On the Law of The Iterated Logarithm for Rapidly Increasing Subsequences

The usual law of the iterated logarithm states that the partial sums S_n of independent and identically distributed random variables can be normalized by the sequence a_n = √nlog log n, such that limsup_n→∞ S_n/a_n = √2 a.s. As has been pointed out by Gut (1986) the law fails if one considers the limsup along subsequences which increase faster than exponentially. In particular, for very rapidly increasing subsequences {n_k≥1} one has limsup_k→∞ S_nk/a_nk = 0 a.s. In these cases the normalizing constants a_nk have to be replaced by √nk log k to obtain a non-trivial limiting behaviour: limsup_k→∞ S_nk/ √nk log k = √2 a.s. We will present an intelligible argument for this structural change and apply it to related results.     

Effective conductivity and skew Brownian motion

We consider heat conduction in a periodic body which is composed of finitely many different components. The effective conductivity is represented in terms of skew Brownian motion. The representation formula is a fluctuation-dissipation relation. The dissipation term in this formula is related to the transmission of heat through the surface separating the different components of the body; it is described by the skew reflections of Brownian motion at these surfaces. The problems caused by the discontinuity of the microscopic conductivity are handled in the framework of Dirichlet forms.     

Simulated annealing for constrained global optimization

Hide-and-Seek is a powerful yet simple and easily implemented continuous simulated annealing algorithm for finding the maximum of a continuous function over an arbitrary closed, bounded and full-dimensional body. The function may be nondifferentiable and the feasible region may be nonconvex or even disconnected. The algorithm begins with any feasible interior point. In each iteration it generates a candidate successor point by generating a uniformly distributed point along a direction chosen at random from the current iteration point. In contrast to the discrete case, a single step of this algorithm may generateany point in the feasible region as a candidate point. The candidate point is then accepted as the next iteration point according to the Metropolis criterion parametrized by anadaptive cooling schedule. Again in contrast to discrete simulated annealing, the sequence of iteration points converges in probability to a global optimum regardless of how rapidly the temperatures converge to zero. Empirical comparisons with other algorithms suggest competitive performance by Hide-and-Seek.This material is based on work supported by a NATO Collaborative Research Grant, no. 0119/89.     

Quenching problem of globally coupled bistable stochastic systems with finite size

The transient process of globally coupled bistable systems from an unstable state to metastable state (i.e, quenching process) is studied analytically for small noise intensity. The influences of noise intensity and system size on the system evolution are investigated. The problem of a large number of coupled Langevin equations is reduced to a simple problem of a one-dimensional ordinary differential equation, subject to a white noise with intensity explicitly given. The analytical results are fully confirmed by direct numerical computations. Received: 3 July 1997 / Revised: 4 December 1997 / Accepted: 15 January 1998     

State estimation for cox processes on general spaces

Let N be an observable Cox process on a locally compact space E directed by an unobservable random measure M. Techniques are presented for estimation of M, using the observations of N to calculate conditional expectations of the form E [M]|$\text{F}$_A], where $\text{F}$_A is the σ–algebra generated by the restriction of N to A. We introduce a random measure whose distribution depends on N_A, from which we obtain both exact estimates and a recursive method for updating them as further observations become available. Application is made to the specific cases of estimation of an unknown, random scalar multiplier of a known measure, of a symmetrically distributed directing measure M and of a Markov–directed Cox process on $\text{R}$. By means of a Poisson cluster representation, the results are extended to treat the situation where N is conditionally additive and infinitely divisible given M.     

A limit theorem for the eigenvalues of product of two random matrices

The existence of limit spectral distribution of the product of two independent random matrices is proved when the number of variables tends to infinity. One of the above matrices is the Wishart matrix and the other is a symmetric nonnegative definite matrix.     

一种基于布朗粒子的混合搜索模型

基于网络上的布朗粒子运动基本原理,提出了一种单粒子和多粒子相结合的混合搜索模型.该模型将一次搜索过程分成单粒子搜索与多粒子搜索两个阶段,既克服了单粒子搜索效率低下的缺点,又降低了多粒子搜索的硬件代价.在各种复杂网络拓扑上实施该模型,并与混合导航模型进行比较.结果表明,混合搜索模型的平均搜索时间收敛更快,硬件代价更小.将度大优先的目标选择策略与混合搜索模型相结合,能进一步提高搜索效率.此外通过仿真发现,在无标度网络上混合搜索模型的效率远高于单粒子随机行走,与多粒子随机行走的效率相当,但硬件代价远小于多粒子行走.最后针对该模型给出了一种能有效降低负载的"吸收"策略.     

电磁散射问题的两种反演方法研究

提出了电磁散射问题的两种反演方法:确定性梯度搜索方法和MonteCarlo随机搜索方法.给出了两种方法的基本原理,并对反演性能包括对散射体的定位能力、反演精度、反演时间进行了系统的分析与比较.数值计算结果表明,确定性方法定位精确,而随机性方法对散射体电磁参数的反演精度高且速度快,利用两者优点可以增强反演效果.     

随机耦合模型在高功率微波效应中的应用

分析了微波混沌腔体系统中关键部位感应电压的统计问题,介绍了随机耦合模型在高功率微波效应研究中的计算方法和应用,并以计算机机箱为实验系统,开展了电磁波耦合入计算机机箱腔体的电磁干扰问题研究,对其电路板上关键部位感应电压的统计计算和实验结果进行了比较,其结果基本一致。随机耦合模型在波混沌系统中感应电压的统计探讨为高功率微波效应、电磁兼容等研究提供了一种新的思路。