# Chapter ᚜(Epilogue): The homeward view
*Teclam Áirme, the Mathematical Compendium — a copy for AI readers of https://www.teclamairme.com/#ch-T28, as of 2026-09-30. Links are relative to https://www.teclamairme.com/fortheaireaders/.*
[Guide for AI readers](index.md) · [Its glossary cards](ch28-cards.md) · [Its references](ch28-refs.md) · [Previous chapter](ch27.md)
*A footnote such as [r-28.3] is reference entry 28.3 and links to it. Section 28 of 28; chapters are named by their Ogham letter, never by number.*
## The view home
᚜ – [Reversed feather mark](ch28-cards.md#g-og-feather-rev): the end, Section 28
## Chapter ᚜(Epilogue): The homeward view
> “Who of us would not be glad to lift the veil behind which the future lies hidden; to cast a glance at the next advances of our science and at the secrets of its development during future centuries?”
> — David Hilbert · [Mathematical Problems](ch28-cards.md#g-hilbert23), address to the International Congress of Mathematicians, Paris, 8 August 1900 [r-28.2](ch28-refs.md#r-28.2)
Formula: [ΛdBN](ch28-cards.md#g-dbn) {brace: [Riemann](ch28-cards.md#g-critline)} · [L(E₃₇, 1)](ch28-cards.md#g-lE37) {brace: [Birch, Swinnerton-Dyer](ch28-cards.md#g-rankE)} · [[every simply connected closed 3-manifold is S³]](ch28-cards.md#g-poincare) {brace: [Poincaré](ch28-cards.md#g-ricci)} · [[P = NP]](ch28-cards.md#g-pnp) {brace: [P versus NP](ch28-cards.md#g-npcomplete)} · [[every Hodge class is algebraic]](ch28-cards.md#g-hodge) {brace: [Hodge](ch28-cards.md#g-hodgeclass)} · [1/T∗](ch28-cards.md#g-nsrate) {brace: [Navier–Stokes](ch28-cards.md#g-weaksol)} · [[a non-trivial quantum Yang–Mills theory exists on ℝ⁴ and has Δ>0]](ch28-cards.md#g-ymbit) {brace: [Yang–Mills](ch28-cards.md#g-ymgap)}
#### Type of number
Status open ·
Key concepts: The seven Millennium Prize Problems, each written as a finite number. It includes symbols you may not have seen before, such as: [ΛdBN](ch28-cards.md#g-dbn), [L(E37, 1)](ch28-cards.md#g-lE37), and [1/T∗](ch28-cards.md#g-nsrate). Zero here does not wait for answers: one proven zero settles the whole product while the open answers stay unknown.
There it is: home, filling the window. As we descend from orbit I can trace the route we took more clearly: the flow and the shapes in the museum, the train’s instructions, the prime mileposts, and the curves at the crossroads. Ideas we met one at a time are meeting again. We can now recognize more of what the hardest questions ask, including those whose answers nobody yet knows.
In 2000 the Clay Mathematics Institute chose seven [Millennium Prize Problems](ch28-cards.md#g-clay) and allocated one million dollars to each. I wanted them at the end for two reasons. First, the symbols and formulas should now feel less foreign. Second, and more importantly, there is an irony in where we finish. We began with the novel and exotic: a number nobody can pin down that still gave a definite answer. We trekked through the well established. Now we end with seven problems whose every moving part is precisely defined, each sitting in a formula we can fluently evaluate, yet on six of them we have bupkus for solutions. [r-28.1](ch28-refs.md#r-28.1)
Field notes
There is no outer fraction this time. Each of the seven problems supplies one finite number to a product; the last section explains why no common denominator is needed.
### [Riemann](ch28.md#ch-T28-riemann) {#ch-T28-riemann}
The old mileposts are our first landmark in this view. The question returns from [zeta](ch18-cards.md#g-zeta): do all its nontrivial zeros lie on [the line Re(s) = 1/2](ch28-cards.md#g-critline)? Their locations control the fluctuations in the distribution of primes, so a statement about a complex function would tell us how tightly an ancient collection of integers follows its expected pattern. [r-28.3](ch28-refs.md#r-28.3)
[ΛdBN](ch28-cards.md#g-dbn) = 0 or a positive real; the de Bruijn–Newman constant, chosen because it turns [the Riemann Hypothesis](ch18-cards.md#g-rh) into the question of whether one number vanishes.
