The Binet--Cauchy theorem for the hyperdeterminant of boundary format multi-dimensional matrices (Q1867300)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | The Binet--Cauchy theorem for the hyperdeterminant of boundary format multi-dimensional matrices |
scientific article |
Statements
The Binet--Cauchy theorem for the hyperdeterminant of boundary format multi-dimensional matrices (English)
0 references
2 April 2003
0 references
Let \(V_i\), \(i=0,\ldots, p\), be a complex vector space of dimension \(k_i+1\), \(k_0=\max_i k_i\), \(A\in V_0\otimes \ldots \otimes V_p\) be a multi-dimensional matrix. The matrix \(A\) is called degenerate if there exist \(v_i\in V_i^*\), \(v_i\neq 0\), \(i=0,\ldots, p\) such that \(A( v_0\otimes \ldots \otimes v_p)=0\). In the case of rectangular matrices ``nondegenerate'' means that \(A\) is of maximal rank. The matrix \(A\) is of boundary format if \(k_0=\sum_{i=1}^p k_i\). It is proved that a matrix of boundary format is nondegenerate if and only if its hyperdeterminant is non-zero. The authors also establish the following generalization of the usual Binet-Cauchy formula about the determinant of a matrix product. Let \(A\) and \(B\) be multi-dimensional matrices of boundary formats \((k_0, \ldots, k_p)\) and \((l_0,\ldots, l_q)\), respectively. It is proved that if \(k_p=l_0\) (so, the convolution \(A*B\) is defined) then \(\text{Det}(A*B)= \text{Det}(A)^{\alpha}\cdot \text{Det}(B)^{\beta}\), where Det denotes the hyperdeterminant, \(\alpha= \frac{l_0!}{l_1!\cdots l_q!}\), \(\beta= \frac{(k_0+1)!}{k_1!\cdots k_{p-1}! (k_p+1)!}\). As a corollary it is proved that \(A*B\) is nondegenerate if and only if both \(A\) and \(B\) are nondegenerate. It is remarked that the ``only if'' part holds without the assumption that matrices are of boundary formats. A counter-example to the Binet-Cauchy formula for the matrices of non-boundary format is provided.
0 references
Binet-Cauchy theorem
0 references
hyperdeterminants
0 references
multi-dimensional matrices
0 references
boundary format matrix
0 references
degenerate matrix
0 references