• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    Percolation transitions in edge-coupled interdependent networks with directed dependency links

    2023-10-11 07:56:42YanLiGao高彥麗HaiBoYu于海波JieZhou周杰YinZuoZhou周銀座andShiMingChen陳世明
    Chinese Physics B 2023年9期
    關(guān)鍵詞:周杰銀座

    Yan-Li Gao(高彥麗), Hai-Bo Yu(于海波), Jie Zhou(周杰), Yin-Zuo Zhou(周銀座), and Shi-Ming Chen(陳世明),?

    1School of Electrical and Automation Engineering,East China JiaoTong University,Nanchang 330013,China

    2School of Physics and Electronic Science,East China Normal University,Shanghai 200241,China

    3Alibaba Research Center for Complexity Sciences,Hangzhou Normal University,Hangzhou 311121,China

    Keywords: edge-coupled interdependent networks with directed dependency links, percolation transitions,cascading failures,robustness analysis

    1.Introduction

    Many facilities and systems in real life can be abstracted into networks, such as transportation networks, biological networks, financial networks, electric networks, and social networks.[1–9]These networks have brought convenience to human production and life,while the propagation phenomena occurring on the networks have also brought negative effects,and complex networks science provides theories and analytical methods for studying complex systems in the real world.

    Early research on complex networks was limited to the topological properties of the single-layer network,[10–13]until it was discovered that the interdependence within networks in the real world posed extremely serious systemic risks,such as large-scale power outages and traffic congestion.[14,15]This is due to the fact that the failures in a single-layer network could spread not only in itself but also in the other networks.Therefore,the study of interdependent networks has attracted much attention of many scholars.

    Buldyrevet al.[16]first developed the analytical framework of interdependent networks,where there are one-to-one dependent nodes between two networks, and found that interdependence greatly reduces the robustness of the network.Following this research, many results have emerged for the study of interdependent networks.According to the difference of interdependent patterns, Parshaniet al.[17]analyzed the node-coupled interdependent networks in which a fraction of the nodes are interdependent in both networks, and found that reducing the coupling between the networks leads to a change from a first-order percolation phase transition to a second-order percolation transition at a critical point, i.e., its robustness is better than the fully node-coupled interdependent networks.[16]Shaoet al.[18]proposed the coupled network systems with multiple support-dependence relations and investigated them analytically and numerically.Buldyrevet al.[19]and Parshaniet al.[20]studied the coupled networks with positively correlated degrees of mutually dependent nodes and found that as the networks become more inter-similar the system would become significantly more robust to random failure.Huet al.[21]found that the phase transition includes the hybrid transition of mixed first and second orders in study of a coupled networks system where both interdependent and interconnected links exist.This hybrid transition phenomenon was also found in the studies of Zhouet al.[22]and Zhanget al.[23]afterward.Donget al.[24]developed a interacting networks model with feedback-dependency links and found that the feedback-dependency links have no effect on robustness of the system for a weak coupling strength corresponding to the second-order transitions, whereas the system is more vulnerable for strong coupling strength corresponding to first-order or hybrid phase transition.Liuet al.[25]found that reducing the coupling strength between interdependent nodes can improve the system’s robustness by studying the interdependent networks with weak node-coupling strength.

    In addition to studying interdependent patterns, scholars have proposed numerous network models.Gaoet al.[26–28]extended interdependent networks to systems with more than two independent networks and analyzed the effect of topology and the number of networks on network robustness.Cellaiet al.[29]showed that the overlap among layers can improve the robustness of interdependent multiplex systems and change the critical behavior of the percolation phase transition in a complex way.Huanget al.[30]studied the effect of clustering on the system robustness of interdependent networks and found that clustering makes the system more vulnerable.Valdet al.[31]set the nodes with high degrees as autonomous nodes and found them to be effective in improving the robustness of the network.Liuet al.[32]developed a theoretical framework of interdependent directed networks and found that the directionality within the network increases its vulnerability and exhibits hybrid phase transitions.

    The above works address the case of node-coupled interdependent networks (NINs), i.e., the failures of nodes in one network lead to the failures of nodes in the other network.Considering that there are also many cases of edge-coupled interdependent networks in the real world, Gaoet al.[33]developed a new model of edge-coupled interdependent networks(EINs) and analyzed it using the self-consistent probabilities method[34]and found that it is stronger than the NIN.

    The existing studies of EINs have only considered the case where both connectivity links and dependency links are undirected.However, there are also some EINs in which the dependency links are directed in the real world.One example is power-transportation interdependent networks.Generally speaking, power system and rail transportation system interdepend on each other.However,by analyzing the interactions between these two networks we found that there are one-way dependencies instead of mutual dependencies for the specific lines.For example,the failures of part power lines can directly affect the functions of some part of rail transportation lines,whereas these transportation lines may not just right affect these power lines.In other words, there are directed dependency relations between transportation lines and power lines.

