This paper studies the solution of real symmetric positive definite Toeplitz matrices by the preconditioned conjugate gradient method.Based on the fact that the real symmetric Toeplitz matrix admits a triangular transform splitting(TTS),we propose the shifted Sine preconditioner TS,the spectral properties of the preconditioned matrix are analyzed,and the arithmetic complexity is discussed.The numerical experiments show that our preconditioner is more effective than T.Chan＇s preconditioner[2].
Analysis on Effects of Opening Residential Area on Road Capacity with Jackson Queueing Networks
In this paper,by applying the discrete event dynamic system modeling method a Jackson queueing network model is established to study the effects of opening residential area on road capacity with two examples. The road traffic states in surrounding area are deeply analyzed by using some performance measures of Jackson queueing networks.Based on our results,some suggestions about whether or not to open residential area according to the environment of surroundings,road traffic flow,road capacity,convenience to the transportation of citizens and the safety are provided.
The Decay Parameter of Stopped MX/M3/1 Queueing Model
Posteriori Error Estimates for the Variable BDF2 Method for Parabolic Equations
A Generalization of The Generalized Entropy Ergodic Theorem for Nonhomogeneous Markov Chains
In this paper we give a generalization to the generalized entropy ergodic theorem for nonhomogeneous Markov chains given by Wang and Yang in[4].
In this paper,two kinds of stability theorems are improved.By the property of the Kfunction,the property of the function’s derivative in original Markin stability theorem is extended from the finite time to the infinite time,and this brings the new method to get the stability of the solution of the nonautonomous differential equation.By the supplement of the integral convergence theorem ,the new theorem is further improved and promoted,and a new determination method about uniform stability is obtained.
A New Auxiliary Function-based Integral Inequality
Neural Network Weight-Threshold Optimization Method Based on LASSO
Based on the LASSO method and neural network,we establish a LASSO-BP algorithm.Compared with the BP neural network algorithm and the RBF algorithm,it has the advantages of time-consuming, strong anti-interference ability and better effects on classification.Its effectiveness is verified by numerical experiments.
Study of Correlations Between the Cash Flow of High-value Payment System in the People’s Bank of China and the CPI
The high-value payment system (HVPS)provides fast,efficient and secure payment services for enterprises and financial markets and prevents payment risks.This paper investigates the correlations between the cash flow changes of the People＇s Bank of China(PBS)in Hunan province and the economic index CPI in Hunan province by building a regression model.The results show that there is a highly positive correlation between the cash flow changes and the CPI.By using the regression model,we can predict the CPI and provide an effective method for the local government to make economic decisions.
A Bidding Model Under Non-cooperation Mode Based on Least Squares
An Ant Colony Optimization Algorithm for Multi-criteria Degree-constrained Minimum Spanning Tree
Both multi-criteria minimum spanning tree problem and degree-constrained minimum spanning tree problem are NP-hard problems in network optimization.They always have many important applications in practice.To solve the minimum spanning tree problem by considering both the multi-criteria and degreeconstraints, this paper proposes an algorithm (coded in Delphi)which is based on the ideas of ant colony optimization. The computational numerical experiments show the efficiency and effectiveness of the algorithm for the multi-criteria degree-constrained minimum spanning tree problem.
Solving Vehicle Routing Problems with the Greedy Randomized Adaptive Bat Algorithm
The Vehicle Routing Problem(VRP)is a worth researching NP-Hard problem in logistic and supply chain.The Bat Algorithm (BA)is a new intelligent optimization algorithm having broad application prospect. But it can’t use to solve discrete problem directly,and just like most intelligent optimization algorithms, it may easily fall into local optimum and its convergence rate is very slow in the late stage of the algorithm.In this paper,in view of the specific characteristics of the VRP,we redefine the coding mode and use GRASP to generate the initial population of BA to improve the algorithm,and then apply it to solve the VRP.
Function fitting is an important data mining technique.In this paper,the modified Gauss type mixed model is applied to the study of the changing pattern of the water level in order to make accurate prediction of the water level.Firstly,we make a pretreatment of the ten years data of the water level and establish a Gauss-like mixed model which satisfies the smallest loss function via curve fit.Then,we construct an overdetermined linear equations and use the least square method to obtain the corresponding weight.Finally,we get a weighted model and examine the rationality of the model and make some improvements.
