Approximations of non-smooth integral type functionals of one dimensional diffusion processes

In this article, we obtain the weak and strong rates of convergence of time integrals of non-smooth functions of a one dimensional diffusion process. We propose the use of the exact simulation scheme to simulate the process at discretization points. In particular, we also present the rates of convergence for the weak and strong errors of approximation for the local time of a one dimensional diffusion process as an application of our method.     

Adaptive estimation of an ergodic diffusion process based on sampled data

We consider adaptive maximum likelihood type estimation of both drift and diffusion coefficient parameters for an ergodic diffusion process based on discrete observations. Two kinds of adaptive maximum likelihood type estimators are proposed and asymptotic properties of the adaptive estimators, including convergence of moments, are obtained.     

扩散过程代数式收敛定性的判别准则

本文定性地讨论非紧空间中可逆扩散过程的代数式收敛的判定 .使用分裂空间的方法 .将全空间分裂成两个部分 :紧的子空间与非紧的余子空间 .在紧子空间中考虑边界反射的Neumann过程 ,它必然是代数式收敛的 .而在非紧子空间中考虑边界吸收的Dirichlet过程 ,如果这一Dirichlet过程以代数式的速度击中边界 ,那么就有原过程在全空间代数式收敛 ;反之 ,原过程代数式收敛 ,非紧子空间中的Dirichlet过程也是代数式收敛的 .因此过程在紧子空间的任意摄动不会影响在全空间的代数式收敛性 .     

Maximum likelihood estimation for multiscale Ornstein–Uhlenbeck processes

We study the problem of estimating the parameters of an Ornstein–Uhlenbeck (OU) process that is the coarse-grained limit of a multiscale system of OU processes, given data from the multiscale system. We consider both the averaging and homogenization cases and both drift and diffusion coefficients. By restricting ourselves to the OU system, we are able to substantially improve the results with strong modes of convergence, and provide some intuition of what to expect in the general case. In particular, in the homogenisation case we derive optimal rates of sub-sampling to minimize the estimation errors.     

A Test of Correlation in the Random Coefficients of an Autoregressive Process

A random coefficient autoregressive process in which the coefficients are correlated is investigated. First we look at the existence of a strictly stationary causal solution, we give the second-order stationarity conditions and the autocorrelation function of the process. Then we study some asymptotic properties of the empirical mean and the usual estimators of the process, such as convergence, asymptotic normality and rates of convergence, supplied with appropriate assumptions on the driving perturbations. Our objective is to get an overview of the influence of correlated coefficients in the estimation step through a simple model. In particular, the lack of consistency is shown for the estimation of the autoregressive parameter when the independence hypothesis in the random coefficients is violated. Finally, a consistent estimation is given together with a testing procedure for the existence of correlation in the coefficients. While convergence properties rely on ergodicity, we use a martingale approach to reach most of the results.     

Rates of convergence of diffusions with drifted Brownian potentials

We are interested in the asymptotic behaviour of a diffusion process with drifted Brownian potential. The model is a continuous time analogue to the random walk in random environment studied in the classical paper of Kesten, Kozlov, and Spitzer. We not only recover the convergence of the diffusion process which was previously established by Kawazu and Tanaka, but also obtain all the possible convergence rates. An interesting feature of our approach is that it shows a clear relationship between drifted Brownian potentials and Bessel processes.

     

Approximation of an Analog Diffusion Network with Applications to Image Estimation

This work is concerned with a numerical procedure for approximating an analog diffusion network. The key idea is to take advantage of the separable feature of the noise for the diffusion machine and use a parallel processing method to develop recursive algorithms. The asymptotic properties are studied. The main result of this paper is to establish the convergence of a continuous-time interpolation of the discrete-time algorithm to that of the analog diffusion network via weak convergence methods. The parallel processing feature of the network makes it attractive for solving large-scale optimization problems. Applications to image estimation are considered. Not only is this algorithm useful for the image estimation problems, but it is widely applicable to many related optimization problems.     

漂移向量与扩散矩阵的非参数估计模型

考虑多维扩散过程的非参数估计问题.利用It扩散的性质,将漂移向量和扩散矩阵的样本表示成带有测量误差的回归模型,并讨论了系统误差的L~r上界以及随机误差项的收敛速度,建立了漂移向量与扩散矩阵非参数估计的通用模型.     

Spectral estimation of the Lévy density in partially observed affine models

The problem of estimating the Lévy density of a partially observed multidimensional affine process from low-frequency and mixed-frequency data is considered. The estimation methodology is based on the log-affine representation of the conditional characteristic function of an affine process and local linear smoothing in time. We derive almost sure uniform rates of convergence for the estimated Lévy density both in mixed-frequency and low-frequency setups and prove that these rates are optimal in the minimax sense. Finally, the performance of the estimation algorithms is illustrated in the case of the Bates stochastic volatility model.     

Quasi-likelihood analysis for the stochastic differential equation with jumps

In this paper, we consider a multidimensional diffusion process X with jumps whose jump term is driven by a compound Poisson process, and discuss its parametric estimation. We present asymptotic normality and convergence of moments of any order for a quasi-maximum likelihood estimator and a Bayes type estimator by assuming an exponential mixing property of X. To show these properties, we use the polynomial type large deviation theory.     

