中国科学院数学与系统科学研究院期刊网

27 May 2025, Volume 45 Issue 5
    

  • Select all
    |
  • ZHANG Ziyang, SONG Jiantao, CHEN Shuangmin, XIN Shiqing, TU Changhe
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1339-1360. https://doi.org/10.12341/jssms240504
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    The Voronoi diagram (VD) is a tool used in computational geometry for spatial division and is widely applied across multiple fields. With advancements in scanning technology, research has expanded from within surfaces to point cloud surfaces. We utilize the dual relationship between VD and Delaunay triangulation (DT) to directly compute the VD restricted by point cloud surfaces. Candidate facets are generated through DT, their conformity with the point cloud surface is assessed, and a mixed-integer programming approach is employed to select the facets that satisfy manifold constraints and are most well-fitted. This method allows us to directly generate restricted VDs from the preserved triangle facets. This is the first study to use point cloud surfaces as constraints for VDs, and its effectiveness has been demonstrated through empirical validation.
  • TAN Yingying, XU Tongyou, KOU Feidan, LIU Song
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1361-1371. https://doi.org/10.12341/jssms240500
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    The least eigenvalue of the Laplacian matrix of a simple undirected graph is called the algebraic connectivity of the graph. For a first-order multi-agent system with an undirected graph as its communication topology, the larger the algebraic connectivity, the faster the consensus convergence rate of the system. In this paper, a graph operation of edge rewiring is used to optimize the undirected graph corresponding to the communication topology of a multi-agent system, so that the algebraic connectivity increases the most, and an algorithm is proposed to increase the algebraic connectivity of the undirected graph corresponding to the communication topology and reduce the communication volume. Simulation experiments on a system consisting of six multi-agents show that the algorithm can improve the speed of multi-agent system error approaching zero, accelerate consensus convergence rate of the system, and reduce the communication volume of the system by decreasing the communication times when the system reaches consistency.
  • XIE Jiacheng, XIONG Juxia, HE Zhenjiang
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1372-1385. https://doi.org/10.12341/jssms240495
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    Aimed at the problems of insufficient optimization performance and accuracy of SMA in solving wind farm layout optimization problem(WFLOP), and the slow convergence speed and premature convergence to local extreme values in SMA, an improved slime mold algorithm based on adaptive contraction and genetic learning strategy is proposed. First, a wind farm layout model is initially established based on the specific environmental conditions. Then, for the problem of premature convergence to local extreme values, a genetic learning strategy is introduced to enhance the convergence speed and global search ability of SMA, resulting in the GLSMA. Finally, aimed at the problems of WFLOP, the maximum rule coding solution vector is adopted, and an adaptive contraction strategy is designed to update the position of slime moulds using the power generation of wind turbines, which improving the solution accuracy. The experimental results show compared to SMA, grey wolf optimization(GWO), salp swarm algorithm(SSA), whale optimization algorithm (WOA), and genetic learning particle swarm optimization(GLPSO), GLSMA has faster convergence speed and higher optimization accuracy in 19 test functions, and the A-GLSMA has higher performance than genetic algorithm(GA) in solving WFLOP under two wind direction distributions.
  • QIN Xiaolin, LIU Yunhao, DENG Lihua, LI Fei
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1386-1399. https://doi.org/10.12341/jssms240136
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    Mathematical human-like answering constitutes a vital component of automated reasoning and has long been a focal point in cognitive intelligence research, drawing extensive attention from scholars, which requires simulating human understanding, representation, and reasoning of mathematical knowledge, where knowledge representation serves as the foundation for semantic comprehension and knowledge inference. Knowledge graphs are effective tools widely cited in fields such as knowledge representation and the construction of knowledge systems. Addressing the logical association challenges in mathematical knowledge representation, this paper proposes a method for constructing a mathematical knowledge graph. By interpreting mathematical predicates and objects as relations and entities, respectively, and employing rule instantiation, the paper unifies the representation of question and rule knowledge through knowledge graph. Single-step reasoning is achieved through structural matching based on subgraph isomorphism, proving effective in the automated solving of mathematical questions devoid of complex expressions, facilitating the generation of human-like answering processes. Experimental results demonstrate that the proposed mathematical knowledge graph method can yield correct and human-like answering styles.
