Multipolynomial resultant algorithms (Q1260754)

The paper describes informally algorithms for computing determinants of matrices with polynomial entries from \(Z(x)=Z(x_ 1,\ldots,x_ n)\). In contrast to methods applied in most computer algebra systems the determinants are computed indirectly by a homomorphism/reconstruction method: map the matrix from \(Z(x)\) to \(Z\) by substituting sets of numbers for \((x_ 1,\dots,x_ n)\), compute for each a numerical determinant in \(Z\) and construct the determinant in \(Z(x)\) by interpolation. Several aspects for the interpolation are discussed: A Vandermonde type equation system (using sets \((p_ 1^ k,\dots,p_ n^ k)\) with primes \(p_ i)\), sparse interpolation, probabilistic interpolation and modular evaluation/Chinese remainder reconstruction. The use of the determinant computation for resultants of multivariate polynomial sets is presented in examples: Implication of rational parametric surfaces, real solutions of polynomial equation systems. The theory of multipolynomial resultants and algorithms for their construction are not covered in this article.