    Therefore, a model of edge-coupled interdependent networks with directed dependency links (EINDDLs) is proposed,and the theoretical framework for the analysis of EINDDLs under random attacks is developed by settingβto change the dependence relationship between two networks based on the self-consistent probabilities method.Both theoretical analysis and computer simulations are conducted on RR-RR and ER-ER EINDDLs(RR:random regular,ER:Erd¨os–Rényi)to explore the relationship betweenβand the phase transition behaviors of networks A and B under random attacks.Some characteristics in the phase transition behaviors of networks A and B are found and the corresponding prevention strategy is proposed.

    2.Model description

    The model of edge-coupled interdependent networks with directed dependency links (EINDDLs) in this paper refers to two-layer edge-coupled interdependent networks(EINs)consisting of networks A and B,where the dependency links are one direction instead of mutual direction.We assume that the number of nodes and the average degree of both networks are the same, so the total numbers of edges in both networks are equal.A connectivity link in network A is randomly selected to establish a directed dependency link with a connectivity link in network B, as shown by the red and blue dashed lines in Fig.1(a).Figure 1 gives the process of cascading failures for the EINDDL.Dependency links make the system follow such cascading rules,e.g.,failure ofa1?a8causes failure ofb2?b3while the failure ofb2?b3cannot lead to failure ofa1?a8.Here,βis defined to control the ratio of two directed types of dependency links, i.e., network A has a 1-βratio of connectivity links depending on network B,and network B has aβratio of connectivity links depending on network A.The dependencies of the two networks will also be changed by changing the value ofβ.

    Fig.1.Cascading failures of the EINDDL.

    According to the bond percolation theory, when a network is under attacks, it may be broken into many connected clusters.The biggest one referred as giant component(GC)is usually considered as functional part of a network.We randomly remove the 1-pfraction of connectivity links in network A in the initial state,this will result in a portion of nodes failing due to being out of GC, while the edges connected to these failed nodes will also fail.Since there are connectivity links in network B that depends on network A,it causes a part of the connectivity links in network B to fail and makes a part of the nodes fail by leaving the GC,which subsequently leads to the failure of connectivity links in network A that depends on network B and causes further damage to network A.Cascading failures repeatedly propagate between two networks as the above process until no more new connectivity links or nodes fail in both networks,which is the steady state.In terms of interdependent networks,the remainder of the network after the cascading failure composes an extensive component, referred as mutually connected giant component(MCGC).The size of the MCGC at a steady state reflects the robustness of the network.

    The solid black lines in Fig.1 represent the connectivity links within the network,the dashed black lines represent the failed connectivity links,and the dashed red and blue lines are the dependency links between two networks, whereβis 0.6.Due to the initial statea3?a4anda4?a5in network A failed,as shown in Fig.1(a).According to the cascading failure rules,b5?b6in network B will not fail andb6?b7will fail.Finally,a steady state is reached,as shown in Fig.1(b).

    3.Theoretical analysis framework

    3.1.Single-layer network

    The degree distributionp(k)is defined as the probability that a node of degreekis randomly selected in the network.The condition that a node of degreekdoes not belong to the GC is that allkconnectivity links of the node do not lead to the GC.We introduce an auxiliary parameterR,defined as the probability that a randomly selected link leads to the GC along one direction, then the probability that the node does not belong to the GC can be expressed as(1-R)k.For any node with degreeksatisfying the degree distributionp(k),the probability that a node does not belong to the GC is obtained by multiplying the probability (1-R)kwith the degree distributionp(k)and summing up,[34–38]whereSdenotes the probability that a randomly selected node in the network belongs to the GC,

    From the generating function theory,[36,37]it is known that the network degree distribution generating function isΦ0(x) = ∑k p(k)xkand the excess-degree distribution generating function isΦ1(x) = ∑k q(k)xk-1, whereq(k) denotes the excess-degree distribution of the nodes, i.e.,q(k)=p(k)·k/〈k〉,and〈k〉is the average degree of the network,thus Eq.(1)can be rewritten as

    Apart from Eq.(2),in order to solveSorR,it is also necessary to build a consistent equation of the probabilityRthat a connectivity link leads to the GC,i.e.,at least one of the remaining links of the node reached by this link leads to the GC,i.e.,

    Thus Eq.(3)can reduce to

    Using the generating functionΦ1(x), Eq.(4) can be rewritten as

    If a fraction 1-pof connectivity links is randomly removed from the network (bond-percolation), the probability of randomly selecting a node in the network that belongs to the GC remains unchanged since no node is removed from the network,and the expression of the order parameterSremains as Eq.(2).Since only a fractionpof connectivity links remains in the network,the probability of randomly selecting a link to reach a node with degreekbecomesptimes the original one, i.e.,p·q(k).Thus, the consistent equation for the auxiliary parameterRis

    The order parameterSand the auxiliary parameterRcan be solved by combining Eqs.(2)and(6).