  • LUO Song, CAO Yanhua
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1400-1412. https://doi.org/10.12341/jssms23550
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    Traditional neural networks process real-valued inputs and outputs through the manipulation of neuron weights. In order to investigate the impact of introducing complex numbers into neural networks, this study employs two deep learning methods to solve the time-dependent Schrödinger equation. The physics informed neural network (PINN) focuses on incorporating physical equations and boundary conditions as constraints into the training process of neural networks, making them more in line with physical laws. On the other hand, the deep Galerkin method (DGM) utilizes the nonlinear fitting capability of neural networks to minimize residuals and approximate the true analytic solution. Numerical experimental results indicate that whether complex numbers are included or excluded from the neural networks has no substantial impact on the resulting numerical solutions. Complex operations can be re-expressed using real-valued tensors. Therefore, these two deep learning methods for solving the time-dependent Schrödinger equation are feasible, greatly simplifying the solution process while avoiding grid-related limitations. The high-precision approximation demonstrated by neural networks in numerical computation is not only simple and easy to implement, but also possesses strong parallel computing capabilities.
  • WU Tingting, GAO Jian
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1413-1421. https://doi.org/10.12341/jssms240369
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    Cyclic codes are an important subclass of linear codes, which have wide applications in data storage systems, communication systems due to their efficient encoding and decoding algorithms. In this paper, quinary cyclic codes $\mathcal{C}_{(1,e,s)}$ with three nonzeros $\alpha$, $\alpha^e$, $\alpha^s$ are discussed, where $\alpha$ is the primitive element of $\mathbb{F}_{5^m}$, $m$ is a positive integer, $s=\frac{5^m-1}{2}$, and $2\leq e\leq 5^m-2$. Firstly, the paper presents the necessary and sufficient condition for the quinary cyclic codes $\mathcal{C}_{(1,e,s)}$ to be optimal when $e=5^h-2$, where $1\leq h\leq m$. Furthermore, based on the proposed necessary and sufficient condition, by analyzing the irreducible factors of certain polynomials, it is proven that when $m$ is an odd integer no less than $3$, and $h$ takes values of $2,~m-2,~m-1$, respectively, the cyclic codes $\mathcal{C}_{(1,e,s)}$ are optimal with parameters $[5^m-1,5^m-2m-2,4]$. Secondly, when $e=4(5^h+1)$, where $0\leq h\leq m-1$, by analyzing the solutions of certain equations, it is demonstrated that when $m$ is an odd integer no less than $3$ and $h$ is $0$, the cyclic code $\mathcal{C}_{(1,e,s)}$ is optimal with parameters $[5^m-1,5^m-2m-2,4]$. This paper has made some progress in solving the two open problems proposed in reference(Wu, et al., 2023).
  • XU Meng, MAO Wei, SHEPHERD Simon, HARRISON Gillian
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1422-1437. https://doi.org/10.12341/jssms23654
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    From the perspective of autonomous vehicles (AV) market development, this study presents a dynamic analysis involving the market development characteristics of private vehicle and shared AV. With the consideration of current development situation of AV, a system dynamics model for the market development of AV is proposed, which is based on a series of indexes including trust, AV purchase willingness, using willingness and shared AV service accessibility, etc. Based on the passengers' vehicles situation in Beijing, the vehicles development scenario from 2014 to 2050 is developed, the variant numbers of AVs from 2024 to 2050 are modeled and its market share is further estimated. The results show that the number of AVs will exceed the number of conventional vehicles in 2040, and the market share of AVs will be 90% by 2050. Further, the study develops scenarios of AV involving technology development and shared AV rate promotion. The results show that the rapid development of AV technology will attract more residents to use private AV, and the promotion of shared AVs can effectively reduce the purchase of private AVs. This study provides reference for the future market development of AV.
  • FANG Xin, ZHANG Chengyuan, CHAI Jian, WANG Shouyang
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1438-1454. https://doi.org/10.12341/jssms250013
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    The volatility and nonlinear characteristics of time series have made modeling and prediction difficult and have attracted widespread attention from scholars. This study combines the decomposition and integration framework to achieve effective information extraction and modeling to improve prediction accuracy. Correspondingly, our proposed methodology involves four main steps: Data decomposition via complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN); Component grouping via sample entropy (SE); Prediction of the reorganized low-, mid-, and high-frequency component groups through persistence (PER), convolutional neural networks (CNN), gated recurrent unit (GRU), and ensemble prediction via weighted addition using ant lion optimization (ALO). Taking the hourly PM2.5 concentration of Xi'an as the sample, experimental results showed that our proposed hybrid decomposition-group-ensemble forecasting framework (i.e., ALO-CEEMDAN-SE-(PER-CNN-GRU)) significantly outperformed the benchmarks, and the final prediction error obtained the lowest value (2.53%). This validates the superiority of the decomposition integration framework with excellent neural network models for PM2.5 prediction.
