ISSN 0253-2778

CN 34-1054/N

2016 Vol. 46, No. 2

Display Method:
The expected discounted penalty function under the compound Poisson risk model with tax payments and a threshold dividend strategy
WANG Wenyuan, LIU Zhang
2016, 46(2): 87-94. doi: 10.3969/j.issn.0253-2778.2016.02.001
The compound Poisson risk model was considered in which taxes were paid according to a loss-carry forward system and dividends were paid under a threshold strategy. For this model, the ruin quantities were discussed by defining an expected discounted penalty function at ruin and the analytical integro-differential equation satisfied by the expected discounted penalty function was derived. Finally, in the case where the individual claims follow an exponential distribution, explicit expressions for the ruin probability were given.
Asymptotical analysis of SEIR model with infectious force in latent and immune periods
MA Yanli, ZHANG Zhonghua
2016, 46(2): 95-103. doi: 10.3969/j.issn.0253-2778.2016.02.002
A type of SEIR epidemic model with infective force in the latent and immune period was studied. And the threshold, basic reproductive number R0 which determines whether a disease is extinct or not, was obtained. By using the Liapunov function method, it was proved that the disease-free equilibrium E0 is globally asymptotically stable and that the disease eventually goes away if R0<1. It was also proved that in the case where R0>1, E0 is unstable and the unique endemic equilibrium E* is locally asymptotically stable by Hurwitz criterion theory. It was shown that when disease-induced death rate and elimination rate are zero, the unique endemic equilibrium E* is globally asymptotically stable and the disease persists. Finally, numerical simulation was given to illustrate the theoretical analysis.
Research on modulo scheduling framework for clustered architecture
WANG Xiangqian, ZHENG Qilong, HONG Yi
2016, 46(2): 104-112. doi: 10.3969/j.issn.0253-2778.2016.02.003
An implementation method for modulo scheduling framework was established for clustered architecture. The description method for its machine resources was introduced. The relationship between module scheduling and loop unrolling was researched, and then the effective decision condition on loop unroll was offered. The modulo variable expansion algorithm framework for vectorization was proposed. The code generation schema based on speculation execution was described. The experiment result shows that the modulo algorithm framework could bring speed-up ratio to 170%~680%.
Recommender algorithms based on tendencies between users and items
LI Qiang, HE Xingsheng, FU Zhongqian
2016, 46(2): 113-119. doi: 10.3969/j.issn.0253-2778.2016.02.004
Recommender systems are one of the most effective technologies to help users filter the overload of information, and collaborative filtering (CF) is one of the most widely used techniques in recommender systems. However, CF algorithms have difficulty dealing with the problems such as the sparseness of data and the scalability for new users. As an alternative, an improved algorithm based on the preference (tendencies-based, TB) algorithm was proposed. In the proposed method: firstly, the user rating set was classified into different groups according to user rating preference and item rated tendencies; then the regression model was obtained by linear regression performed on each class. The improved model not only achieves higher accuracy in rating prediction on sparse datasets, but also greatly reduces computational complexity and space complexity. Through extensive experiments on three benchmark data sets, the results show that the improved approach increases recommendation accuracy by an average of 3.97% compared with TB algorithm.
Improving firefly algorithm by keeping individual activity
LU Kezhong, ZHANG Zheqing, SUN Jun
2016, 46(2): 120-129. doi: 10.3969/j.issn.0253-2778.2016.02.005
There are some disadvantages in the basic firefly algorithm(FA), such as low solving precision, premature convergence and etc. To overcome these disadvantages, a novel improved FA(IFA) was proposed that keeps individual activity. Firstly, an adaptive control for gamma value was designed by using swarm distance. Secondly, a position calculation for fireflies was updated by using the search process information. Thirdly, a special mutation for the firefly swarm was executed to activate individuals and to make them explore the search space when losing activity. Finally, a perturbation and local search method for the best individual was proposed. Based on ten multi-model test functions, the test results show that the IFA has a better convergence speed and precision than the basic FA, PSO, ABC and other improved FA. The improved FA is a good method for complex function optimization.
Extraction of instantaneous velocity in ultrasonicdoppler signal based on generalized S transform
WANG Yuanguo, YU Wenli, ZHENG Chichao, ZHAO Wei, PENG Hu