Asymptotic behavior of M-estimator and related random field for diffusion process

The M-estimate which maximizes a positive stochastic process Q is treated for multidimensional diffusion models. The convergence in distribution of the process of ratio of Q's after normalizing is proved. The asymptotic behavior of M-estimates is stated. We present the asymptotic variance in general cases and in estimation by misspecified models.     

BANDWIDTH SELECTION IN NONPARAMETRIC SPECTRAL DENSITY ESTIMATION OF THE STATIONARY GAUSSIAN PROCESS

ＢＡＮＤＷＩＤＴＨＳＥＬＥＣＴＩＯＮＩＮＮＯＮＰＡＲＡＭＥＴＲＩＣＳＰＥＣＴＲＡＬＤＥＮＳＩＴＹＥＳＴＩＭＡＴＩＯＮＯＦＴＨＥＳＴＡＴＩＯＮＡＲＹＧＡＵＳＳＩＡＮＰＲＯＣＥＳＳ￥ＹＵＤＡＮ（于丹）（ＩｎｓｔｉｔｕｔｅｏｆＳｙｓｔｅｍｓＳｃｉｅｎｃｅ，...     

Wavelet estimation of the diffusion coefficient in time dependent diffusion models

The estimation problem for diffusion coefficients in diffusion processes has been studied in many papers,where the diffusion coefficient function is assumed to be a 1-dimensional bounded Lipschitzian function of the state or the time only.There is no previous work for the nonparametric estimation of time-dependent diffusion models where the diffusion coefficient depends on both the state and the time.This paper introduces and studies a wavelet estimation of the time-dependent diffusion coefficient under a more general assumption that the diffusion coefficient is a linear growth Lipschitz function.Using the properties of martingale,we translate the problems in diffusion into the nonparametric regression setting and give the L~r convergence rate.A strong consistency of the estimate is established.With this result one can estimate the time-dependent diffusion coefficient using the same structure of the wavelet estimators under any equivalent probability measure.For example, in finance,the wavelet estimator is strongly consistent under the market probability measure as well as the risk neutral probability measure.     

Malliavin calculus for difference approximations of multidimensional diffusions: Truncated local limit theorem

The truncated local limit theorem is proved for difference approximations of multidimensional diffusions. Under very mild conditions on the distributions of difference terms, this theorem states that the transition probabilities of these approximations, after truncation of some asymptotically negligible terms, possess densities uniformly convergent to the transition probability density for the limiting diffusion and satisfy certain uniform diffusion-type estimates. The proof is based on a new version of the Malliavin calculus for the product of a finite family of measures that may contain nontrivial singular components. Applications to the uniform estimation of mixing and convergence rates for difference approximations of stochastic differential equations and to the convergence of difference approximations of local times for multidimensional diffusions are presented. Published in Ukrains’kyi Matematychnyi Zhurnal, Vol. 60, No. 3, pp. 340–381, March, 2008.     

Optimal rates for parameter estimation of stationary Gaussian processes

We study rates of convergence in central limit theorems for partial sums of polynomial functionals of general stationary and asymptotically stationary Gaussian sequences, using tools from analysis on Wiener space. In the quadratic case, thanks to newly developed optimal tools, we derive sharp results, i.e. upper and lower bounds of the same order, where the convergence rates are given explicitly in the Wasserstein distance via an analysis of the functionals’ absolute third moments. These results are tailored to the question of parameter estimation, which introduces a need to control variance convergence rates. We apply our result to study drift parameter estimation problems for some stochastic differential equations driven by fractional Brownian motion with fixed-time-step observations.     

A new approach for numerical identification of conductivity

This paper studies the parameter estimation problem for a steady state flow in an inhomogeneous medium. For identifying the spatially varying diffusion coefficient from an observed solution to the forward problem, we propose a direct method using local Green's function technique. This could be used when the diffusion coefficient is discontinuous. The convergence order is calculated and numerical simulations are performed.     

On smoothed probability density estimation for stationary processes

Aspects of estimation of the (marginal) probability density for a stationary sequence or continuous parameter process, are considered in this paper. Consistency and asymptotic distributional results are obtained using a class of smoothed function estimators including those of kernel type, under various decay of dependence conditions for the process. Some of the consistency results contain convergence rates which appear to be more delicate than those previously available, even for i.i.d. sequences.     

On identification of the threshold diffusion processes

We consider the problems of parameter estimation for several models of threshold ergodic diffusion processes in the asymptotics of large samples. These models are the direct continuous time analogues of the well known in time series analysis threshold autoregressive models. In such models, the trend is switching when the observed process attaints some (unknown) values and the problem is to estimate it or to test some hypotheses concerning these values. The related statistical problems correspond to the singular estimation or testing, for example, the rate of convergence of estimators is T and not ?T{\sqrt{T}} as in regular estimation problems. We study the asymptotic behavior of the maximum likelihood and Bayesian estimators and discuss the possibility of the construction of the goodness-of-fit test for such models of observation.     

Estimating functions for noisy observations of ergodic diffusions

In this article, general estimating functions for ergodic diffusions sampled at high frequency with noisy observations are presented. The theory is formulated in terms of approximate martingale estimating functions based on local means of the observations, and simple conditions are given for rate optimality. The estimation of the diffusion parameter is faster than the estimation of the drift parameter, and the rate of convergence is classical for the drift parameter but not classical for the diffusion parameter. The link with specific minimum contrast estimators is established, as an example.