  • GUO Wenqiang, CHEN Siqi, LEI Ming, LIANG Yunze, GAO Yaqi
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1455-1470. https://doi.org/10.12341/jssms240290
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    In examining the evolutionary processes of cooperative behavior between enterprises and low-carbon service providers in establishing supply chain alliances, this study employs evolutionary game theory and catastrophe theory to transform the traditional game's replication dynamic equation into a cusp catastrophe model. Additionally, it establishes stochastic dynamics that incorporate Gaussian white noise. The system introduces an elasticity measurement index to quantitatively assess the degree to which the system absorbs disturbances. Simulation experiments further analyze the impact of relevant parameter changes on the nonlinear evolution and elasticity of the alliance. The results indicate that when the game parameter combination lies within the mutation set, a bimodal phenomenon and disturbing mutation occur. Conversely, when the game parameter combination crosses the boundary of the mutation set, a structural mutation arises within the alliance state. Furthermore, when excess income and punishment intensity surpasses a certain threshold, they positively influence system elasticity. However, alliance member synergy negatively affects system elasticity up to a certain threshold; Beyond this point, an increase in synergy leads to a decrease in elasticity. This change can prompt alliance members to transition from a "not participating" strategy to a `participation' strategy, ultimately stabilizing at the "participation" strategy.
  • WANG Yuyan, DING Luping, HUO Baofeng
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1471-1493. https://doi.org/10.12341/jssms240085
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    Enterprises must choose a live streaming method that matches their own development to make profits through live streaming sales. This paper considers three live streaming methods: Manufacturer self-broadcasting, entrusted internet celebrity live streaming, and self-broadcasting + internet celebrity live streaming. Based on game theory, a live streaming e-commerce supply chain model is constructed to study the streamer's ability to sell products and fans effects, the impact of supply chain members' decisions and the best way for manufacturers to carry live sales. The research found that: 1) A streamer's improvement in product delivery ability will help increase product prices and the streamer's effort level; The stronger the fans effect of Internet celebrities, the higher the product price and product sales. 2) The price of products in the internet celebrity's live broadcast room is not always lower than the price of the manufacturer's self-streaming. The relationship is related to the internet celebrity's ability to sell products. 3) Self-broadcasting + internet celebrity live broadcasting is the most beneficial way for manufacturers to make profits and expand market share. The conclusions of this paper can help members of the live broadcast e-commerce supply chain make reasonable decisions and help enterprises cooperate better.
  • YUAN Ruiping, ZENG Wang, YANG Yang, LI Juntao, LIANG Kaibo
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1494-1507. https://doi.org/10.12341/jssms240576
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    Online freight platforms face challenges such as fierce price competition due to homogeneous pricing strategies. This paper addresses this issue by developing a differentiated pricing model based on two-sided market theory. The model considers the interplay between platform technology level and user quality level. The study examines three user affiliation structures: Single-homing, multi-homing, and both sides of users are partially multi-homing. It investigates the impact of technology and user quality on the platform's equilibrium pricing and profits using analytical methods and simulations. When both sides of users are single-homing, a two-part pricing strategy is more advantageous, with profits increasing with better technology and user quality. Under multi-homing, the optimal pricing strategy depends on an equilibrium condition. Higher user quality favors two-part pricing, while higher technology level favors registration fee-only pricing. When both sides of users are partially multi-homing, registration fee-only pricing is more optimal. Both technology and user quality positively impact profits, with technology playing a bigger role. The research provides insights for online freight platforms to implement differentiated pricing based on their market position.
  • ZHANG Yaojia, GONG Zaiwu, LI Ming
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1508-1523. https://doi.org/10.12341/jssms240168
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    The political situation along the "the Belt and Road" is changeable, the economic development is unbalanced, natural disasters occur frequently, and the geo environment is complex. The geo risk assessment is the premise and foundation to promote the construction of the "the Belt and Road". In response to the problems and difficulties of incomplete information and uncertain knowledge in the risk assessment of China-Myanmar geopolitical relationship and investment security, this paper comprehensively adopts the technical approach of cross fusion of uncertainty intelligent assessment methods such as Bayesian networks, cloud models, and DS evidence theory to construct a data and knowledge driven China-Myanmar geopolitical relationship and investment security risk assessment model, and carries out risk assessment and scenario situation deduction, Intended to provide technical support and policy recommendations for overseas investment security risk warning.