Take a function built from zeta and let it flow the way heat spreads. Run the flow far enough and every one of its zeros lies on the [real line](ch01-cards.md#g-reals); the point where that starts to happen is [ΛdBN](ch28-cards.md#g-dbn). [The Riemann Hypothesis](ch18-cards.md#g-rh) says the zeros are there already, before any flow at all, which is the statement ΛdBN ≤ 0. Rodgers and Tao proved ΛdBN ≥ 0, so the hypothesis is true exactly when ΛdBN = 0. If it is false, the number lies above 0 and no higher than 0.2; either way it is a definite finite real. [r-18.20](ch18-refs.md#r-18.20)
**Fine print.** Turning this conjecture into “does one real number vanish?” is a property of this particular construction, not a general recipe for turning any question into a zero.
### [Birch and Swinnerton-Dyer](ch28.md#ch-T28-birch-and-swinnerton-dyer) {#ch-T28-birch-and-swinnerton-dyer}
At the crossroads, the curve chapter kept two things apart: a curve’s [rank](ch28-cards.md#g-rankE), the number of independent rational points of infinite order it has, and its point counts modulo each prime. Birch and Swinnerton-Dyer, working from those counts, conjectured that the second knows the first: package the counts into the curve’s L-function, and the order to which it vanishes at s = 1 should equal the rank. [r-28.5](ch28-refs.md#r-28.5) [r-28.27](ch28-refs.md#r-28.27)
[L(E₃₇, 1)](ch28-cards.md#g-lE37) = 0; a known, established zero, chosen so the whole product is settled whatever the open factors turn out to be.
Our curve is [E₃₇](ch28-cards.md#g-E37): [y² + y = x³ − x](ch28-cards.md#g-E37), with conductor 37. Its [rank](ch28-cards.md#g-rankE) is 1, and its L-function vanishes at s = 1 to order exactly 1: for this curve the two sides agree, as the conjecture predicts. It is one example, not a number whose vanishing would decide the conjecture for every elliptic curve. [r-19.8](ch19-refs.md#r-19.8)
### [Poincaré](ch28.md#ch-T28-poincar) {#ch-T28-poincar}
The gallery of shape brought us loops and surfaces. Poincaré asked about a [closed three-dimensional manifold](ch28-cards.md#g-threemfld): a space locally like ordinary three-dimensional space, compact and without boundary. If it is connected and every loop can contract to a point, must it be the three-sphere S³, the three-dimensional analogue of the usual sphere’s surface? [r-28.6](ch28-refs.md#r-28.6)
[[every simply connected closed 3-manifold is S³]](ch28-cards.md#g-poincare) = 1; the one Millennium Problem already solved, a reminder that an open question can acquire an answer.
Conjectured in 1904, it was settled by Perelman’s work in 2002–2003, which drove [Ricci flow](ch28-cards.md#g-ricci) through its singularities. [r-28.26](ch28-refs.md#r-28.26) The statement is true, so the bracket is 1.
**Fine print.** “Is S³” means is [homeomorphic](ch12-cards.md#g-homeo) to S³: one can be stretched into the other without cutting or gluing, as a topologist’s coffee cup becomes a doughnut; precisely, there is a continuous [bijection](ch27-cards.md#g-bijection) between them with continuous inverse.
### [P versus NP](ch28.md#ch-T28-p-versus-np) {#ch-T28-p-versus-np}
Back in the locomotive, the computation chapter showed that [P ⊆ NP](ch09-cards.md#g-pnpsub). The unanswered direction asks whether polynomial-time checking of yes answers always comes with a polynomial-time method for finding them. Given a proposed route visiting every vertex of a graph exactly once, we can check it efficiently; finding one, or showing that none exists, is another task. Solving any one of [the hardest problems in NP](ch28-cards.md#g-npcomplete) quickly would give P = NP. [r-28.7](ch28-refs.md#r-28.7)
[[P = NP]](ch28-cards.md#g-pnp) = 0 or 1; the most famous open question in computer science, written as a single bit.
The bracket is 1 if every problem whose answer can be checked quickly can also be solved quickly, and 0 if not. It is a definite integer that nobody knows; most complexity theorists expect 0.
**Fine print.** In the book’s classical setting, lack of a proof does not create a third truth value. Polynomial time also does not mean every instance would become practically easy; the statement concerns asymptotic resources.
