Part 15 of 15. The capstone. Where E₈ shows up outside lattice theory, what we deliberately skipped, and where to go next if you want to keep climbing.
The series promised E8 as a concrete object you can compute with. Fourteen explainers later you can generate the 240 roots, apply a Weyl reflection, project onto the Coxeter plane, peel off root subsystems, and derive the packing density π⁴/384.
This final explainer is the narrative capstone: where E8 connects outward to the rest of mathematics and physics, what we skipped, and where to go next.
In 1984 and 1985, the "heterotic string" model of David Gross, Jeffrey Harvey, Emil Martinec, and Ryan Rohm appeared as part of the first superstring revolution. A heterotic string is a 10-dimensional theory in which the left-moving and right-moving modes live in different-dimensional spaces: 10 dimensions for the right-movers, 26 for the left-movers, with the 16-dimensional difference compactified onto a 16-torus. Anomaly cancellation forces the gauge algebra to have dimension 496, and modular invariance selects the two even unimodular 16-dimensional gauge lattices. The resulting gauge groups are E8 × E8 and Spin(32)/ℤ2 (often described at the Lie-algebra level as so(32)). Both theories are internally consistent; the E8 × E8 version is more popular for phenomenological reasons related to how it breaks down to Standard Model physics.
The 16-dimensional torus that the left-movers live on is literally the E8 × E8 lattice, as a 16-dimensional even unimodular lattice. Compactifying onto this lattice is how the E8 × E8 gauge symmetry becomes the gauge group of the heterotic theory. This is a direct application of the lattice structure we've built: the lattice is the mathematical object that makes the string theory work.
On "applications" in physics. The heterotic string is a mathematically elegant model of quantum gravity. It is not a verified description of reality. No experiment has confirmed string theory, and whether it will ever make a testable prediction remains an open question. When we say "E₈ appears in physics," we mean it appears in physicists' models, not that those models are known to describe the universe. The same caveat applies to every physics application in this explainer.
In January 2007, the American Institute of Mathematics announced that a team of 18 mathematicians led by Jeffrey Adams had completed a four-year computation of the Kazhdan–Lusztig–Vogan polynomials for the split real form of E8. These are specific polynomials whose coefficients encode the structure of the unitary representations of the associated real Lie group, an enormous classification project that had been pending for decades. The computation produced 60 gigabytes of output, with individual polynomials occasionally having coefficients into the tens of millions (the largest was 11,808,808).
The headlines focused on size: the output file was briefly the largest single piece of mathematics ever computed, and the announcement compared it to "an area the size of Manhattan" if written out by hand. What matters is what it enables: detailed study of the unitary representations of E8's real form, previously out of reach.
The computation is a direct descendant of Kazhdan–Lusztig's 1979 conjectures, which proposed a combinatorial formula for certain representation-theoretic quantities. Proving and computing Kazhdan–Lusztig polynomials for large root systems is mechanically demanding but conceptually systematic. E8 was the hardest case because its Weyl group has 696,729,600 elements and the polynomial computation scales badly.
We met Maryna Viazovska in explainer 14 as the author of the 2016 E8 sphere packing proof. In 2022, she received the Fields Medal at the International Congress of Mathematicians in Helsinki (the ICM was originally scheduled for St Petersburg and moved in the wake of Russia's invasion of Ukraine). She became the second woman to receive the Fields Medal, after Maryam Mirzakhani in 2014, and the citation highlighted her proof that the E8 lattice gives the densest 8-dimensional sphere packing, along with related extremal and interpolation work. The 24-dimensional Leech lattice theorem was joint work with Cohn, Kumar, Miller, and Radchenko soon afterward.
Viazovska was born in Kharkiv (1984), trained in Kyiv and Bonn, and is at EPFL. The "magic function" technique has since been extended to spherical designs and other discrete-geometry problems. The 2016 paper is 23 pages and famously readable.
In November 2007, physicist Garrett Lisi posted a paper to the arXiv titled "An Exceptionally Simple Theory of Everything," proposing that E8 could serve as the unified gauge group for a theory combining the Standard Model of particle physics with general relativity. The idea was that elements of the E8 Lie algebra could be identified with known fundamental particles and forces, with the exceptional structure of E8 somehow constraining the physics in ways that matched observation.
In 2009 Distler and Garibaldi's "There is no 'Theory of Everything' inside E8" showed that the E8 embedding strategy they analyse cannot produce the observed chiral Standard Model spectrum. E8's representation theory does not provide three generations of chiral fermions with the right charges in the proposed framework.
The Distler–Garibaldi result is a genuine obstruction to Lisi's specific embedding. Variant proposals have appeared but none has accommodated Standard Model phenomenology while passing consistency checks.
Scope of the disagreement. Distler and Garibaldi's critique rules out Lisi's embedding and the straightforward E8 unification strategy analysed in their paper. It is not a theorem that E8 can never appear in future physics (the heterotic string already uses it), but it is a serious representation-theoretic obstruction to this class of proposals.
A quasicrystal is a solid whose atomic arrangement is highly ordered but not periodic; the atoms sit at specific positions that come from a long-range ordering rule that does not repeat. Shechtman and collaborators discovered physical quasicrystals in 1982 (confirmed 1984), eventually earning the 2011 Nobel Prize in Chemistry. One way to construct quasicrystal patterns mathematically is to project a higher-dimensional periodic lattice onto a lower-dimensional irrational subspace: a cut-and-project construction.