  • ZHANG Shaojun, JIANG Rui, GAO Wentao, YANG Zejiang
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1524-1542. https://doi.org/10.12341/jssms23739
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    In order to promote the high-level opening up of the insurance industry and build a fair market competition environment, the CBRC revised and issued the 《Detailed Rules for the Implementation of the Regulations of the People's Republic of China on the Administration of Foreign funded Insurance Companies》 in 2019, unifying some of the regulatory policies for Chinese and foreign investment. Under the macro background of both opening up and strengthening supervision, this paper, based on the natural experiment of the 《Detailed Rules for the Implementation of the Regulations of the People's Republic of China on the Administration of Foreign funded Insurance Companies》, analyzes the changes of life insurance companies' business models and their impact mechanisms from the perspective of unified supervision by using micro data of life insurance companies. The research results indicate that: 1) The unification of regulatory policies has effectively increased the operational stability of foreign life insurance companies, making them more inclined towards debt driven business models. 2) The unification of regulatory policies enhances the operational stability of life insurance companies by increasing their liquidity creation level. 3) Compared to smaller and tightly regulated life insurance companies, the unified regulatory policy has a smaller impact on larger and more lenient life insurance companies. From the perspective of regulatory unity, this study provides policy implications for further expanding the high-level institutional opening of the insurance market, preventing regulatory arbitrage, and promoting insurance companies to return to security to effectively prevent systematic risk.
  • LI Delong, CHENG Yu, CHAI Ruirui, SHI Zhihong, WANG Tianhua
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1543-1565. https://doi.org/10.12341/jssms23612
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    The face recognition system plays a sentry and signal deterrent role in the pre-defense against terrorism and riot in the subway. However, considering the imbalance in the allocation of security resources at subway stations, the effectiveness of collaborative configuration among multiple stations still needs to be further enriched. This paper uses a face recognition system as a front-end port for identifying terrorists and a signal carrier for anti-terrorism and riots prevention. A subway anti-terrorism and riot prevention signal game model based on the emotions of terrorists and limited attack resources is constructed, and the impact of the relative cost coefficients and emotional parameters of terrorist dispatch on relevant thresholds is studied through numerical analysis. It is found that, firstly, there is no separate equilibrium path in which both types of security departments adopt camouflage strategies. Emotional parameters affect the posterior probability boundary of violent terrorists towards vulnerable security departments under different signals in the mixed equilibrium path, and the camouflage cost of security departments determine their own signal strategy. Secondly, when the cost of opening a face recognition system exceeds a certain threshold, the benefits of the security departments are positively correlated with the proportion of vulnerable security departments naturally endowed, and at the same time, this threshold is also positively correlated with the total amount of terrorist attack resources. Thirdly, the higher the recognition rate of a face recognition system for highly recognizable terrorists and the lower the recognition rate for low recognizable terrorists, the less sensitive the impact of the relative cost coefficient on the posterior probability decision threshold, and the greater the sensitivity of the probability judgment scale on the impact of the relative cost coefficient decision threshold.
  • MA Qiang, GAO Ya, WANG Hong, HAN Haitao
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1566-1587. https://doi.org/10.12341/jssms240020
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    Since the launch of a new round of power system reform in China in 2015, the establishment of a single-track-based electricity spot market has gradually become the focus of attention in domestic electricity markets. However, up to now, a mature electricity price forecasting model has not been established in the unified spot market, and power generation and sales companies, power trading centers, and power users cannot make full use of electricity price forecasting data for auxiliary decision-making to obtain the best benefits. Therefore, this paper proposes an electricity price forecasting model based on electricity price formation mechanism and XGBoost algorithm. Firstly, according to the marginal clearing price formation mechanism and unique bidding rules adopted in the unified spot market, the unified cumulative bidding curve of the whole network is fitted by piecewise function, and the unified clearing price prediction model of the whole network is established by combining the bidding strategy of power generation enterprises. Secondly, according to the relevant data published in the unified power spot market, the XGBoost algorithm is used to select features and solve the different daily ladder bidding strategies of power generation enterprises. Finally, the hyper-parameters of the model are optimized by the highly automated Optuna algorithm. The experimental results show that the electricity price prediction model in this paper has stronger interpretability and accuracy than the XGBoost algorithm directly substituted into the data, and proves that the XGBoost algorithm has higher prediction accuracy for the bidding strategy than the gradient boosting regression tree algorithm and the random forest algorithm, thus verifying the superiority and effectiveness of the model in the electricity price prediction of the unified power spot market.