### [Hodge](ch28.md#ch-T28-hodge) {#ch-T28-hodge}
The crossroads and the gallery of shape each hold half of this question: equations can define a shape, while topology studies its global structure. A [smooth complex projective variety](ch28-cards.md#g-smoothproj) is a shape cut out by polynomial equations in complex coordinates, with no corners or self-crossings and nowhere to run off to infinity. It has algebraic structure from its equations and topological structure studied through [cohomology](ch12-cards.md#g-cohomology), which records global features as classes rather than literal subshapes.
[[every Hodge class is algebraic]](ch28-cards.md#g-hodge) = 0 or 1; the question of whether two ways of seeing a space always agree, written as a bit.
The plain question: of the global features a shape has, which can be accounted for by pieces cut out by polynomial equations? Every such piece, an algebraic subvariety, leaves a class in the [cohomology](ch12-cards.md#g-cohomology), and those classes always have one special form, called type (p, p); the rational classes of that form are the [Hodge classes](ch28-cards.md#g-hodgeclass). The conjecture asks for the converse: is every Hodge class built, with fractional weights, from the classes those pieces leave? Having the special form is necessary for a class to come from pieces; the question is whether it is enough. The bracket is 1 if so and 0 if some class is not. [r-28.21](ch28-refs.md#r-28.21) [r-28.8](ch28-refs.md#r-28.8)
**Fine print.** The abbreviated bracket includes the rational coefficients and the smooth projective setting. It does not claim that every topological feature is itself a single polynomially defined object.
### [Navier–Stokes](ch28.md#ch-T28-navier-stokes) {#ch-T28-navier-stokes}
The gallery of motion returns with a question about how a smooth flow can change over time. Can a smooth three-dimensional incompressible viscous flow develop a singularity in finite time, or does its smooth solution persist? Viscosity tends to smooth the velocity, while nonlinear transport and vortex stretching can concentrate motion. [Weak solutions](ch28-cards.md#g-weaksol) are known to exist forever; whether they are also smooth is the prize. [r-28.9](ch28-refs.md#r-28.9) In September 2026 a finite-time singularity was announced for a flow driven by an external force, and the Clay Mathematics Institute relabeled its problem page Active. The announcement had not been refereed as of late September 2026. If it holds, it settles Clay’s problem through the forced case, but the number below is defined over unforced flows, so the question this book is counting is still open. [r-28.22](ch28-refs.md#r-28.22)
[1/T∗](ch28-cards.md#g-nsrate) = 0 or a positive real; a blow-up rate designed for this book, so that the question becomes one finite number.
I wanted one finite number that records whether any unforced flow on ℝ³ in a chosen class breaks down: an endless smooth lifetime contributes zero, a finite one a positive reciprocal. Start with smooth, divergence-free, rapidly decaying velocity fields on ℝ³, fix viscosity to 1, and restrict their [enstrophy](ch28-cards.md#g-enstrophy), ∫|∇u₀|², to at most 1. Let T(u₀) be the maximal smooth lifespan, and define
(1)/(T∗) := supu₀ (1)/(T(u₀)), (1)/(∞) := 0
The supremum is the least upper bound, and no single worst starting flow need reach it. Leray showed that capping the enstrophy guarantees every flow in the class a minimum smooth lifetime, so the value is finite. And any smooth, rapidly decaying, unforced flow that does break down can be rescaled, stretched in time, until its starting [enstrophy](ch28-cards.md#g-enstrophy) fits under the cap, still breaking down. So the value is zero exactly when every such flow, of any starting enstrophy, continues smoothly forever. [r-28.10](ch28-refs.md#r-28.10)
**Fine print.** This number is built for the **unforced ℝ³** setting and should not be presented as a detector for every version of the prize problem; the official formulation has distinct existence and breakdown statements on ℝ³ and in periodic space. Without the normalization, scaling would defeat the finite bound. An externally forced construction adds a term to the equations and cannot automatically settle this unforced diagnostic.
### [Yang–Mills](ch28.md#ch-T28-yang-mills) {#ch-T28-yang-mills}
Our physics and symmetry chapters meet here, bringing the ground beneath our first steps together with the transformations of the hall of mirrors. Yang–Mills theory describes fields organized by [gauge symmetries](ch28-cards.md#g-gauge). Its quantum form is central to the mathematics of fundamental interactions, but the prize asks for a rigorous construction with specified properties, not just a formal expression used in calculations. [r-28.11](ch28-refs.md#r-28.11)
[[a non-trivial quantum Yang–Mills theory exists on ℝ⁴ and has Δ>0]](ch28-cards.md#g-ymbit) = 0 or 1; the whole problem in one bracket, because the gap cannot be a number until the theory exists.
