Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2601.05272

A Rank 23 Algorithm for Multiplying 3 x 3 Matrices with an Arithmetic Complexity of 59

Erik Mårtensson·Paul Stankovski Wagner·Joshua Stapleton
View original source ↗Read source document ↗Export metadata ↓

Abstract

In 1969 Strassen showed surprisingly that it is possible to multiply two 2 x 2 matrices using seven multiplications and 18 additions, instead of the naive eight multiplications and four additions. The number of additions was later reduced to 15. Karstadt and Schwartz further reduced the number of additions to 12 using a change-of-basis method. Both the number of multiplications and the number of additions have been shown to be optimal for the 2 x 2 case. For multiplying 3 x 3 matrices, the lowest number of multiplications found so far is 23. Using 23 multiplications, Schwart et al. showed how to reduce the number of additions to 61 using a change-of-basis method. Mårtensson and Stankovski Wagner showed how to achieve 62 additions, without changing basis. Using the optimization method by Mårtensson and Stankovski Wagner, Stapleton found an algorithm requiring only 60 additions. In this work we continue to combine the methods of Mårtensson, Stankovski Wagner and Stapleton, finding an algorithm requiring only 59 additions, still without a basis change. Technical details on the method and tools used for finding this scheme, and a discussion on the impact of this discovery, will come in an upcoming publication.

Explore related subjects

cs.SCcs.DM

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Erik Mårtensson, Paul Stankovski Wagner, Joshua Stapleton. 2025-12-18. A Rank 23 Algorithm for Multiplying 3 x 3 Matrices with an Arithmetic Complexity of 59. https://arxiv.org/abs/2601.05272

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Complete Reductions and Idempotent Representations for $RΠΣ^*$-towers

$RΠΣ^*$-extensions form a rich class of difference rings that provide a unified algebraic framework for modeling indefinite nested sums, transcendental products, and nested products over roots of unity structures that frequently appear in combinatorics, number theory, and particle physics. For a large subclass of these extensions whose ring of constants is a field, we introduce a complete reduction approach to resolve the telescoping problem without solving any difference equations. More precisely, we explicitly construct a complement to the subspace of differences over the constant field and develop an algorithm that decomposes any element of the extension into the sum of a difference and a component lying in this complement. Consequently, summability holds if and only if this complementary component is zero. This structural approach yields significant speed-ups for parameterized telescoping and, notably, creative telescoping for deriving linear recurrences of definite sums. Finally, we compute an explicit idempotent representation that extends existing telescoping algorithms and our complete reduction framework to the general class of $RΠΣ^*$-extensions, opening up previously untreatable classes of sums and products.

cs.SC↗
arXiv2026

Refined complexity bounds for rational reconstruction and XGCD through Padé approximants and Cauchy interpolants

When computing with univariate polynomials, two fundamental and related problems are the XGCD and rational reconstruction, classically solved in quasi-linear complexity using the half-gcd algorithm. These problems have various applications in algebraic computations and bear strong connections to linearly recurrent sequences, structured matrices, and continued fractions. This article first gives a collection of algorithmic reductions, showing that rational reconstruction and XGCD can be solved via the computation of bases of relations modulo a freely-chosen polynomial $M(x)$. In particular, one recovers the folklore idea that bases of Padé approximants (i.e., $M(x) = x^d$) can be used to perform quasi-linear rational reconstruction or XGCD, extending to fast algorithms the well-known link between the Berlekamp-Massey algorithm and the extended Euclidean algorithm. One highlight of these reductions is that, instead of approximants, one may rely on Cauchy interpolants (i.e., $M(x)$ vanishes at chosen points). In a second part, this article describes divide-and-conquer algorithms for approximants and interpolants along with complexity analyses showing an explicit leading constant in front of the dominant term. For interpolants, the best leading constant is obtained through a variant that stores polynomials represented by evaluations, and exploits fast extrapolation in order to avoid repeated conversions to the monomial basis; this requires special points, in geometric or arithmetic progression, or FFT points when the base field allows them. Combining the analyses with the reductions leads to the best complexity bounds we are aware of for rational reconstruction and XGCD. Perhaps surprisingly, even Padé approximants or Berlekamp-Massey-like computations, which intrinsically involve $M(x) = x^d$, are accelerated by reducing them to Cauchy interpolation at well-chosen points.

cs.SC↗
arXiv2026

Parallel Integration over Simple Radical Extensions II: Mixed Towers

In Part I we extended the structure theorems underlying the Risch--Norman (parallel Risch) method to a simple radical extension $L=K(y)$, $y^m=q$, of a differential field $K=F(t_1,\dots,t_n)$ closed under the derivation. Here we remove the closure hypothesis: the radical may occupy any position in the tower, so that the derivatives of the generators above it involve $y$ --- the setting of Bronstein's algorithm for mixed elementary functions. The working ring is $\cA=\cO[t_{j+1},\dots,t_n]$, the integral closure of $F[t_1,\dots,t_n]$ in $L$: a Krull domain, free over the polynomial ring on Trager's basis, so that all factorisation remains in a unique factorisation domain. The denominator of the derivation is no longer an element but a divisor $\fd_D$ on $\cA$, and the valuation lemma takes the unified form $v_P(Dg)=v_P(g)-(1+v_P(\fd_D))$ at normal height-one primes, subsuming the shifts $\{1,e_P\}$ of Part I; the proof localises and requires no cancellation analysis. Stability of the class group and the unit group, $\Cl(\cA)\cong\Cl(\cO)$ and $\cA^*=\cO^*$, splits the admissible logands into $S$-units of $\cO$ --- computed by the machinery of Part I --- and irreducible polynomials moving in the upper variables, whose residues must be constants. We prove degree bounds in the top variable and describe the resulting algorithm, which --- unlike the classical parallel method, whose failure proves nothing --- returns certificates of non-elementarity in two situations: a residue outside the constant field, and, when every bound in force is proved, a residue-free remainder that the linear system shows to be non-exact.

cs.SC↗