2016, 46(2): 130-137. doi: 10.3969/j.issn.0253-2778.2016.02.006
Extraction of instantaneous velocity in doppler echo signal is of great significance to the analysis of target motion state. However, general analytical methods can not provide sufficient time-frequency focusing performance due to the restriction of nonstationarity. A method based on generalized S transform (GST) was suggested to extract instantaneous velocity from the ultrasonic doppler signal, and computer simulation and experiment verification were carried out. In the simulation, a model system was set up, which contains a reciprocating dot. The doppler echo signal was produced by transmitting ultrasonic waves to the reciprocating dot. In the experiment a machinery device with reciprocating movements was designed, which gets echo signal by transmitting ultrasonic waves. The instantaneous frequency was obtained by processing the echo signal with GST and the instantaneous velocity was calculated. The computer simulation and experiment demonstrate the effectiveness of extraction of instantaneous velocity of the moving target in the doppler echo signal by GST.
Power allocation based on energy and spectral efficiency in multi-beam satellite systems
KAN Xi, XU Xiaodong
2016, 46(2): 138-147. doi: 10.3969/j.issn.0253-2778.2016.02.007
A new algorithm aiming to optimize the power allocation to each beam was proposed in the frame work of a multi-beam satellite communication system. The algorithm focuses on both energy efficiency (EE) and spectral efficiency (SE). Suppose a shadowed Rice model is satisfied by land mobile satellite channels. Under the constraint of total power and rate of system, a double-objective optimization problem was set up, which optimizes power allocation of beams to maximize both EE and SE. Then it was analyzed that under this frame EE is quasi-concave in SE. Based on this analysis and Lagrangian duality theory, Pareto optimal solution sets can be acquired. Finally, the trade-off analysis based on the Pareto front and scalarization of this bi-criterion problem resulted in a trade-off solution. In addition, the impact of system parameters such as elevation angle, static circuit power and number of beams on algorithm performance was discussed. The simulation results demonstrate the effectiveness of this power allocation scheme for achieving desirable EE and SE and confirm the validity of the conclusion.
The minimal spanning sets of constacyclic codes over polynomial residue rings
LI Hongju, DING Jian
2016, 46(2): 148-155. doi: 10.3969/j.issn.0253-2778.2016.02.008
The minimal spanning sets and ranks of (1+λu) constacyclic codes with an arbitrary length N=pen over R=Fpm[u]/〈uk〉 were investigated, where uk=0 and λ is a unit of R. Specifically, the case with k=2 and λ=1 was given, while a small mistake about the minimal spanning sets in [Abular T, Siap I. Constacyclic codes over F2+uF2. Journal of the Franklin Institute, 2009, 345: 520-529] was pointed out. Furthermore, based on the analysis of the equivalence between cyclic codes and constacyclic codes over R, the generator polynomials and minimal spanning sets of some other constacyclic codes over R were obtained. As a consequence, the generator polynomials and minimal spanning sets of (1+ζu2) constacyclic codes with odd length and length N≡2(mod 4) over F2m[u]/〈u3〉 were given for any ζ∈F*2m.
Multi-period mean-variance optimization with cardinality constraints
HAO Jing, ZHANG Peng
2016, 46(2): 156-164. doi: 10.3969/j.issn.0253-2778.2016.02.009
The multi-period mean-variance portfolio selection was presented by taking into account transaction cost, threshold constraints, borrowing constraints and cardinality constraints. Because of the transaction costs, the multi-period portfolio selection is the mix integer dynamic optimization problem with path dependence. The discrete approximate iteration method was designed to obtain the optimal portfolio strategy. Finally, the comparison analysis of the differently desired number of assets in the portfolio selection was provided by a numerical example to illustrate the efficiency of the proposed approaches and the designed algorithm.
Reliability analysis of group experts and testing of evaluation outcomes
ZHAO Yajuan
2016, 46(2): 165-172. doi: 10.3969/j.issn.0253-2778.2016.02.010
Information structural bias and cognitive structural bias and other issues lead to randomness and inconsistency of expert evaluations, calling into question the reliability of evaluation outcomes of expert groups. To solve this problem, psychological statistics reliability was used to measure expert reliability for individual experts and expert groups; then, two-factor fixed effects model was used to further analyze and test the sources of error. This can simultaneously analyze and evaluate the fairness and reasonableness of evaluation outcome and the reliability of experts. Finally, an empirical analysis was carried out with a tender evaluation case.