Millennium Prize Problems: The Greatest Unsolved Questions in Mathematics

In the year 2000, the Clay Mathematics Institute identified seven problems so fundamental — and so stubbornly resistant to proof — that it offered $1,000,000 for a correct solution to each. Twenty-five years later, only one has been solved. This page covers the structure, history, and mathematical substance of all seven problems, explains what makes them so difficult, and addresses the most persistent misconceptions about what "solving" one actually requires.

Definition and scope
Core mechanics or structure
Causal relationships or drivers
Classification boundaries
Tradeoffs and tensions
Common misconceptions
Checklist or steps
Reference table or matrix

Definition and scope

Seven problems. One million dollars each. A near-perfect record of resistance to human ingenuity for a quarter century. The Clay Mathematics Institute (CMI) announced the Millennium Prize Problems on May 24, 2000, in Paris — a deliberate echo of David Hilbert's famous 1900 lecture at the International Congress of Mathematicians, where Hilbert posed 23 problems that shaped the next hundred years of mathematical research.

The CMI's seven problems were selected by a scientific advisory board of leading mathematicians, not by poll or committee popularity contest, but by a principled criterion: each problem sits at a junction where resolution would unlock or reorganize a substantial region of mathematics or theoretical physics. They are not puzzles. They are structural questions about the architecture of mathematical reality — questions about how fluids behave, whether certain equations have solutions, and how hard it is for a computer to check its own answers.

The scope covers pure mathematics (Riemann Hypothesis, Hodge Conjecture, Birch and Swinnerton-Dyer Conjecture, Poincaré Conjecture), computational theory (P vs NP), mathematical physics (Yang–Mills Existence and Mass Gap, Navier–Stokes Existence and Smoothness), and arithmetic geometry (BSD Conjecture). That breadth is intentional. The problems were chosen to represent the full landscape of mathematics as of 2000, not any single school or tradition. For a broader map of where these problems sit within the discipline, the key dimensions and scopes of mathematics provides useful orientation.

Core mechanics or structure

Each problem has a precise official formulation, published by the CMI and written by a specialist mathematician. The Poincaré Conjecture, for example, was formally stated by John Milnor; the P vs NP problem statement was written by Stephen Cook. These aren't informal sketches — they are rigorous logical propositions that admit of exactly two outcomes: proved true, or proved false (with a valid counterexample).

The seven problems, briefly characterized:

P vs NP — Asks whether every problem whose solution can be verified quickly can also be solved quickly. This is the central question of computational complexity theory, and it governs the theoretical security of cryptographic systems used across the internet.
Hodge Conjecture — In algebraic geometry, asks whether certain topological cycles on a complex algebraic variety can be represented by algebraic cycles. It bridges differential geometry and algebraic topology in ways that remain partially uncharted.
Riemann Hypothesis — Proposes that all non-trivial zeros of the Riemann zeta function have real part equal to 1/2. First stated by Bernhard Riemann in 1859, it is tied to the distribution of prime numbers. As of 2024, more than 10 trillion zeros have been computed and all lie on the critical line — but computation is not proof (Odlyzko, AT&T Labs, published numerical verifications).
Yang–Mills Existence and Mass Gap — Asks whether quantum Yang–Mills theory (the mathematical framework underlying the Standard Model of particle physics) can be rigorously constructed in four-dimensional space, and whether it predicts a positive mass gap for particles.
Navier–Stokes Existence and Smoothness — Asks whether smooth solutions to the Navier–Stokes equations (which govern fluid motion) always exist in three dimensions, or whether they can develop singularities (discontinuities) in finite time.
Birch and Swinnerton-Dyer Conjecture — Concerns elliptic curves: it proposes a relationship between the algebraic rank of an elliptic curve's rational points and the behavior of its associated L-function near a particular point. Partial results were proved by Andrew Wiles and others but the full conjecture remains open.
Poincaré Conjecture — The only solved problem. It states that any simply connected, closed 3-manifold is homeomorphic to a 3-sphere. Grigori Perelman published a proof between 2002 and 2003 using Richard Hamilton's Ricci flow technique, which was verified by three independent teams by 2006. Perelman declined the $1,000,000 prize.

Causal relationships or drivers

Why are these seven problems unsolved when mathematicians have tackled them for decades or, in the Riemann Hypothesis's case, over 160 years? The answer is structural, not accidental.

Each problem requires either a fundamentally new mathematical tool or a resolution of a deep tension between two existing frameworks. The Navier–Stokes problem, for instance, sits at the intersection of partial differential equations and turbulence theory — two areas where existing analytical techniques run out before the question does. The P vs NP problem may require separating two complexity classes in a way that evades known "relativizing" proof techniques, a barrier identified by Baker, Gill, and Solovay as early as 1975 (Baker, Gill, Solovay, SIAM Journal on Computing, 1975).

The Yang–Mills problem is unusual in that it asks mathematicians to formalize physics that already "works" experimentally — the Standard Model makes predictions verified to extraordinary precision, yet its mathematical foundations are not rigorous. The gap between physical intuition and mathematical proof is the problem itself.

The broader mathematical research landscape, including where these problems sit relative to other open questions, is surveyed in mathematics research fields.

Classification boundaries

The seven problems divide meaningfully across four domains:

Pure mathematics (2 problems): Riemann Hypothesis and Hodge Conjecture deal with intrinsic mathematical structure, independent of physical application.

Arithmetic geometry (1 problem): Birch and Swinnerton-Dyer operates at the intersection of number theory and algebraic geometry — the same neighborhood where Andrew Wiles proved Fermat's Last Theorem in 1995.

Mathematical physics (2 problems): Yang–Mills and Navier–Stokes require rigorous mathematical treatment of physical phenomena. Both are existence problems: not "what is the answer" but "does an answer of a certain kind exist at all."

Theoretical computer science (1 problem): P vs NP is the only problem squarely in computational theory, though its resolution would have immediate implications for cryptography, optimization, and artificial intelligence.

Topology (1 problem — solved): Poincaré Conjecture lived in geometric topology and was resolved through differential geometry, specifically Ricci flow with surgery.

That one problem spans two major domains (Riemann, linking analysis and number theory) illustrates why classification is useful but not airtight.

Tradeoffs and tensions

A correctly solved Millennium Problem must be submitted to a CMI-approved journal, pass peer review, and then withstand a two-year waiting period before the prize is formally awarded. This process is deliberately slow — and Perelman's case showed why. His proof was submitted informally on arXiv (not a traditional journal) between 2002 and 2003 and required multiple independent verification teams working over several years before consensus emerged.

That same case exposed a genuine tension between the culture of mathematical publishing and the speed of modern dissemination. A proof posted on arXiv is publicly available but not formally peer-reviewed; the CMI's rules were designed for a pre-arXiv era. This tension has not been formally resolved in CMI's published guidelines.

A separate tension exists around P vs NP specifically: the majority of theoretical computer scientists believe P ≠ NP, but "majority belief" has no weight in mathematics. The problem could in principle be independent of the standard axioms of mathematics (ZFC), which would mean neither a proof nor a disproof exists within normal mathematical reasoning — an outcome that would itself be a landmark result.

Common misconceptions

"Solving one of these problems means solving a hard calculation." Every Millennium Problem is a proof problem, not a computation problem. The Riemann Hypothesis cannot be resolved by computing more zeros of the zeta function, no matter how many. Only a general argument covering all cases constitutes a solution.

"The problems were chosen because they are the hardest problems in mathematics." They were chosen because they are important junction problems — questions whose resolution reorganizes other mathematics. Difficulty was a factor; centrality was the criterion.

"Perelman solved the Poincaré Conjecture alone." Perelman's work was decisive, but it was built explicitly on Richard Hamilton's Ricci flow program, developed through the 1980s and 1990s. Perelman himself credited Hamilton. Mathematical proof is rarely a solo act at this level.

"A disproof counts as a solution." It does, if rigorous. A valid counterexample to the Birch and Swinnerton-Dyer Conjecture would earn the $1,000,000 just as a proof would. The prize is for resolution, not confirmation.

"These problems are connected." They are largely independent. A proof of the Riemann Hypothesis would not solve P vs NP, though both involve deep questions about structure and decidability. The problems were grouped by importance, not by mathematical kinship.

The mathematical proof techniques page covers the underlying logical machinery relevant to how any of these would ultimately be resolved.

Checklist or steps

How a Millennium Prize solution moves from claim to award:

A mathematician (or team) produces a complete proof or disproof and posts or submits it publicly.
The proof is submitted to a mathematics journal approved by the CMI for peer review.
The journal's peer review process evaluates the argument — a process that can take one to three years for a result of this magnitude.
Following publication, a two-year waiting period begins during which the broader mathematical community can scrutinize the result.
The CMI's Scientific Advisory Board reviews community response and the published work.
If the board concludes the proof is correct and complete, the $1,000,000 prize is awarded.
The solver may accept or decline the prize (Perelman declined; CMI confirmed the award was valid regardless).

Reference table or matrix

Problem	Domain	Status (as of 2024)	Key Difficulty
P vs NP	Computational complexity	Unsolved	Requires new proof techniques beyond relativizing methods
Hodge Conjecture	Algebraic geometry	Unsolved	Bridging topological and algebraic cycles
Riemann Hypothesis	Analytic number theory	Unsolved	General proof over all non-trivial zeros required
Yang–Mills Existence and Mass Gap	Mathematical physics	Unsolved	Rigorous 4D quantum field theory construction
Navier–Stokes Existence and Smoothness	PDEs / fluid dynamics	Unsolved	Existence of smooth global solutions in 3D
Birch and Swinnerton-Dyer Conjecture	Arithmetic geometry	Unsolved	L-function behavior tied to rational point structure
Poincaré Conjecture	Geometric topology	Solved (2003)	Proved by Perelman via Ricci flow with surgery

For those exploring how these problems connect to competitive mathematics and academic pathways, the mathematics competitions in the US page covers the landscape from high school olympiads to research-level engagement. The broader index of mathematical topics is available at the Mathematics Authority home.