    3.2.Edge-coupled interdependent networks with directed dependency links

    Following the single-layer network method,the EINDDL is a two-layer interdependent networks, where we introduce two auxiliary parametersRAandRB,which denote the probabilities that randomly selected link in network A and network B leads to the MCGC along one direction, respectively.Network A has 1-βfraction of connectivity links depending on network B,and network B hasβfraction of connectivity links depending on network A.For simplicity, assuming that both networks have the same degree distribution and the same number of nodes,their generating function expressions are also the same.In this case, when a connectivity link is randomly selected in network A, whether the node reached by this link belongs to MCGC depends on the following two factors.

    (i) If a randomly selected connectivity link does not depend on network B,i.e., the corresponding link in network B will not cause this link to fail.The fraction of this part of connectivity links isβ.It is only necessary to consider whether the end nodes of these links in network A belong to the GC.

    (ii) If a randomly selected connectivity link depends on network B, i.e., the failure of the corresponding connectivity link in network B will cause this link to fail.The ratio of this part of connectivity links is 1-β.It is necessary to consider the probability that the end nodes of this part of links in network A belong to GC, but also the probability that the end nodes of the corresponding links in network B belong to GC.

    Therefore, based on Eq.(5), the probabilityRAAthat a fraction ofβconnectivity links leads to the GC is

    Since a 1-βfraction of connectivity links depends on network B,at least one of the nodes to which the corresponding connectivity links the network B lead must be in the GC,as shown in Fig.2.

    Fig.2.The survival of connectivity link dependent on network B.

    The probability that one end ofg′not in the GC isΦ1(1-RB) and the probability that both ends are not in the GC is [Φ1(1-RB)]2.Thus the probabilityRLBthat at least one end of the connectivity linkg′in the GC is

    Similarly, the node to which the 1-βratio of connectivity links leads has to be in the GC of network A, and its probabilityRABis known according to Eq.(5),

    Accordingly, the probability that the nodes to which the connectivity links of the 1-βfraction lead are in the MCGC has to satisfy the conditions in both networks A and B, i.e.,RAB·RLB.

    Combining the two aspects considered above, the auxiliary parameterRAcan be expressed as

    Similarly, whether the nodes reached by a randomly selected connectivity link in network B belong to the MCGC requires consideration of the above two aspects.Using the same analysis,we can obtain that theβfraction of connectivity links in network B leading to the GC needs to satisfy both constraints in networks A and B,and the probability is

    The 1-βratio of connectivity links in network B leading to the GC only needs to meet the constraint in network B leading to the GC,and the probability is

    Thus, the expression for the auxiliary parameterRBcan be expressed as

    Following the above formulas,if the 1-pfraction of the connectivity links in network A are randomly removed in the initial state,the fraction of remaining connectivity links in network A isp,and the connectivity links ofβand 1-βfractions becomep·βandp·(1-β),respectively.From Eq.(10),the expression of the auxiliary parameterRAat this time is

    Since there are no attacks in network B,what affects network B is the failed links in theβscaled connectivity links.Therefore, the surviving connectivity links in network B are only related to the connectivity links of the 1-βscale and the connectivity links of thep·βscale.From Eq.(13), the expression of the auxiliary parameterRBis

    In addition,after randomly removing the 1-pfraction of connectivity links in network A,the order parameterSwill not be affected by the deletion of connectivity links because the nodes between the two layers are not directly related.From Eq.(2),the expressions for the order parametersSAandSBof networks A and B are,respectively,

    Given the values ofβandp,combing Eqs.(14)–(17),we can solve the values of the auxiliary parametersRA,RBand the order parametersSAandSB.

    In the percolation theory,[16,35]a fraction 1-pof nodes or links are randomly removed which may trigger cascading failures due to the dependency between networks.An abrupt collapse may occur where the size of GC changes to zero discontinuously upon a removal of a critical fraction 1-pcof nodes/links.This discontinuous percolation transition is named as the fist-order phase transition andpcis known as the percolation threshold.If the size of GC changes to zero continuously that is typical on single-layer networks, we call it the second-order phase transition.

    The type of phase transition and the phase transition thresholdpcof the EINDDL can be investigated by comparingSandp.The percolation behaviors of the networks vary with different values of the parameterβ.If the network undergoes a first-order phase transition,the curve of Eq.(14)tangentially will meet the curve represented by Eq.(15).If the network undergoes a second-order phase transition,the value of the auxiliary parameterRAorRBwill be 0,and the second-order phase transition threshold can be found by the above equations.

