One function, all the primes.
For \(\operatorname{Re}s>1\), a sum over integers equals a product over primes. Analytic continuation then carries the zeta function into the critical strip.
PRIMES, ZEROS, AND THE CRITICAL LINE
An extraordinary claim. A vast formal argument. A place to begin understanding it.
Explore the mathematics behind the Riemann hypothesis, follow the structure of Alpöge’s released Lean source, and trace the nonvanishing theorem and its dependencies, then work through the final deduction.
This first edition is a guided reading of the source. The explanations and website checks here are separate from an independent build of the complete Lean project.
First edition · 20 November 2026 Overview, nonvanishing dependencies, and a source-linked deduction.
Read the scope and sourcesThe zeta function brings the primes into analysis. Its Euler product records prime factorization; its zeros govern the oscillations in prime-counting formulas. The Riemann hypothesis predicts an exact location for all of its nontrivial zeros.
Every zero of the zeta function in the critical strip has real part one half.
The negative even integers are the trivial zeros. The point \(s=1\) is a pole. The technical notes explain how these exceptions appear in the source’s full statement.
For \(\operatorname{Re}s>1\), a sum over integers equals a product over primes. Analytic continuation then carries the zeta function into the critical strip.
Dirichlet characters weight the integers and give functions \(L(s,\chi)\). The released source states nonvanishing and critical-line results for this family as well as RH.
This guide follows the declarations in the released statement file: the Riemann zeta function and Dirichlet L-functions. “Dirichlet GRH” here means the critical-line assertion for those L-functions. It is not a claim here about every automorphic L-function or every possible generalization of RH.
The defining series above converge in the right half-plane \(\operatorname{Re}s>1\). Statements about zeros in the critical strip concern their analytic continuations.
For a primitive character, the functional equation relates the completed L-function at \(s\) to the one for the inverse character at \(1-s\). A hypothetical zero to the left of the line therefore produces a zero to its right.
Assume that every Dirichlet L-function is nonzero when \(\operatorname{Re}s>1/2\), apart from the pole exception. A reflected zero in that region is then impossible. This strong nonvanishing input is the hard theorem; the reflection argument is the final deduction.
Original: 0.30 + it
Reflected: 0.70 − it
A zero at real part 0.30 would give a zero for the inverse character at real part 0.70. The nonvanishing input excludes that reflected zero.
Follow every step of the argumentThe released source brings together algebraic number theory, automorphic forms, arithmetic geometry, and complex analysis. Start with the short final deduction, then follow the deeper inputs that supply its nonvanishing theorem.
The central input says that every Dirichlet L-function is nonzero to the right of one half, excluding the principal-character point \(s=1\):
For a primitive character, the functional equation sends a zero in the left half of the critical strip to a right-hand zero for the inverse character. The input rules that out. Zeros already on the right are excluded directly, leaving only the critical line. Finite Euler factors handle imprimitive characters; the zeta case yields RH.
Three kinds of input meet here: bounds on Galois cohomology for selected subfields of cyclotomic fields; geometric bounds on covolumes and slopes of integral lattices; and matching results for theta integrals on unitary groups. Local Langlands parameters for \(\mathrm{GL}_n\), with the source’s stated compatibility hypotheses, connect the automorphic and Galois data.
The intermediate output is an upper bound for a smoothed prime-sum error on arbitrarily late, long intervals. A Landau argument converts these bounds into nonvanishing: a forbidden zero would create a pole whose derivative growth contradicts the interval estimates.
The cohomology groups are Tate–Shafarevich-like localization kernels: they measure global Galois cohomology classes whose restrictions vanish at the specified places above \(\ell\). They are not simply the usual Tate–Shafarevich groups of elliptic curves. The fields vary through a particular family of cyclotomic subfields.
The window construction passes through lifting and automorphy arguments. Its inputs include Galois representations attached to regular algebraic, polarized cuspidal automorphic representations of \(\mathrm{GL}_n\) over CM fields, together with base change, Jacquet–Shalika, Clozel’s purity, Bernstein–Zelevinsky, the main theorem of complex multiplication, and Deligne–Ribet.
The resulting bound controls weighted logarithmic sizes of these cohomology groups throughout a long interval. Further comparisons turn that arithmetic budget into the prime-sum estimate required by Landau.
At the inspected assembly point, the Galois-representation input is supplied by a theorem combining a geometric realization with theta lifting between unitary groups. It is not left as a free hypothesis of the final theorem.
The geometric branch supplies Frobenius–Hecke compatibility and crystalline/Hodge conditions above \(\ell\), under its local hypotheses. The construction then uses geometric seeds, congruences, trace limits, and descent to reach the Galois representations needed by the arithmetic window.
The guiding idea is to realize suitable automorphic information in cohomology, where Galois acts, and carry that information through the lifting and comparison steps. The exact local conditions remain part of the statement.
The geometric and deformation branches invoke named inputs for resolution of singularities in characteristic zero, toroidal compactifications of Siegel varieties, arithmetic Hilbert–Samuel, Moriwaki’s bigness theorem, Kawamata–Viehweg and Nadel vanishing, Grothendieck–Messing, and prismatic Dieudonné theory.
These inputs serve different purposes. Compactifications and resolution supply geometric models; positivity, vanishing, and Hilbert–Samuel estimates control sections and their growth; deformation and comparison results connect integral and cohomological structures.
On the covolume branch, the goal is a quantitative bound for lattices attached to the automorphic construction. Together with the cohomological estimate and theta matching, it constrains the prime-sum error.
Local divisibility measures arithmetic constraints on periods, with cohomology bounds controlling the losses. Theta matching puts the periods into one lattice, where correlations and covolumes constrain their sizes. Finite Fourier inversion then expresses the arithmetic prime contributions in terms of the corrected character sums.
The remaining normalized costs include terms behaving as \(x/m\) and \(m\), up to small powers of \(x\). A chain of about \(\sqrt{x}\) stages balances them. This yields the prime-sum bound on long intervals used by Landau to exclude zeros to the right of one half. The functional equation then gives the critical-line conclusion.
Follow the arithmetic-to-prime-sum calculationThis is a source-based dependency outline. A full Lean build and an audit of the final theorem’s assumptions have not been performed here, so “no additional axioms” and “complete formalizations of all the named theorems” are not conclusions of this guide. Source identity and review status
Work through the critical-line lemma in ordinary mathematical prose. The notes explain the character, gamma factors, functional equation, and both possible sides of the line, then match the reasoning to Lean.
Read the technical companionIt applies to every character, including the inverse one.
Gamma factors are finite and nonzero in the open strip.
Right-hand zeros contradict the input directly; left-hand zeros do so after reflection.
Start with the language, follow the deduction, or inspect the exact declarations. Each reading route leads back to the same mathematical statements.
Zeta, characters, analytic continuation, and the functional equation. The vocabulary used throughout the guide.
Read the foundationsThe nonvanishing theorem’s main dependencies, followed by the complete conditional deduction to the critical line.
Read the explained argumentSelected Lean declarations, line references, and a record of the archive used for this edition.
Explore the source indexAlpöge’s released Lean source states RH and a Dirichlet generalization. This website explains selected material from that development. The archive has been downloaded and extracted, and the prose has been checked against the selected source statements; the full Lean project has not been compiled or independently audited here.
The expanded technical notes trace the analytic estimate, the arithmetic cohomology window, Galois compatibility, theta matching and geometric bounds. They identify the deeper construction proofs that still need separate treatment.
Read the source and review record