Millennium Prize Problem · $1,000,000 Reward
The deepest unsolved problem in mathematics, connecting the prime numbers to the geometry of the complex plane.
↓ Scroll to explore ↓
Section I
In 1859, the German mathematician Bernhard Riemann submitted a single eight-page paper to the Berlin Academy — his only paper in number theory. In it, he introduced a function of a complex variable that would forever change mathematics. Embedded in the paper, almost as an aside, was what he called "very probable" — the statement now known as the Riemann Hypothesis.
That "temporary" deferral has lasted 165 years. Despite the collective efforts of virtually every major mathematician since, the hypothesis remains stubbornly unproven — and unprovable with current tools.
Riemann's paper concerned the distribution of prime numbers. Primes — integers divisible only by 1 and themselves — become increasingly sparse as you go further along the number line. Mathematicians had long sought a formula to predict their distribution. Gauss had conjectured the Prime Number Theorem: that primes thin out approximately as x / ln(x). Riemann's approach transformed this vague approximation into something breathtakingly precise.
The zeros of the Riemann zeta function are literally music encoded in the primes — each non-trivial zero is a frequency in an infinite symphony whose harmonics determine exactly where each prime number sits.
1740
Euler defines ζ(s) for real s > 1. Discovers connection to primes: the "Euler product" formula.
1859
Riemann analytically continues ζ(s) to the entire complex plane. States the hypothesis.
1896
Hadamard and de la Vallée Poussin independently prove the Prime Number Theorem — but not the full Hypothesis.
1900
Hilbert lists it as Problem 8 on his famous list of 23 unsolved problems.
1914
Hardy proves infinitely many zeros lie on the critical line — but not all of them.
1972
Montgomery discovers that zeros are spaced like eigenvalues of random Hermitian matrices — a physics connection.
2000
Clay Mathematics Institute designates it a Millennium Prize Problem, with a $1,000,000 reward.
2024
Over 10¹³ zeros verified computationally — all on the critical line. The hypothesis remains unproven.
Section II · Interactive
The zeta function begins simply. For real numbers s greater than 1, it is just an infinite sum:
Riemann's genius was to allow s to be a complex number — a point in the 2D complex plane with both a real part and an imaginary part. Through a technique called analytic continuation, the function extends to (almost) the entire complex plane, becoming a richly structured landscape with peaks, valleys, and zeros.
The visualization below shows |ζ(s)| as a heat map on the complex plane. Bright gold regions have large magnitude; dark regions have magnitude near zero. The critical line at Re(s) = ½ runs vertically through the center, marked in gold. Hover or drag to inspect values.
The Functional Equation
The zeta function satisfies a remarkable symmetry — a "reflection" about the line Re(s) = ½:
Section III · Interactive
When does the zeta function equal exactly zero? There are two kinds of zeros:
Trivial Zeros
The negative even integers: s = −2, −4, −6, …
These arise from the sine term in the functional equation and are well-understood.
Non-Trivial Zeros
The mysterious zeros in the critical strip: 0 < Re(s) < 1. The Hypothesis claims all of them have Re(s) = ½ exactly.
Below you can see the known non-trivial zeros plotted in the complex plane. Every single one that has ever been computed sits exactly on the critical line Re(s) = ½. The first few zeros are at approximately:
Why "Critical Strip"?
The functional equation forces zeros to be symmetric about Re(s) = ½. The Prime Number Theorem tells us there are no zeros on the lines Re(s) = 0 or Re(s) = 1. So all non-trivial zeros must lie in the open strip 0 < Re(s) < 1. The "critical line" Re(s) = ½ is its exact center.
Section IV
The Riemann Hypothesis is not merely an abstract curiosity. Its truth (or falsehood) has cascading consequences across mathematics, physics, cryptography, and computer science.
If the Riemann Hypothesis is true, we gain extraordinarily precise control over how primes are distributed. The error in the Prime Number Theorem becomes bounded by √x · ln(x) — a tight guarantee. Without it, the best known error bounds are weaker by exponential factors. Every prime gap estimate, prime-counting formula, and number-theoretic inequality sharpens dramatically.
Modern cryptography — from your HTTPS connection to Bitcoin — relies on the practical hardness of factoring large numbers. The distribution of primes determines how many large primes exist and how to find them. The truth of RH affects the rigor of security proofs for primality testing algorithms. Several primality tests have been proven correct assuming RH. If proven, these conditional guarantees become unconditional.
Most strikingly: if RH is false, and a zero exists far from the critical line, this could in principle hint at structure in primes exploitable for factoring attacks.
In 1972, Hugh Montgomery proved that the gaps between zeros of ζ(s) follow the same statistical distribution as energy levels of heavy atomic nuclei — specifically, random matrix theory from quantum mechanics. This GUE (Gaussian Unitary Ensemble) correlation was confirmed numerically by Andrew Odlyzko to stunning precision.