  • WANG Lu, GUO Jixing
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1588-1606. https://doi.org/10.12341/jssms240047
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    Considering the differences in the credit policies of the automobile industry in China and the United States, the Cournot game models of duopoly automobile manufacturers are constructed in three scenarios: Corporate average fuel economy regulation, the separate management regulation of corporate average fuel economy and new energy vehicle credit, and dual-credit policy. The optimal decisions under different situations are solved and compared, and the following conclusions are obtained. The effects of the separate credit management regulation in promoting the increase of new energy vehicle output, restraining the production of fuel vehicles and increasing the total output value are stable and not affected by policy parameters, while the effects of the dual-credit policy can only be released when the credit trading price is higher than a certain threshold. Through the adjustment of the price signal of the credits, the dual-credit policy can achieve better effects than the separate credit management regulation. Therefore, the dual-credit policy shows higher flexibility and adaptability. The implementation of both policies contributes to the reduction in the price of new energy vehicles, thereby exerting a guiding effect on consumers' environmentally-friendly purchasing behavior from the perspective of demand. The impact of policies on different market participants varies. Implementing the separate credit management regulation is beneficial for new energy vehicle manufacturers, while the dual-credit policy provides stronger incentives for traditional manufacturers when the credit price reaches a certain level. If certain accounting discount multiples are given to new energy vehicles, it is easier for the dual-credit policy to exceed the "valley of death" of the lowest credit trading price. The research presented in this paper is instrumental in conducting a comprehensive analysis of the micro and macro mechanisms underlying policy actions within the automobile industry, thereby offering valuable insights for optimizing and enhancing the efficacy of the dual-credit policy.
  • GUO Xiaole, RAN Bo
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1607-1618. https://doi.org/10.12341/jssms240855
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    This paper deals with robust optimality condition and duality theory for a class of minimax fractional semi-infinite optimization problems with uncertain data. By virtue of robust optimization, Dinkelbach technique, and robust type constraint qualification conditions, we first establish robust optimality conditions for this uncertain optimization problem. Then, we introduce a mixed type robust dual problem for this uncertain optimization problem, and explore robust duality properties between them. As a special case, we investigate robust optimality conditions and sum of squares relaxation properties for minimax fractional semi-infinite optimization problems with sum of squares convex polynomial structures.
  • GUO Feng, HE Liang, SUN Xiangkai
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1619-1628. https://doi.org/10.12341/jssms240783
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    This paper deals with a Tikhonov regularized primal-dual dynamical system featuring variable mass for solving convex optimization problems with linear equality constraints. Under suitable assumptions and through the use of energy functions, the convergence rates are first established for the primal-dual gap, the residual of the objective function, the feasibility measure, the velocity vector, and the gradient norm of the objective function along the trajectories. Then, the strong convergence of the primal trajectory of the dynamical system towards the minimal norm solution of the linear equality constrained convex optimization is demonstrated. Moreover, numerical experiments are conducted to illustrate the obtained results.
  • GAO Jianqing, CHEN Liting, ZHENG Jing
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1629-1642. https://doi.org/10.12341/jssms23780
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    The existing models for predicting the economic loss index of typhoon disasters are mostly based on the prediction of evaluation data, neglecting the application of historical data. The existing case-based reasoning(CBR) research on typhoon disasters rarely considers the scenario of case adaptation, and compared to the most similar historical case, the adjusted scheme can improve the accuracy of decision-making. Therefore, the case adaptation is introduced to improve the prediction of typhoon disaster economic loss index. The dynamic time warping algorithm and K-nearest neighbor algorithm are applied to construct the suggested solution for the target typhoon. The attribute difference revision(ADR) method is innovatively introduced to obtain the corrected value of the target typhoon. Furthermore, the suggested solution and corrected value are aggregated to the predicted value. Through comparative analysis and empirical research with other prediction models, it was found that the prediction model.
  • XU Chen
    Journal of Systems Science and Mathematical Sciences. 2025, 45(5): 1643-1650. https://doi.org/10.12341/jssms23727
    Abstract & Html ( ) Download PDF ( )   Knowledge map   Save
    This work studies two-agent multitasking scheduling on a single machine with job rejection, where two agents can't interrupt each other. The model with the objective of the makespan is considered. An upper bound on the total permitted rejection cost is assumed, and there are two options for each job: Acception or rejection. Since the problem is NP hard, we pay attention to providing pseudo polynomial dynamic programming algorithm, $n_A^2$-approximation algorithm and the fully polynomial-time approximation scheme.