Conditionally positive definite dot product kernels

This paper deals with conditionally positive definite kernels on Euclidean spaces. The focus here is on dot product kernels, that is, those depending on the inner product of the variables. Among the results, we include some properties relating conditional positive definiteness and standard convolution in the line and also results related to the characterization of the conditionally positive definite dot product kernels with respect to finite-dimensional polynomial spaces. We also introduce and characterize two large classes of strictly conditionally positive definite dot product kernels.     

Weinstein positive definite functions

Positivity - In this paper, we introduce the notion of the Weinstein positive definite functions and we state a version of Bochner’s theorem. Furthermore, we study the strictly Weinstein...     

On positive definite measures

In this paper we study the Fourier transform of unbounded measures on a locally compact groupG. After a short introductory section containing background material, especially results established byL. Argabright andJ. Gil De Lamadrid we turn to the main subjects of the paper: first we characterize \(\Re \left( G \right), \mathfrak{J}\left( G \right)\) andB(G) cones in \(\mathfrak{W}\left( G \right)\) . After that we establish the subspace \(\mathfrak{W}_\Delta \left( G \right)\) of \(\mathfrak{W}\left( G \right)\) which contains \(\mathfrak{W}_p \left( G \right)\) , the linear span of all positive definite measures.     

The minimal positive integer represented by a positive definite quadratic form

In this paper we shall give an upper bound on the size of the gap between the constant term and the next nonzero Fourier coefficient of a holomorphic modular form of given weight for the group G₀(2) \Gamma_{0}(2) . We derive an upper bound for the minimal positive integer represented by an even positive definite quadratic form of level two. In our paper we prove two conjectures given in [1]. In particular, we can prove the following result: let Q \mathcal{Q} be an even positive definite quadratic form of level two in v v variables, with v o 4(mod 8) v \equiv 4(\textrm{mod}\, 8) , then Q \mathcal{Q} represents a positive integer 2n ￡ 3+v/4 2n \leq 3+v/4 .     

On Voronoi reduction of positive definite quadratic forms

There are two methods of reduction of positive definite quadratic forms due to Voronoi. One of these methods is based on the perfect forms and the other on the type of Voronoi polyhedra associated with the form. It was conjectured by Voronoi that these two methods are strongly connected.     

Elliptic curves and positive definite ternary forms

For two given ternary quadratic formsf(x, y, z) andg( x, y, z), letr(f, n) andr(g, n) be the numbers of representations of n represented byf( x, y, z) and g( x, y, z) respectively. In this paper we study the following problem: when will we haver(f, n) =r(g, n) orr( f, n)≠r(g, n). Our method is to use elliptic curves and the corresponding new forms.     

On positive definite hermitian forms

Kneser's method of constructing adjacent lattices will be used to determine class numbers of unimodular positive definite hermitian lattices of rank 2 and 3 over rings of integers in some imaginary quadratic fields. The same method will also be applied in order to construct indecomposable unimodular positive definite hermitian lattices of rank 2 and 3 over almost all orders in imaginary quadratic fields, even non-maximal ones. All exceptional cases will be determined explicitly. In part supported by NSF grant DMS 8805262. I would like to thank Prof. Martin Kneser for the support and the many helpful hints I received during the work on my Diplom thesis on which this paper is based, and also Prof. T.-Y. Lam for providing me with the grant without which I would not have been able to devote part of my time to writing this paper.     

Strictly Hermitian positive definite functions

LetH be any complex inner product space with inner product <·,·>. We say thatf: ℂ→ℂ is Hermitian positive definite onH if the matrix

Factorisation of positive definite operators

In this paper we prove Reade’s result for the positive definite C ¹ kernels by using the factorisation method used by Kühn. Received: 10 April 2007, Revised: 3 December 2007     

Extension of positive definite functions

Let Ω⊂Rⁿ$\Omega \subset {\mathbb{R}}^n$ be an open, connected subset of Rⁿ${\mathbb{R}}^n$, and let F:Ω−Ω→C$F:\Omega -\Omega \to \mathbb{C}$, where Ω−Ω={x−y:x,y∈Ω}$\Omega -\Omega =\{x-y:x,y\in \Omega \}$, be a continuous positive definite function. We give necessary and sufficient conditions for F   to have an extension to a continuous positive definite function defined on the entire Euclidean space Rⁿ${\mathbb{R}}^n$. The conditions are formulated in terms of existence of a unitary representations of Rⁿ${\mathbb{R}}^n$ whose generators extend a certain system of unbounded Hermitian operators defined on a Hilbert space associated to F. Different positive definite extensions correspond to different unitary representations.     