The zeros of ζ(s) may literally be the eigenvalues of some vast, undiscovered quantum system — an operator whose spectrum encodes the primes. This "Hilbert-Pólya conjecture" suggests the RH might have a physical proof before a purely mathematical one.
Over 1,000 published theorems begin with the words "Assuming the Riemann Hypothesis..." A proof would simultaneously validate all of them as unconditional truths. These span: analytic number theory, algebraic geometry, L-functions, automorphic forms, arithmetic of elliptic curves, and more. The RH is a keystone that, when removed, would cause the collapse of vast portions of conditional mathematics.
The Extended Riemann Hypothesis (ERH) is used to prove the correctness and efficiency of numerous algorithms in computational number theory. The Miller-Rabin primality test — used billions of times daily in cryptographic software — can be made deterministic and provably fast under ERH. Several graph theory results, including bounds on expander graphs, also depend on ERH.
If ERH fails, some of the theoretical foundations of fast number-theory algorithms would need to be rebuilt from scratch.
Section V
Every major school of modern mathematics has attacked the Riemann Hypothesis, and every one has encountered a fundamental obstacle. Here are the principal strategies and where they break down.
Riemann's own paper contains an explicit formula expressing π(x) exactly in terms of the zeros. The hope: show algebraically that any zero off the critical line would produce an inconsistency. This works beautifully for proving the PNT, but the constraints are too weak to force all zeros onto the line.
David Hilbert and George Pólya independently conjectured that the imaginary parts of the zeros are eigenvalues of some self-adjoint (Hermitian) operator — and since Hermitian operators always have real eigenvalues, this would prove the RH. The problem: nobody has been able to construct such an operator. Alain Connes made progress using noncommutative geometry, producing an operator whose spectrum is related to the zeros, but the connection remains incomplete.
The Weil Approach — Partial Victory
André Weil proved an analogue of the Riemann Hypothesis for function fields (curves over finite fields) in 1948. This "Riemann Hypothesis for curves" was the first complete proof of an RH-type statement. It suggested that Grothendieck's algebraic geometry might crack the number-field case, but the translation has defied every attempt for 76 years.
The Riemann zeta function is the simplest member of a vast family of L-functions. Proving the RH for all L-functions (the "Generalized Riemann Hypothesis") would follow from a grand unification program. The Langlands program represents the deepest current effort here, aiming to connect L-functions to automorphic forms. Taylor, Wiles, and others have made spectacular progress — but the GRH remains out of reach.
Using increasingly clever algorithms (Odlyzko-Schönhage), mathematicians have verified the RH for the first 10 trillion zeros. All lie exactly on the critical line. While this provides extraordinary numerical confidence, it cannot substitute for a proof — we cannot compute infinitely many zeros.
Crucially, no one has even proven that the fraction of zeros on the critical line equals 100%. Hardy proved infinitely many are on it; Selberg showed a positive fraction (at least about 40%) are; Conrey improved this to over 40.88%. But "all" remains unproven.
Section VI · Speculative
No one knows how to prove the Riemann Hypothesis. What follows is a synthesis of the most promising modern directions — not a claimed proof, but a roadmap of where expert consensus suggests a proof might eventually emerge.
Important Caveat
This section represents speculative mathematical directions, not an actual proof. Many brilliant mathematicians have believed they were close — and were not. The following reflects current research frontiers as understood by the mathematical community.
The most promising direction, advocated by Berry, Keating, and Connes, involves finding an explicit Hermitian operator H such that its eigenvalues are exactly the imaginary parts of the non-trivial zeros. Connes has proposed the candidate:
The challenge: identifying precisely which Hilbert space the operator acts on. Connes's approach in noncommutative geometry ("the adelic approach") constructs a space using the adèles — a number-theoretic construction encompassing all prime completions of ℚ simultaneously. Recent work by Connes and Consani has refined this.
Weil's explicit formula for L-functions can be rewritten as a positivity condition: a certain distribution must be non-negative. If true, this positivity implies the GRH. The so-called "Weil positivity" approach involves showing:
Langlands predicted a vast web of correspondences between automorphic forms and Galois representations. For function fields, these correspondences are proven (Drinfeld, Lafforgue, Ngô — all Fields Medalists). If the global Langlands correspondence for number fields can be established, the GRH would likely follow as a consequence of the "Ramanujan conjecture" for automorphic forms.
Current Frontier
As of 2024, the most active promising directions are: (1) Connes's noncommutative geometry approach, (2) the "explicit L-functions" program of Bhargava, Shankar, and collaborators, and (3) arithmetic-geometric approaches via perfectoid spaces (Scholze). No single breakthrough is imminent — but the tools available in 2024 are vastly more powerful than those Riemann had in 1859.
Based on the above, the most plausible proof structure would probably: