Matrix multiplication exponent equals 2

Determine whether the exponent ω of square matrix multiplication over the reals is equal to 2.

Background

Recent nearly-optimal linear programming algorithms have arithmetic complexity that depends on the matrix multiplication exponent ω. If ω were 2, these LP algorithms would achieve near-quadratic time (up to polylogarithmic factors).

The long-standing conjecture that ω = 2 remains open, and resolving it would immediately sharpen the best-known LP complexity bounds described in the paper.

References

Were the exponent of matrix multiplication ultimately shown to be 2 (as is conjectured, although progress in reducing it has slowed considerably), the complexity of the algorithms in would become \tilde{O}(n{2+1/6}) while for the algorithm of it would become O(n{2+1/18}).

— Optimization in Theory and Practice  (2510.15734 - Wright, 17 Oct 2025) in Section 4, Linear Programming, Subsection "Recent Developments in Complexity Analysis of Interior-Point Methods"

No larger lower bound is known, leading many to optimistically conjecture that ω=2.

— More Asymmetry Yields Faster Matrix Multiplication  (2404.16349 - Alman et al., 2024) in Section 1 (Introduction)

We further show that the above algorithm for $t = 2$ can be implemented in time $n{\omega + o(1)}$ for $\omega$ the constant of square matrix multiplication and succeeds when $\lambda > 0.3320$; under the folklore conjecture that $\omega = 2$, this runs in the nearly-linear time of the algorithm of Deshpande--Montanari while finding smaller cliques.

— Improved polynomial-time algorithms for detecting and recovering planted $Θ(\sqrt{n})$-cliques  (2609.24780 - Kunisky et al., 21 Sep 2026) in Abstract; Section 1, subsection “Main results”

More precisely, the asymptotic rank of the $n\times n$ matrix multiplication tensor is $n\omega$ for some (unknown) constant $\omega$, the exponent of matrix multiplication.

— Asymptotic completions of preordered semirings  (2609.31021 - Vrana, 25 Sep 2026) in Section 1, Introduction, paragraph discussing the semiring of tensors and matrix multiplication

At the time of writing, Dupont et al. hold the record for the smallest value of $\omega$ achieved by an algorithm, and it remains open whether the lower bound $\omega \geq 2$ is attainable.

— Fast and Stable Arbitrary Precision Approximate Matrix Multiplication  (2610.13905 - Dumitriu et al., 9 Oct 2026) in Section 1, Introduction