Matrix A is said to be additively D-stable if A − D remains Hurwitz for all nonnegative diagonal matrices D. In reaction-diffusion models, additive D-stability of the matrix describing the reaction dynamics guarantees stability of the homogeneous steady-state, thus ruling out the possibility of diffusion-driven instabilities. We present a new criterion for additive D-stability using the concept of compound matrices. We first give conditions under which the second additive compound matrix has nonnegative offdiagonal entries. We then use this Metzler property of the compound matrix to prove additive D-stability with the help of an additional determinant condition. This result is then applied to investigate stability of cyclic reaction networks in the presence of diffusion.
I. IThe concept of diagonal stability and its variants are commonly used in the study of dynamic models in economics, ecology, and control theory, as surveyed by Kaszkurewicz and Bhaya [1]. A square matrix A is said to be diagonally stable if there exists a diagonal matrix S > 0 such that A T S + S A < 0. Two related properties that are less restrictive than diagonal stability but possibly more restrictive than the Hurwitz property of A are multiplicative D-stability, which means that DA is Hurwitz for all diagonal matrices D > 0, and additive D-stability, which means that A − D is Hurwitz for all diagonal D ≥ 0.Additive D-stability is particularly useful for the study of reaction-diffusion systems where the matrix A represents the linearization of the reaction dynamics at a steady-state. Denoting by D the diagonal matrix of diffusion coefficients for each species, Casten and Holland [2] showed that the stability of the reactiondiffusion PDE is determined by the simultaneous stability of the family of matrices A − λ k D, where λ k ≥ 0, k = 1, 2, 3, · · · are the eigenvalues for Laplace's equation with Neumann boundary condition on the given spatial domain. Additive D-stability thus guarantees stability of the spatially homogeneous steady-state and rules out the possibility of diffusion-driven instabilities which constitute the basis of Turing's mechanism for pattern formation [3], [4]. Wang and Li [5] further studied the connection between additive D-stability and reactiondiffusion models, and gave several algebraic sufficient conditions that guarantee either stability or instability in the presence of diffusion.In this paper, we present a new sufficient condition for additive D-stability using the concept of additive compound matrices [6]-[8] defined below. The key property employed in this condition is a special sign structure of matrix A which ensures that its second additive compound matrix A [2] is Metzler; that is, the off-diagonal entries of A [2] are nonnegative. Among the systems that exhibit this sign structure are cyclic reaction networks with negative feedback, where the end product of a sequence of reactions inhibits the first reaction upstream. For this class of networks, [9] established stability of the homogeneous stead...