arXiv++ Combinatorics

The Cullis' determinant is a generalization of the ordinary determinant for rectangular matrices. It is defined as the alternating sum of maximal minors of given matrix. In this paper we express the Cullis' determinant of a matrix $X$ as the Pfaffian of the matrix obtained from $X$ by matrix multiplication and transposition. Relying on this result, we present an efficient polynomial-time algorithm for calculating the Cullis' determinant of given matrix.

Let $\mathbb F$ be a field and $P \in \mathbb F [x_1,\ldots, x_n]$ be a homogeneous polynomial such that $|\mathbb F| > °(P)$ and $φ, ψ\colon \mathbb F^n \to \mathbb F^n$ be two maps such that $P(\mathbf{x} + λ\mathbf{y}) = P(φ(\mathbf{x}) + λψ(\mathbf{y}))$ for all $λ\in \mathbb F$ and $\mathbf{x}, \mathbf{y} \in \mathbb F^n.$ We provide the characterization of all such $φ$ and $ψ$ for all polynomials in the case if $\mathrm{char}(\mathbb F) = 0$ and for all polynomials satisfying certain condition in the case if $\mathrm{char}(\mathbb F) > 0$. This characterization generalizes the existing results regarding the linear maps on matrices preserving the determinant, the immanant and other homogeneous polynomial functions of matrix entries. To obtain the main result of this paper, we introduce the vector space $\mathcal L_{P} \subseteq {\mathbb F^n}^*$ spanned by the range of the gradient field of $P \in \mathbb F[x_1,\ldots, x_n]$. Being a linear invariant associated with $P,$ this space has several remarkable properties and may also be used for studying the linear maps preserving $P$. In addition, we demonstrate how the main result could be applied to the particular polynomial matrix invariants. Namely, we provide an explicit description of corresponding pairs of nonlinear maps $φ, ψ$ for the case where $P$ is equal to the Cullis' determinant of $n\times k$ rectangular matrix (with the assumption that $n \ge k + 2$ and $k \ge 3$).

Let $V$ be a vector space of rectangular $n\times k$ matrices annihilating the Cullis' determinant. We show that $\dim(V) \le (n-1)k$, extending Dieudonn{é}'s result on the dimension of vector spaces of square matrices annihilating the ordinary determinant. Furthermore, for certain values of $n$ and $k$, we explicitly describe such vector spaces of maximal dimension. Namely, we establish that if $k$ is odd, $n \ge k + 2$ and $\dim(V) = (n-1)k$, then $V$ is equal to the space of all $n\times k$ matrices $X$ such that alternating row sum of $X$ is equal to zero. Our proofs rely on the following observations from the matroid theory that have an independent interest. First, we provide a notion of matroid corresponding to a given linear variety. Second, we prove that if the linear variety is transformed by projections and restrictions, then the behaviour of the corresponding matroid is expressed in the terms of matroid contraction and restriction. Third, we establish that if $M$ is a matroid, $I^*$ its coindependent set $M|S$ and its restriction on a set $S$, then the union of $I^*\setminus S$ with every cobase of $M|S$ is coindependent set of $M$.

This paper is the first in the series of papers devoted to the explicit description of linear maps preserving the Cullis' determinant of rectangular matrices with entries belonging to an arbitrary ground field which is large enough. The Cullis' determinant is defined for every matrix of size $n\times k$, where $n \ge k \ge 1$ and is equal to the ordinary determinant if $n = k$. In this paper we solve the linear preserver problem for the Cullis' determinant for $k \ge 4, n \ge k + 2$ and $n + k$ is even. It appears that in this case all linear maps preserving the Cullis' determinant are non-singular and could be represented by two-sided matrix multiplication. Note that the cases where $n = k$ or $n = k + 1$ admit slightly different description allowing (sub)matrix transposition and were completely studied before: the case where $n = k$ is a classical linear preserver problem for the ordinary determinant and was solved by Frobenius; the complete characterisation for the case where $n = k + 1$ was obtained in the previous paper by the authors.

This paper is the second in the series of papers devoted to the explicit description of linear maps preserving the Cullis' determinant of rectangular matrices with entries belonging to an arbitrary ground field which is large enough. In this part we solve the linear preserver problem for the Cullis' determinant defined on the spaces of matrices of size $n\times k$ with $k \ge 4,\; n \ge k + 2$ and $n + k$ is odd. In comparison with the case when $n + k$ is even, in this case linear maps preserving the Cullis' determinant could be singular and are represented as a sum of two linear maps: first is two-sided matrix multiplication and second is any linear map whose image consists of matrices, all rows of which are equal.