<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<item>
  <id>02187725</id>
  <dt>j</dt>
  <an>02187725</an>
  <augroup>
    <au>Pan, V.Y.</au>
  </augroup>
  <ti>On theoretical and practical acceleration of randomized computation of the determinant of an integer matrix.</ti>
  <so>Zap. Nauchn. Semin. POMI 316, 163-187 (2004); translation in J. Math. Sci., New York 134, No. 5, 2411-2424 (2006).</so>
  <py>2004</py>
  <pu>Nauka, Sankt-Peterburg</pu>
  <lagroup>
    <la>EN</la>
  </lagroup>
  <ccgroup>
  </ccgroup>
  <utgroup>
    <ut>integer matrix</ut>
    <ut>determinant</ut>
    <ut>Las Vegas algorithm</ut>
    <ut>block Hankel matrix</ut>
    <ut>generating matrix polynomial</ut>
    <ut>divide-and-conquer techniques</ut>
  </utgroup>
  <cigroup>
    <ci>Zbl 1012.65505</ci>
  </cigroup>
  <ligroup>
  </ligroup>
  <abgroup>
    <ab>A new Las Vegas algorithm for computing the determinant of an $n\times n$ integer matrix $A$ is proposed. The development is based on an algorithm presented [{\it E. Kaltofen} and {\it G. Villard}, ``On the complexity of computing determinants". (Extended abstract). (English) Shirayanagi, Kiyoshi (ed.) et al., Computer mathematics. Proceedings of the 5th Asian symposium (ASCM 2001), Matsuyama, Japan, September 26--28, (2001). Singapore: World Scientific. Lect. Notes Ser. Comput. 9, 13--27 (2001; Zbl 1012.65505)] but proposes a more efficient approach for computing the coefficients of the minimum generating matrix polynomial used within this algorithm. This reduces the complexity from $O^*(n^{10/3} L)$ to $O^*(n^{16/5} L)$ bit operations, where $L=\log(n\max \vert a_{ij}\vert )$. Roughly speaking, this improvement is achieved by using divide-and-conquer techniques for solving linear systems having block Hankel structure.</ab>
    <rv>Daniel Kressner (Zagreb)</rv>
  </abgroup>
</item>