confirming $d$ is the length of the chord between sphere images.

2 · Order at infinity and the zero-pole balance

For $R(z)=z^3/(z^2+1)$, locate every zero and pole on $\hat{\mathbb{C}}$ with multiplicity, determine the order at $\infty$, and confirm the balance of equation (7). Then exhibit a Möbius transformation that is defined at $\infty$.

Step 1. Finite zeros: $z^3=0$ gives a zero of order 3 at $z=0$. Finite poles: $z^2+1=0$ gives $z=\pm i$, each a simple pole, so 2 finite poles.

Step 2. Order at infinity: with $p=\deg P=3$, $q=\deg Q=2$, the order is $q-p=-1$, a simple pole at $\infty$. Explicitly $g(w)=R(1/w)=\dfrac{w^{-3}}{w^{-2}+1}=\dfrac{1}{w(1+w^2)}$, a simple pole at $w=0$, confirming the degree rule.

Step 3. Tally on $\hat{\mathbb{C}}$: zeros $=3$ (all at $z=0$); poles $=2$ (at $\pm i$) $+\,1$ (at $\infty$) $=3$. Zeros $=$ poles $=3$, so $R$ is a degree-3 branched self-cover of the sphere — equation (7) holds exactly.

Step 4. A Möbius example: the Cayley map $M(z)=\dfrac{z-i}{z+i}$ with $ad-bc=(1)(i)-(-i)(1)=2i\neq0$. It is defined everywhere on $\hat{\mathbb{C}}$: $M(-i)=\infty$ (the finite pole) and $M(\infty)=a/c=1$ (from the leading coefficients). So $M$ sends $i\mapsto0$, $-i\mapsto\infty$, $\infty\mapsto1$ — a genuine bijection of the sphere, the value at $\infty$ an ordinary point of its graph. Only on $\hat{\mathbb{C}}$ are $M(-i)$ and $M(\infty)$ legitimate values at all.

The point at infinity is where geometry, analysis, and physics converge: an astronomer's projection trick from antiquity, made into the founding object of a branch of mathematics by a single dissertation, and now the coordinate on every qubit and every patch of the night sky.

Texts. Ahlfors, Complex Analysis, 3rd ed., §1.2 (the Riemann sphere and stereographic projection) · Needham, Visual Complex Analysis, Ch. 3 (geometry of the sphere, conformality, and Möbius maps) · Stone & Goldbart, Mathematics for Physics, complex-analysis chapters · Arfken, Weber & Harris, Mathematical Methods for Physicists, 7th ed. (singularity at infinity, residue at infinity) · Hassani, Mathematical Physics, complex-analysis part · Byron & Fuller, Mathematics of Classical and Quantum Physics · Nakahara, Geometry, Topology and Physics (Riemann surfaces as the first example of a complex manifold). Four internal consistency checks tie the results together: (1) the plane-to-sphere map lands on $S^2$ via the root $t=2/(|z|^2+1)$ and the identity $X^2+Y^2+Z^2=1$; (2) the cancellation in the differential computation yields $ds^2=4(dx^2+dy^2)/(1+|z|^2)^2$; (3) the order at $\infty$ of $z^3/(z^2+1)$ follows from $g(w)=1/(w(1+w^2))$, giving the tally zeros $=$ poles $=3$; (4) the chordal distance $d(1,i)=\sqrt2$ agrees with the raw 3-D chord between $(1,0,0)$ and $(0,1,0)$.

Prerequisites. Complex Numbers and the Complex Plane · Functions of a Complex Variable and Limits · Elementary Complex Functions. Leads to. Laurent Series · Zeros, Poles, and Isolated Singularities · Consequences: Liouville, Maximum Modulus, Morera.

Ask the tutor: Curious how the same Riemann sphere becomes a qubit's Bloch sphere or a moving observer's night sky — ask, and we will trace the Möbius map from complex analysis to physics.