General Full Implicit Strong Taylor Approximations for Stiff Stochastic Differential Equations

In this paper, we present the backward stochastic Taylor expansions for a Ito process, including backward Ito-Taylor expansions and backward Stratonovich-Taylor expansions. We construct the general full implicit strong Taylor approximations (including Ito-Taylor and Stratonovich-Taylor schemes) with implicitness in both the deterministic and the stochastic terms for the stiff stochastic differential equations (SSDE) by employing truncations of backward stochastic Taylor expansions. We demonstrate that these schemes will converge strongly with corresponding order $1,2,3,\ldots$ Mean-square stability has been investigated for full implicit strong Stratonovich-Taylor scheme with order $2$, and it has larger mean-square stability region than the explicit and the semi-implicit strong Stratonovich-Taylor schemes with order $2$. We can improve the stability of simulations considerably without too much additional computational effort by using our full implicit schemes. The full implicit strong Taylor schemes allow a larger range of time step sizes than other schemes and are suitable for SSDE with stiffness on both the drift and the diffusion terms. Our numerical experiment shows these points.     

Gaussian beam decomposition of high frequency wave fields

In this paper, we present a method of decomposing a highly oscillatory wave field into a sparse superposition of Gaussian beams. The goal is to extract the necessary parameters for a Gaussian beam superposition from this wave field, so that further evolution of the high frequency waves can be computed by the method of Gaussian beams. The methodology is described for R^d${\mathbb{R}}^d$ with numerical examples for d=2$d=2$. In the first example, a field generated by an interface reflection of Gaussian beams is decomposed into a superposition of Gaussian beams. The beam parameters are reconstructed to a very high accuracy. The data in the second example is not a superposition of a finite number of Gaussian beams. The wave field to be approximated is generated by a finite difference method for a geometry with two slits. The accuracy in the decomposition increases monotonically with the number of beams.     

Mixtures of truncated basis functions

In this paper we propose a framework, called mixtures of truncated basis functions (MoTBFs), for representing general hybrid Bayesian networks. The proposed framework generalizes both the mixture of truncated exponentials (MTEs) framework and the Mixture of Polynomials (MoPs) framework. Similar to MTEs and MoPs, MoTBFs are defined so that the potentials are closed under combination and marginalization, which ensures that inference in MoTBF networks can be performed efficiently using the Shafer-Shenoy architecture.Based on a generalized Fourier series approximation, we devise a method for efficiently approximating an arbitrary density function using the MoTBF framework. The translation method is more flexible than existing MTE or MoP-based methods, and it supports an online/anytime tradeoff between the accuracy and the complexity of the approximation. Experimental results show that the approximations obtained are either comparable or significantly better than the approximations obtained using existing methods.     

A Discrete Chebyshev Approximation Problem by Means of Several Polynomials

The problem of approximation of a given function on a given set by a polynomial of a fixed degree in the Chebyshev metric (the Chebyshev polynomial approximation problem) is a typical problem of Nonsmooth Analysis (to be more precise, it is a convex nonsmooth problem). It has many important applications, both in mathematics and in practice. The theory of Chebyshev approximations enjoys very nice properties (the most famous being the Chebyshev alternation rule). In the present article the problem of approximation of a given function on a given finite set of points by several polynomials is studied. As a criterion function, the maximin deviation is taken. The resulting functional is nonsmooth and nonconvex and therefore the problem becomes multiextremal and may have local minimizers which are not global ones. A necessary and sufficient condition for a point to be a local minimizer is proved. It is shown that a generalized alternation rule is still valid. Sufficient conditions for a point to be a strict local minimizer are established as well. These conditions are also formulated in terms of alternants. An exchange algorithm for finding a local minimizer is constructed. An k -exchange algorithm, allowing to find a "better" local minimizer is stated. Numerical examples illustrating the theory are presented.     

A numerical evaluation of some collinear scaling algorithms for unconstrained

Collinear scaling algorithms for unconstrained minimization were first proposed by Davidon (1977,80) so that they may incorporate more information about the problem than is possible with quasi–Newton algorithms. Sorensen (1980,82), and Ariyawansa (1983,90) have derived collinear scaling algorithms as natural extensions of quasi–Newton algorithms. In this paper we describe the results of a comprehensive numerical evaluation of four members in the classes of collinear scaling algorithms derived by Sorensen (1980,82) and Ariyawansa (1983,90), relative to the quasi–Newton algorithms they extend.     

Ramanujan formula for the generalized Stirling approximation

The aim of this paper is to improve Ramanujan’s formula for approximation of the factorial function, starting from Burnside’s formula in contradistinction with the classical formula that starts from Stirling’s formula.     

Detectability of critical points of smooth functionals from their finite-dimensional approximations

Given a critical point of a C²-functional on a separable Hilbert space, we obtain sufficient conditions for it to be detectable (i.e. ‘visible’) from finite-dimensional Rayleigh-Ritz-Galerkin (RRG) approximations. While examples show that even nondegenerate critical points are, without any further restriction, not visible, we single out relevant classes of smooth functionals, e.g. the Hamiltonian action on the loop space or the functionals associated with boundary value problems for some semilinear elliptic equations, such that their nondegenerate critical points are visible from their RRG approximations.     

Approximations of stochastic Navier–Stokes equations

In this paper we show that solutions of two-dimensional stochastic Navier–Stokes equations driven by Brownian motion can be approximated by stochastic Navier–Stokes equations forced by pure jump noise/random kicks.     

On the minimum corridor connection problem and other generalized geometric problems

In this paper we discuss the complexity and approximability of the minimum corridor connection problem where, given a rectilinear decomposition of a rectilinear polygon into “rooms”, one has to find the minimum length tree along the edges of the decomposition such that every room is incident to a vertex of the tree. We show that the problem is strongly NP-hard and give a subexponential time exact algorithm. For the special case when the room connectivity graph is k-outerplanar the algorithm running time becomes cubic. We develop a polynomial time approximation scheme for the case when all rooms are fat and have nearly the same size. When rooms are fat but are of varying size we give a polynomial time constant factor approximation algorithm.