Some quasicrystal models with icosahedral symmetry (5-fold rotational, forbidden in crystals) come from projecting the E8 lattice onto a 3- or 4-dimensional subspace. E8's icosahedral symmetries become the rotational symmetries of the quasicrystal pattern.
The quasicrystal application is smaller than string theory but empirically grounded. It is where E8 touches the physical world in a verified way.
E⊥ slice: a 2D view of the 6D "hidden" space, with the acceptance window drawn as a circle. Blue dots lie inside (kept); grey dots lie outside (rejected).
Figure 1. Cut-and-project construction, live. We enumerate every E8 lattice point with squared length ≤ 8 (Ntotal ≈ 26k), split each into a physical piece in a 2-plane E∥ ⊂ ℝ8 and a hidden piece in the 6D complement E⊥, and keep only the points whose hidden piece lies inside the acceptance window W = {ξ ∈ E⊥ : ‖ξ‖ ≤ r}. The kept points, drawn at their E∥ coordinates, form an aperiodic tiling with "forbidden" (non-crystallographic) rotational symmetries. The window slider changes r. The plane slider interpolates E∥ between a golden-ratio 2-plane (t = 0) and the Coxeter plane (t = 1) — two natural choices that each give a different symmetry in the pattern. Play reveals the points one lattice shell at a time, showing how each shell of the 8D lattice contributes to the pattern.
The series treats E8 as a concrete geometric object. Several directions were deliberately left out:
E8 has an associated Lie algebra 𝔢8, a 248-dimensional real vector space with a bracket operation. The identity 248 = 240 + 8 reflects the decomposition of 𝔢8 into its 240 "root spaces" (one per root) and its 8-dimensional "Cartan subalgebra." The Lie algebra's representations (spaces on which 𝔢8 acts linearly) are a vast classical subject that we did not touch. In particular, we did not discuss weights, characters, highest weight representations, or the Weyl character formula. The adjoint representation (248-dimensional) is E8's smallest nontrivial irreducible representation, which is unusual among Lie algebras.
Start with: Fulton & Harris, Representation Theory: A First Course, part III. For E8 specifically: Humphreys, Introduction to Lie Algebras and Representation Theory, chapters 9–12.
Attached to E8 (and the Leech lattice) is a vertex operator algebra (VOA), a sophisticated algebraic structure introduced by Frenkel, Lepowsky, and Meurman. The E8 VOA is related to the moonshine module and ultimately to the monster group, the largest sporadic simple group, of order roughly 8 × 1053. Richard Borcherds proved the "monstrous moonshine" conjectures in 1992, establishing a deep and originally mysterious connection between the monster group, modular forms, and these lattice-based VOAs. A series about E8 could have spent ten explainers on moonshine alone; we chose not to.
Start with: Ronan, Symmetry and the Monster (narrative, accessible). Then: Frenkel, Lepowsky, & Meurman, Vertex Operator Algebras and the Monster.
E8 has an alternate realisation using the octonions, the largest of the four normed division algebras (after ℝ, ℂ, and ℍ). Specifically, the 240 roots of E8 can be identified with 240 unit octonions forming a specific finite subset closed under multiplication up to sign. This construction is beautiful but requires teaching octonion arithmetic first, which is its own climb of about three explainers, which would distract from the spine of this series.
Start with: John Baez, The Octonions (Bull. AMS 2002), freely available online. The final section discusses E8 and the exceptional Lie algebras from the octonionic perspective.
Every complex Lie algebra has several "real forms": real Lie algebras whose complexification is the given complex one. E8 has three real forms: the compact form (realising all of E8's symmetries in a compact space), the split form (all symmetries realised as unbounded linear operators), and E8(−24) (an intermediate form). The 2007 AIM computation was specifically about the split form E8(8).
Start with: Helgason, Differential Geometry, Lie Groups, and Symmetric Spaces, chapter X. For the AIM story specifically: the AIM E8 project page at aimath.org/E8/.
In 4-manifold topology, Freedman's E8 manifold is a topological 4-manifold whose intersection form is the E8 lattice. It is not a smooth manifold with an exotic smooth structure: Rokhlin's theorem and Donaldson's diagonalisation theorem imply that this definite E8 form cannot occur as the intersection form of a closed smooth simply connected 4-manifold. This gap between topological and smooth 4-manifolds is one of the landmarks of low-dimensional topology.
Start with: Donaldson & Kronheimer, The Geometry of Four-Manifolds. For a narrative: Freedman & Quinn, Topology of 4-Manifolds.
Individual historical timelines appeared in explainers 11, 12, and 13. Merged into one, the story from Kepler's cannonball conjecture (1611) to Viazovska's Fields Medal (2022) reads as a 411-year climb, with long stretches of quiet and a few dramatic leaps.
Figure 2. Four centuries of sphere packing and root system history. The 20th-century acceleration is real — most of the development happened between 1887 (Killing's Lie algebra classification) and 2022 (Viazovska's Fields Medal). The 1979 kissing number result and the 2016 packing density result sit 37 years apart in the narrative but use related LP-bound machinery.
Fourteen explainers ago, E8 was a name on T-shirts and arXiv papers. You now have:
This is not a complete picture of E8 — there isn't one — but a computationally grounded picture you can keep building on.
Everything the reader now knows, in one glance:
Figure 3. The numerical signature of E8. Each tile is a fact the series built up. If you can explain where each of these numbers comes from and how to compute it, you have the computational fluency the series aimed for.