基于参数化分位回归模型的非寿险准备金评估

准备金及其风险边际对保险公司的偿付能力具有决定性影响.均值回归模型在非寿险准备金评估中的应用较为普遍,但需要通过Bootstrap等方法计算准备金的风险边际.分位回归模型可以一次性求得准备金及其风险边际的预测值,所以在非寿险准备金评估中具有独特的应用价值.基于GB2(Generalized Beta type 2)分布建立了一种参数化分位回归模型,该模型首先对GB2分布中的位置参数和尺度参数同时引入流量三角形数据中的事故年和进展年作为解释变量,增加了模型的灵活性;其次,根据模型参数的极大似然估计结果,借助分位数函数的表达式,计算了不同分位数水平下的准备金预测值;最后,利用极大似然估计的渐近性质,通过Delta方法给出了准备金预测值的误差.基于一组增量赔款数据的实证研究结果表明,GB2参数化分位回归模型在非寿险准备金评估及其风险边际的预测中具有良好的应用价值.     

判定矩阵可逆的几个充分条件

利用矩阵范数和非奇异M-矩阵的性质以及Raleigh商值定理,给出了判定矩阵非奇异的几个充分条件.     

Constructal entransy optimizations for insulation layer of steel rolling reheating furnace wall with convective and radiative boundary conditions

Based on the entransy dissipation extremum principle for thermal insulation process, the constructal optimizations for a plane insulation layer of the steel rolling reheating furnace wall with convective and radiative boundary conditions are carried out by taking the minimization of entransy dissipation rate as optimization objective. The optimal construct of the plane insulation layer is obtained. The results show that for the convective heat transfer boundary condition, the optimal constructs of the insulation layer obtained based on the minimizations of the entransy dissipation rate and heat loss rate are obvi- ously different. Comparing the optimal construct obtained based on the minimization of the entransy dissipation rate with that based on the minimization of the heat loss rate, the entransy dissipation rate is reduced by 5.98 %, which makes the global thermal insulation performance of the insulation layer improve. For the combined convective and radiative heat transfer boundary condition, compared the insulation layer having an increasing thickness with that having constant thickness and a decreasing thickness, the entransy dissipation rates are reduced by 16.59 % and 39.72 %, respectively, and the global thermal insulation performance of the insulation layer is greatly improved.There exits an optimal constant coefficient α2,opt which leads to the minimum dimensionless entransy dissipation rate of the insulation layer. The difference between the optimal constant coefficients α2,opt obtained based on the minimizations of the entransy dissipation rate and the maximum temperature gradient of the insulation layer is small. This makes the corresponding thermal stress obtained based on the minimum dimensionless entransy dissipation rate also be small, and the global thermal insulation performance and thermal safety of the insulation layer are improved simultaneously. The results obtained can provide some guidelines for the optimal designs of the insulation layers.     

广义Lucas数列中平方数的确定

设数列{ Uk}和{Vk}是关于参数为P,Q的广义Lucas数列,确定了当2| P,且Q=-1时广义Lucas数列中的平方数.     

The generalized partial transposition criterion for infinite-dimensional quantum systems

The positive partial transposition(PPT)criterion and the realignment criterion constitute two of the most important criteria for detecting entanglement.The generalized partial transposition(GPT)criterion which contains the PPT criterion and the realignment criterion as special cases,provides a necessary condition for a multipartite state to be separable.Here we extend the GPT criterion to the infinitedimensional multipartite(bipartite)case and show that it includes multipartite(bipartite)PPT criterion and multipartite(bipartite)realignment criterion as special cases as well.     

基于离散阵列广义同步的彩色图像加密算法

为设计出具有较高安全性能的图像算法,提出了一种基于离散阵列广义同步系统的彩色图像加密算法。首先利用离散阵列广义同步定理,在3D-Lorenz映射的基础上构造了一个具有16×16×6的阵列广义同步系统。其次,结合二维离散分数阶傅里叶变换设计了一种彩色图像加密算法。该算法对像素矩阵进行了新混沌序列的位置变换,提高了密钥空间。加密算法的密钥敏感性、直方图和信息熵的仿真实验表明:该加密算法安全性高,能够达到较好的加解密效果。     

二维无解区域上一类广义Navier-Stokes方程的衰减率

主要研究二维无解区域Ω上一类广义Navier-Stokes方程速度梯度的L2衰减率.当u0∈L2(Ω)时利用新的能量方法和精确的计算得到了其速度梯度在L2范数下的衰减率为(1+t)-12.     

空间中混合均衡问题和相对拟非扩张半群公共解的一种逼近方法

在Banach空间中,通过广义投影算子,建立一种求解混合均衡问题和相对拟非扩张半群的公共解的迭代算法,并证明迭代序列的强收敛性.     

约定性结构还原与广义科学统一

还原与反还原的争论,科学统一与反科学统一的争论,是当今科学技术哲学界的两大争论不休的问题。争论的每一方都有著名的哲学家支持,各执一方。我在此两大争论基础上,提出约定性结构还原,狭义科学统一,广义科学统一。同时从约定性结构还原与广义科学统一的密切关系中,用新的视角,试图寻找还原与科学统一的研究新进路。     

广义三角模的近似表示

给出了在理论和应用方面都具有重要意义的连续广义三角模的一种近似表示，这种表示可以“逼近”到事先任意指定的精度。