Investigation and Analysis on the Life Quality of Orphans in the First Social Welfare Institute of Changsha
An Empirical Analysis on the Impact of China′s Import and Export Trade on Taxation
Properties of Quadratic Weighted Markov Branching Processes with Immigration and Resurrection
In this paper we study the regularity,uniqueness,recurrence and ergodicity of the quadratic weighted Markov branching processes with immigration and resurrection(QWMBPIR).Firstly,we investigate the properties of the generating function for QWMBIR q-matrix.It is proved that the QWMBPIR is regular and unique.Then we discuss the recurrence and ergodicity of QWMBPIR and give a sufficient condition for the ergodicity.
General Exchange Option Pricing in Sub-fractional Brownian Motion Environments
his paper studies the pricing problem of general exchange options in sub-fractional Brownian motion environments.Under the condition that the two stock pricing processes obey the stochastic differential equation driven by the sub-fractional Brownian motion,the pricing formula of the general exchange options is obtained by insurance actuary pricing.
Stability Analysis of Solutions for a Class of Stochastic Difference Equations
In this paper we study the image denoising algorithm based on total variation.The corresponding optimization model is solved by the steepest descend method,the difference iterative method and the split Bregman method,respectively.Experiment results show that the difference iterative method convergences rapidly,and achieves better denoising performances.
A Risk Model of Two Type Claims Under a Ruin Threshold
In this paper,we study a risk model of two type claims under a time-dependent ruin bound,where one term policy’s arrival intensity follows a Poisson process of parameterλand its refunding,its abnormal claims and normal claims are respectively relate to the policies ofρ1-thinning process,ρ2-thinning process andρ3 -thinning process,while the other term policy and claim arrival both follow the compound negative binomial distribution.For the risk model we investigate the properties of the surplus process by applying the martingale approach and derive the formula of ruin probability and the Lundberg inequality.
Ruin Probability of a Dependent Risk Model with Generalized Polya-Aeppli Distribution
In this paper,we study the ruin probability of a risk model with two dependent compensation processes based on the generalized Polya-Aeppli distribution.Firstly,the joint probability distribution function and the precise expressions of moments for a class of dependent processes are derived by applying the probability generating function defined by Kocherlakota (1995).Then two kinds of ruin models are formulated,and the corresponding ruin probabilities are obtained by using the Laplace transformation to covert computing ruin probability to calculating probability distribution function of the cumulative claims when the claims follow the exponential distribution.The generalized Polya-Aeppli distribution defines a class of discrete distributions with correlation.It overcomes the over-dispersion problem of the actual data which cannot be modeled by Poisson process,and is easy to estimate the parameters.Thus our approach has a wide range of applicability.
A New Cryptosystem Based on a Kind of Integer Matrix Equations
Numerical Simulation of Heavy Metal Migration Process in Submarine Near Deep Sea Mining
Extension of the Wallis Formula and Stirling Formula
Let {Xk,1≤k ≤n}be independent and identically distributed random variables with bilaterally truncated Cauchy distribution of parameters μ,λ,A,B,X1,n,X2,n,…,Xn,n be their order statistics.In this paper we obtain the density function of Xk,n,the joint density function of X1,n,X2,n,…,Xn,n,the asymptotic distributions of their extreme order statistics X1,n and Xn,n,and the asymptotic distributions of Xk,n and Xn-k+1,n.We also show that X1,n and Xn,n is asymptotically independent.
Solving the Optimization Model of Equipment Preventive Maintenance Period with the Genetic Algorithm and Nonlinear Programming
Branching and order-reducing are widely used for solving NP-Hard Problems.The main idea of the approach is to solve the problem by decomposing it into two or more sub-problems,and the sub-problems can be recursively solved.But we cannot get a better result by using the normal complexity analyticalmethod since it is not so accurate.The measure and conquer approach is a new technique for algorithm design and complexity analysis.This paper designs an algorithm to solve the weighted maximum clique problem and uses traditional analysis technology to analyze the worst-case running time of the algorithm and gets O(1.4656np(n))running time,where p(n)is the polynomial function of node number nin the problem.We employ the measure and conquer approach to improve the time complexity of the algorithm from O(1.4656np(n)) to O (1.3765np(n))without modifying the algorithm. Our results show that the measure and conquer approach can give more precise time complexity.
Research on Credit Risk Prevention of Commercial Bank in China Based on Multiple Linear Regression
Research of Human Machine Identification for Sliding Verification Code Based on Neural Network
In the process of sliding verification,it is very important to distinguish whether the drill author is “machine”or“personal”.In this paper,we use the neural network algorithm and MATLAB software to make an empirical study and analysis on the characteristics of the sliding trajectory extraction left by human and machine,and establish the classification model of the neural network to predict the operator.The results show that the BP neural network model has high prediction accuracy and provides a guarantee for network security to some extent.