Strictly positive definite functions on the unit circle

We study strictly positive definite functions on the unit circle in the Euclidean space of dimension two. We develop several conditions pertaining to the determination of such functions. The major result is obtained by considering the set of real numbers as a vector space over the field of rational numbers and then applying the Kronecker approximation theorem and Weyl's criterion on equidistributions.

     

Unsymmetric positive definite linear systems

Is it necessary to pivot when solving an unsymmetric positive definite linear system Ax=b? Define $\text{T=}\text{(A+A}{}^{\mathrm{T}}\text{)}\text{2}$ and $\text{S=}\text{(A±A}{}^{\mathrm{T}}\text{)}\text{2}$. It is shown that pivoting is unnecessary if the quantity $\text{6ST}{}^{\mathrm{?1}}\text{S6}{}_2\text{6A6}{}_2$ is suitably small with respect to the working machine precision.     

On eigenvalues of differentiable positive definite kernels

If a positive definite kernelK(x, y) has thepth order partial derivative ( ^p /y ^p )K(x,y) continuous on the square [0,1]², we show that the eigenvalues of the integral operator generated byK(x, y) are asymptoticallyo(1/n ^p+1 ). We also obtain the anticipated asymptotic estimate when ( ^p /y ^p )K(x,y) satisfies further a Lipschitz condition iny of order 0<1. These results, which extend some classical estimates of I. Fredholm and H. Weyl under the additional positive definiteness assumption, are based on two interesting inequalities of K. Fan.     

矩阵正定性的进一步推广

矩阵的正定性在很多领域中都有广泛的应用,其定义得到了一系列的推广.进一步推广了矩阵的正定性,给出了更广义的正定矩阵的定义,并得到了它的若干性质.     

The positive definite completion problem revisited

In [R. Grone, C.R. Johnson, E. Sa, H. Wolkowicz, Positive definite completions of partial Hermitian matrices, Linear Algebra Appl. 58 (1984) 109-124] the positive definite (semi-) completion problem in which the underlying graph is chordal was solved. For the positive definite case, the process was constructive and the completion was obtained by completing the partial matrix an entry at a time. For the positive semidefinite case, they obtained completions of a particular sequence of partial positive definite matrices with the same underlying graph and noted that there is a convergent subsequence of these completions that converges to the desired completion. Here, in the chordal case, we provide a constructive solution, based entirely on matrix/graph theoretic methods, to the positive (semi-)definite completion problem. Our solution associates a specific tree (called the “clique tree” [C.R. Johnson, M. Lundquist, Matrices with chordal inverse zero-patterns, Linear and Multilinear Algebra 36 (1993) 1-17]) with the (chordal) graph of the given partial positive (semi-)definite matrix. This tree structure allows us to complete the matrix a “block at a time” as opposed to an “entry at a time” (as in Grone et al. (1984) for the positive definite case). In Grone et al. (1984), using complex analytic techniques, the completion for the positive definite case was shown to be the unique determinant maximizing completion and was shown to be the unique completion that has zeros in its inverse in the positions corresponding to the unspecified entries of the partial matrix. Here, we show the same using only matrix/graph theoretic tools.     

Operator-valued positive definite kernels on tubes

In this paper we generalize the concept of an infinite positive measure on a -algebra to a vector valued setting, where we consider measures with values in the compactification of a convex coneC which can be described as the set of monoid homomorphisms of the dual coneC ^* into [0, ]. Applying these concepts to measures on the dual of a vector space leads to generalizations of Bochner's Theorem to operator valued positive definite functions on locally compact abelian groups and likewise to generalizations of Nussbaum's Theorem on positive definite functions on cones. In the latter case we use the Laplace transform to realize the corresponding Hilbert spaces by holomorphic functions on tube domains.     

Compactly supported positive definite radial functions

We provide criteria for positive definiteness of radial functions with compact support. Based on these criteria we will produce a series of positive definite and compactly supported radial functions, which will be very useful in applications. The simplest ones arecut-off polynomials, which consist of a single polynomial piece on [0, 1] and vanish on [1, ∞). More precisely, for any given dimensionn and prescribedC ^k smoothness, there is a function inC ^k (? ⁿ ), which is a positive definite radial function with compact support and is a cut-off polynomial as a function of Euclidean distance. Another example is derived from odd-degreeB-splines.     

On extending backwards positive definite sequences

The two-sided Hamburger moment problem¹, also called the strong one [4], has been extensively studied in recent years in connection with rational approximation. We propose to consider the question of when a sequence, say {a _n } _n=0 can be extended backwards so that the resulting sequence {a _n } _n=–N has an integral representation of the Hamburger type. This was settled (without any proof) under different circumstances in [6]. Here we wish to discuss this completely, as well as the possibility of extending {a _n } _n=0 to {a _n } _n– .