The task: for every compact simple gauge group, construct a nontrivial quantum [Yang–Mills theory](ch28-cards.md#g-ymgap) on ℝ⁴ satisfying appropriate rigorous axioms, and prove that it has a positive [mass gap](ch28-cards.md#g-ymgap): even the lightest particle the theory allows has a mass strictly above zero, so there is a gap between empty space and the first thing the theory can make. The bracket is 1 if both parts hold and 0 if not, so it is never undefined. [r-28.12](ch28-refs.md#r-28.12)
**Fine print.** The bracket includes both tasks and the quantification over gauge groups, even though its displayed label is abbreviated.
### [The completed product](ch28.md#ch-T28-the-completed-product) {#ch-T28-the-completed-product}
Not every entry stands for its whole problem: Riemann’s number and the four brackets do, [Birch and Swinnerton-Dyer](ch28-cards.md#g-bsdconj)’s is one proven example, and Navier–Stokes’s is scoped to one setting. There is no common denominator. The reciprocal-lifespan factor has its own explicit convention and finite-bound argument; the other factors are finite real quantities or truth values. We never divide by an unresolved quantity whose nonvanishing would have to be assumed.
The completed product of seven factors is 0: [L(E₃₇, 1)](ch28-cards.md#g-lE37) is already zero and every other factor is finite, [Poincaré’s bracket](ch28-cards.md#g-poincare) among them already known to be one, so the unsettled factors cannot change the outcome. Its value comes from established mathematics, while its contents leave room for what we have not yet learned.
What this chapter adds to the formula: [ΛdBN](ch28-cards.md#g-dbn) · 0 · 1 · [[P = NP]](ch28-cards.md#g-pnp) · [[Hodge]](ch28-cards.md#g-hodge) · [1/T∗](ch28-cards.md#g-nsrate) · [[Yang–Mills]](ch28-cards.md#g-ymbit) = 0
I began with Rose’s name beside a number our chosen language could not reveal, and from this window I find myself thinking of the elder tree again. Thank you for letting me come along with you: through the garden and the park, across the city and the museum, up the ladder, and out here into orbit. The journey has changed what I notice on an ordinary walk from home: a loop of rope, a turn, a gap in the ground, a handful of counters, a point with no chance of being drawn exactly. The things were there before we left.
Mathematics has a reputation for being cold, dry, and purely logical. It is logical, but it can also be whimsical, playful, beautiful, and surprisingly often accidental. This formula has seated some two dozen frameworks in the same room, and a few of them have already started talking: matrices came back in a weather forecast, the same [G₂](ch14-cards.md#g-g2) returned as the symmetry of a multiplication, and the curves at the crossroads finished [Fermat’s story](ch08-cards.md#g-fermat) from the observation car. Maybe sitting side by side will start a few more conversations. Maybe it will nudge you toward a kind of number nobody has named yet. One example to start with… I dunno… might involve a relation to time, a cousin of the numbers we “kinda” knew back in the park: something like… “a number we do not know… yet… but assuredly will in exactly ten and a half minutes.” How would we even write that? What would its purpose be—does it even have one?
Your flight bag is still empty. That seems right for an equation that, at its heart, says 0 = 0, and yet we are bringing home more ways of seeing than any bag could hold. The product was settled at the first sign we passed, thirteen zeros back at the trailhead; everything after it was the reason to keep walking. Stay by the window a little longer. I can picture the garden, the park beyond it, and the elder tree where this began. There is a great deal of forest still to explore. So the next time any math looks daunting, remember [Rose’s number](ch01-cards.md#g-rose) and our pleasant journey. Instead of a wall of confusion, I hope you’ll see an inviting gateway to a fabulous garden.
Watch and read
- [The Million Dollar Equations - with Tom Crawford](https://www.youtube.com/watch?v=f251NkeDVB8) The Royal Institution: one lecture that walks through all seven prizes for a general audience
- [Riemann Hypothesis - Numberphile](https://www.youtube.com/watch?v=d6c6uIyieoo) Numberphile with Edward Frenkel: the prize that opens the epilogue, and the one most likely to be worth a million
- [Millennium Prize Problems](https://en.wikipedia.org/wiki/Millennium_Prize_Problems) Wikipedia
- [De Bruijn–Newman constant](https://en.wikipedia.org/wiki/De_Bruijn%E2%80%93Newman_constant) Wikipedia