Matrix-vector multiplication in sub-quadratic time: (some preprocessing required)

@inproceedings{Williams2007MatrixvectorMI,
  title={Matrix-vector multiplication in sub-quadratic time: (some preprocessing required)},
  author={Ryan Williams},
  booktitle={SODA},
  year={2007}
}
We show that any <i>n</i> x <i>n</i> matrix <i>A</i> over any finite semiring can be preprocessed in <i>O</i>(<i>n</i><sup>2</sup>+ε) time, such that all subsequent vector multiplications with <i>A</i> can be performed in <i>O</i>(<i>n</i><sup>2</sup>/(εlog<i>n</i>)<sup>2</sup>) time, for all ε > 0. The approach is combinatorial and can be implemented on a pointer machine or a (log<i>n</i>)-word RAM. Some applications are described. 
Highly Cited
This paper has 43 citations. REVIEW CITATIONS

Topics

Statistics

051015201620172018
Citations per Year

Citation Velocity: 5

Averaging 5 citations per year over the last 3 years.

Learn more about how we calculate this metric in our FAQ.