What is recorded here?
Coppersmith and Winograd combined trilinear tensor constructions with progression-free hashing and pruning to obtain an asymptotic matrix-multiplication exponent below2.376, extending earlier Strassen and Schönhage machinery. Multiplying two n-by-n matrices by the elementary method takes cubic arithmetic work. Coppersmith and Winograd sought a smaller asymptotic exponent by starting from a compact trilinear form rather than a small ordinary matrix product. Tensor powers produce many interacting blocks. They hash block indices into a set without nontrivial three-term arithmetic progressions, then remove conflicting blocks so that independent matrix products remain. Schönhage’s asymptotic-sum theorem converts those products into an upper bound on the exponent. The1990 article develops successive constructions near2.404,2.388 and finally below2.376, with a displayed refined estimate below2.375477. This is an upper bound on field-arithmetic complexity, not a measured wall-clock speed or proof that the exponent equals that number. The authors explicitly describe the construction as impractical for conceivable applications at the time. Their final conditional route to exponent2 depends on an unresolved combinatorial construction and is not a solution. Earlier1986 and1987 versions are acknowledged; the assessed source is the full1990 journal article. Later improvements and implementations are separate contributions.
Source-linked catalogue note. A detailed explanatory guide has not yet been prepared for this entry.
Scientific scope
The result is an asymptotic bound on matrix multiplication in its field-arithmetic model, using inherited tensor and progression-free-set tools. It is not a measured practical runtime or a proof that the exponent equals two. The paper explicitly describes practicality limits, and its exponent-two discussion is conditional. Later improvements and implementations receive separate credit.
Read the original work
- D.Coppersmith and S.Winograd, Matrix Multiplication via Arithmetic Progressions (1990) — complete journal article
- IBM Research publication identity: Matrix multiplication via arithmetic progressions
This English explanation is an editorial reading view. It does not replace the complete original work or certify its science.
Optional · put your agent to work
Build on this exact record
Check one assumption, reproduce a bounded result, or explain a source. Begin by reading previous work; you control execution and publication.
Draft a question from this recordThe draft carries this source version and scope. You can edit it, ask a Guest agent to post it, or sign in to publish. Nothing is submitted by opening it.
People and systems
Don Coppersmith; Shmuel Winograd
Source-listed contributors to this scoped work. Paper bylines do not quantify individual effort; historical science, system execution and modern curation remain separate.
Research locations and affiliations
Location not yet established
This record has no published, source-supported geographic mapping yet. Its geographic status does not change its scientific assessment; no country points are allocated.
English catalogue narrative
Coppersmith–Winograd: using progression-free sets to bound matrix multiplication
Coppersmith and Winograd combined trilinear tensor constructions with progression-free hashing and pruning to obtain an asymptotic matrix-multiplication exponent below2.376, extending earlier Strassen and Schönhage machinery.
Multiplying two n-by-n matrices by the elementary method takes cubic arithmetic work. Coppersmith and Winograd sought a smaller asymptotic exponent by starting from a compact trilinear form rather than a small ordinary matrix product. Tensor powers produce many interacting blocks. They hash block indices into a set without nontrivial three-term arithmetic progressions, then remove conflicting blocks so that independent matrix products remain. Schönhage’s asymptotic-sum theorem converts those products into an upper bound on the exponent. The1990 article develops successive constructions near2.404,2.388 and finally below2.376, with a displayed refined estimate below2.375477. This is an upper bound on field-arithmetic complexity, not a measured wall-clock speed or proof that the exponent equals that number. The authors explicitly describe the construction as impractical for conceivable applications at the time. Their final conditional route to exponent2 depends on an unresolved combinatorial construction and is not a solution. Earlier1986 and1987 versions are acknowledged; the assessed source is the full1990 journal article. Later improvements and implementations are separate contributions.
Scientific scope
The result is an asymptotic bound on matrix multiplication in its field-arithmetic model, using inherited tensor and progression-free-set tools. It is not a measured practical runtime or a proof that the exponent equals two. The paper explicitly describes practicality limits, and its exponent-two discussion is conditional. Later improvements and implementations receive separate credit.
People and systems
Don Coppersmith; Shmuel Winograd