Asymptotic Expansions of Large Deviations for Sums of Nonidentically Distributed Random Variables

The work is designated for obtaining asymptotic expansions and determination of structures of the remainder terms that take into consideration large deviations both in Cramer zones and Linnik power zones for the distribution function of sums of independent nonidentically distributed random variables (r.v.). In this scheme of summation of r.v., the results are obtained first by mainly using the general lemma on large deviations considering asymptotic expansions for an arbitrary r.v. with regular behaviour of its cumulants [11]. Asymptotic expansions in the Cramer zone for the distribution function of sums of identically distributed r.v. were investigated in the works [1,2]. Note that asymptotic expansions for large deviations were first obtained in the probability theory by J. Kubilius [3].     

信息熵通过系统变化规律的研究

对孤立热力系统中的不可逆过程，热力系统熵自发不可逆地增加；而对自发进行的物理变化过程，物理场熵是减小的。迄今，信息系统中信息熵的变化规律尚不十分明确。对此进行研究，得出结论：带限平稳正态随机信号通过因果最小相位线性系统后，熵可增加或减小，可用线性逆系统使熵可逆变化；平稳非正态独立同分布随机信号通过因果非最小相位线性系统后，熵恒增加，并趋近平稳正态信号，使熵趋于最大。原线性系统不能使熵可逆变化，亦不     

A cumulant-based parameter estimation algorithm for near-field sources

To efficiently use the array aperture and avoid the complicated parameter pairing computation, this paper proposes a new cumulant-based algorithm for localizing near-field narrowband sources. Firstly, this algorithm proposes a symmetric uniform linear array (ULA), and constructs three cumulant matrices by using the fourth-order cumulants of some properly chosen sensor outputs; secondly, unlike the conventional parallel factor (PARAFAC) analysis models in data or subspace domain, it forms a three-way array (TWA) in the fourth-order cumulant domain using the three matrices, and analyzes the uniqueness of its low-rank decomposition; thirdly, it jointly estimates the frequency, the direction-of-arrival (DOA), and the range of each near-field source from the matrices via the low-rank decomposition of the TWA. The simulation results are presented to validate the performance of our proposed method.     

基于累积量增强全矢量正交子空间方法的MA建模

提出了一种新的线性代数方法——累积量增强全矢量正交子空间方法(CEAVOS)用于非最小相位非高斯滑动平均(MA)建模，该方法利用组合特性映射的累积量增强，并用全矢量正交于空间法估计MA参数．数值仿真结果表明，CEAVOS的性能优于组合累积量切片法WS和全矢量正交子空间法(AVOS)这2种现有的性能最好的MA参数估计方法，尤其是在估计的偏差上；在低信噪比与短数据的情况下，性能也表现良好．     

基于高阶累积量的地震子波提取

在地震勘探中,子波处理是提高地震资料分辨率的一个重要手段,而该技术的关键是地震子波的估计.由于影响地震波的因素太多且复杂,使得子波的提取始终是一个十分困难的问题.传统的提取子波的方法主要利用地震记录的二阶统计信息,然而由于二阶统计量并不包含相位信息,因而无法解决非最小相位的物理问题.鉴于信号高阶累积量的性质和特点,利用高阶累积量来达到提取地震子波的目的.     

有限温度格点U(1)及定模U(1)–Higgs模型Symanzik作用量的解析研究

用变分累积展开方法(VCE)解析研究了Symanzik作用量形式的有限温度格点U(1)及定模U(1)-Higgs规范模型,并计算了序参量Polyakov线及临界指数β.结果表明,采用Symanzik作用量的变分累积展开比Wilson作用量收敛快.     

基于规范高阶累积量的盲分离算法

本文给出信号的r阶规范累积量定义,证明了在信号瞬时线性混叠情况下r阶规范累积量绝对值不超过最大源信号的r阶规范累积量,因此可以通过最大化r阶规范累积量的绝对值实现盲分离.最大化r阶规范累积量的绝对值可以得到两种盲分离算法,一种高效快速算法———特征值分解盲分离算法,另外一种盲提取算法.本文为高阶累积量盲分离奠定了严格的理论基础.仿真结果验证了理论的正确性和算法的有效性.     

Statistical mechanics of a nonlinear stochastic model

A multivariable Fokker-Planck equation (FPE) is used to investigate the equilibrium and dynamical properties of a nonlinear stochastic model. The model displays a phase transition. The equilibrium distributions are found to be non-Gaussian; the deviation from Gaussian is especially significant near the transition point. To study the nonequilibrium behavior of the model, a self-consistent dynamic mean field (SCDMF) theory is derived and used to transform the FPE to a systematic hierarchy of equations for the cumulant moments of the time-dependent distribution function. These equations are numerically solved for a variety of initial conditions. During the time evolution of the system from an initial unstable equilibrium state to the final equilibrium state, three distinct time stages are found.Supported by a grant from the National Research Council of Canada (to RCD) and by the Sherman Fairchild Foundation (to RZ).Also Sherman Fairchild Distinguished Scholar, 1974–75, at the California Institute of Technology, where the early part of this research was done.     

Stochastic Optimal Control of Nonlinear Systems via Short-Time Gaussian Approximation and Cell Mapping

A novel strategy to obtain global solutions of stochasticoptimal control problems with fixed state terminal conditions and controlbounds is proposed in this paper. The solution is global in the sense that theoptimal control solutions for all the initial conditions in a region of thestate space are obtained. The method makes use of Bellman's principle ofoptimality, the cumulant neglect closure method and the short-time Gaussianapproximation. A Markov chain with a control dependent transition probabilitymatrix is built using the generalized cell mapping method. This allows toevaluate the transient and steady state response of the controlled system. Themethod is applied to several linear and nonlinear systems leading to excellentcontrol performances.