A comprehensive probabilistic analysis of approximate SIR-type epidemiological models via full randomized discrete-time Markov chain formulation with applications

This paper provides a comprehensive probabilistic analysis of a full randomization of approximate SIR-type epidemiological models based on discrete-time Markov chain formulation. The randomization is performed by assuming that all input data (initial conditions, the contagion, and recovering rates involved in the transition matrix) are random variables instead of deterministic constants. In the first part of the paper, we determine explicit expressions for the so called first probability density function of each subpopulation identified as the corresponding states of the Markov chain (susceptible, infected, and recovered) in terms of the probability density function of each input random variable. Afterwards, we obtain the probability density functions of the times until a given proportion of the population remains susceptible, infected, and recovered, respectively. The theoretical analysis is completed by computing explicit expressions of important randomized epidemiological quantities, namely, the basic reproduction number, the effective reproduction number, and the herd immunity threshold. The study is conducted under very general assumptions and taking extensive advantage of the random variable transformation technique. The second part of the paper is devoted to apply our theoretical findings to describe the dynamics of the pandemic influenza in Egypt using simulated data excerpted from the literature. The simulations are complemented with valuable information, which is seldom displayed in epidemiological models. In spite of the nonlinear mathematical nature of SIR epidemiological model, our results show a strong agreement with the approximation via an appropriate randomized Markov chain. A justification in this regard is discussed.     

Probability and exams: The work of Antonio Bordoni

The Italian mathematician Antonio Bordoni is mainly known for his adherence to the Lagrangian approach to the foundations of calculus and for his role in creating an important school of mathematics. In this paper, I consider his less known work on the application of probability to design exams and analyze their outcomes. Within this framework, he obtained in 1837, as Mondésir and Poisson, the result that would lead Catalan to formulate his “new principle” of probability (Jongmans and Seneta, 1994). Moreover, in 1843, Bordoni also gave an early complete proof of the finite rule of succession.     

基于多项式混沌拓展模型和bootstrap的结构可靠性分析方法

多项式混沌拓展(polynomial chaos expansion,PCE)模型现已发展为全局灵敏度分析的强大工具,却很少作为替代模型用于可靠性分析。针对该模型缺乏误差项从而很难构造主动学习函数来逐步更新的事实,在结构可靠性分析的框架下提出了基于PCE模型和bootstrap重抽样的仿真方法来计算失效概率。首先,对试验设计(experimental design)使用bootstrap重抽样步骤以刻画PCE模型的预测误差;其次,基于这个局部误差构造主动学习函数,通过不断填充试验设计以自适应地更新模型,直到能够精确地逼近真实的功能函数;最后,当PCE模型具有足够精确的拟合、预测能力,再使用蒙特卡洛仿真方法来计算失效概率。提出的平行加点策略既能在模型更新过程中找到改进模型拟合能力的"最好"的点,又考虑了模型拟合的计算量;而且,当失效概率的数量级较低时,PCE-bootstrap步骤与子集仿真(subset simulation)的结合能进一步加速失效概率估计量的收敛。本文方法将PCE模型在概率可靠性领域的应用从灵敏度分析延伸到了可靠性分析,同时,算例分析结果显示了该方法的精确性和高效性。     

非时齐复合泊松过程下对偶风险模型的破产特征量研究

将经典的对偶风险模型中的收益到达过程推广为非时齐的泊松过程.运用经典方法和时变方法,计算了该模型下的破产概率,并定义了时变后相应模型的广义期望折罚函数,验证了时变方法对非时齐泊松风险模型的有效性,最后又考虑了该模型在带壁分红策略下的情形,当单次索赔额服从指数分布时,得到了它的期望折罚函数以及期望折现分红函数.     

凸概率密度分布簇下的多损失鲁棒优化等价模型及在直营连锁企业中的应用

首先,证明了凸概率密度分布簇的单周期期望均值下单损失鲁棒优化等价模型定理,以及凸概率密度分布簇的单周期期望均值下多损失鲁棒优化等价模型。然后,提出了直营连锁企业的产品在凸概率密度分布簇下的期望均值的单周期生产分配供应问题,建立了直营连锁企业的单周期生产分配供应期望均值鲁棒模型,在获得近似周期概率分布簇情形下给出了单周期生产分配供应鲁棒模型,这种近似鲁棒模型等价于一个线性规划问题。最后,通过已知一个产品的4个周期构成的混合分布簇进行了数值实验,数值结果表明了期望均值准则下的生产分配供应鲁棒模型的生产分配供应策略更加稳健。     

计算数学期望和概率的条件化方法

借助于条件数学期望和随机事件A的示性函数IA,通过对随机变量的适当"条件化"处理,应用全期望公式和推广的全概率公式,讨论了计算数学期望和概率的条件化方法.     

一类古典概型的公式解

针对一类古典概型,通过将基本事件数转化为不定方程的解的个数,并利用组合方法将其求出,进而可以得到该古典概型的公式解.     

常微分方程形式的M/M/1排队算子的谱分析

在l~1空间研究了常微分方程形式的M/M/1排队模型确定的算子A的谱问题.通过细致的谱分析,表明算子A的谱是一个椭圆型,椭圆内部点全是算子A的本征值.0位于椭圆的右边界点是边界上唯一的本征值,从而0不能与其它谱点相分离.这一结果表明常微分方程形式的M/M/1排队系统在有限时间不可能看到系统的稳定状态.     

带干扰双Poisson风险模型生存概率的Feller表示

研究了带干扰双Poisson风险模型,运用鞅论的方法给出了该模型生存概率的Feller表示式.     

威布尔分布无失效数据失效概率的估计

通过讨论威布尔分布函数形状参数m的大小,给出各检测时刻失效概率pi的相互关系作为先验信息,得到pi(i=1,2,…,n)的Bayes估计,并且试验数据显示此种方法是可行的.