    3.3.Theoretical framework verification

    Figure 3 gives the plots ofSAandSBversuspfor the RRRR EINDDL and the ER-ER EINDDL with differentβfor the number of nodesNA=NB=100000 and the average degree〈kA〉=〈kB〉=4,where the solid lines are the theoretical results and the dots are the simulated values.

    Fig.3.Plots of SA and SB versus p with different β.

    It can be seen from Fig.3 that the theoretical values can match well with the computer simulation values.The theoretical framework can be verified.We apply this framework on the RR and ER networks to analyze their phase transition behaviors.

    4.Robustness analysis of the EINDDL

    4.1.The RR-RR EINDDL

    In the RR network, all node degrees are〈k〉.Therefore,the degree distribution is

    Assuming that the numbers of nodes of networks A and B are the same and the average degree is〈k〉, the generating function of the degree distribution of the RR network isΦ0(x)=x〈k〉and the generating function of the excess degree distribution isΦ1(x)=x〈k〉-1from the theory of generating functions.According to Eqs.(14)and(15),the expressions for the auxiliary parametersRAandRBof the RR-RR EINDDL are

    From Eqs.(16)and(17),the expressions of the order parametersSAandSBare

    From Eqs.(19)–(22), the values of the order parametersSAandSBare determined by the auxiliary parametersRAandRB, and only non-zero values ofRAandRBcan be obtained for non-zero values ofSAandSB.Obviously, Eqs.(19) and(20)possess a trivial solutionRA=RB=0,corresponding to the case in which the GC is 0 in the network.

    The first-order phase transition thresholdof the system can be obtained when the curves represented by Eqs.(19)and(20)are tangent.The GC of network A in the critical case will be 0 if the system undergoes a second-order phase transition.At this time, the auxiliary parameterRA→0.The nontrivial solutionRB0of the auxiliary parameterRBcan be found by substituting it into Eq.(20), and then the second-order phase transition thresholdpIIcis obtained as follows:

    Assuming that the number of nodes in both network A and network B in the RR-RR EINDDL isN=100000 and the average degree is the same,Fig.4 gives a plot of the order parametersSAandSBversuspat differentβfor〈k〉=3,4,and 5.

    4.2.The ER-ER EINDDL

    The degree distribution of a Erd¨os–Rényi (ER) network is Poisson distribution, i.e.,Similarly, assuming that the number of nodes of networks A and B are the same and the average degree is〈k〉,the generation function of the degree distribution of the ER network and the generation function of the excess degree distribution are the same from the theory of generating functions, that is,Φ0(x)=Φ1(x)=e〈k〉(x-1).According to Eqs.(14)and(15),the expressions for the auxiliary parametersRAandRBare,respectively,

    From Eqs.(16)and(17),the expressions of the order parametersSAandSBare

    Given the values ofβand the average degree〈k〉,Eqs.(24)and(25)can be rewritten as

    Eliminating the auxiliary parameterRBfrom the system of Eq.(28),we obtain

    When Eq.(29)satisfies the following conditions

    Equations (24) and (25) appear as a nontrivial solution,corresponding to the critical case when the system undergoes a first-order phase transition, wherepis the first-order phase transition threshold.

    If the system undergoes a second-order phase transition, the same analytical method of substitutingRA→0 into Eq.(25) can give the solutionRB0of the auxiliary parameterRB,which leads to the second-order phase transition threshold,

    Figure 5 gives a plot of the order parametersSAandSBversuspat differentβfor〈k〉=3,4 and 5.

    Fig.4.Plots of SA and SB versus p for the RR-RR EINDDL at different β and〈k〉.

    4.3.Robustness analysis

    4.3.1.Robustness analysis of network A

    It can be seen that, no matter whether it is the RR-RR EINDDL or the ER-ER EINDDL, network A gradually converges to the first-order phase transition from the second-order phase transition asβincreases,as seen in Figs.4 and 5.Network A will change from the first-order to the hybrid phase transition and then from the hybrid to the second-order phase transition whenβkeeps increasing,as shown in Fig.6.

    This process can be explained by the graphical method.SAis a monotonically increasing function aboutRAwhenkandβare inevitable, so the value ofRAcan directly reflect the phase transition.Taking the RR-RR EINDDLk=4 as an example, make the plot of the curve represented by Eqs.(19)and (20) and label the solution of Eq.(19) (i.e., the value ofRA),as shown in Fig.7.It can be seen thatRAchanges from zero to non-zero value continuously aspincreases in Fig.7(a).Therefore,the second-order phase transition occurs in network A.RAjumps from zero to a non-zero value aspincreases in Fig.7(b), and the first-order phase transition occurs in network A.In Fig.7(c),RAchanges continuously from zero to non-zero values and jumps when the two curves are tangent so that network A undergoes the hybrid phase transition.In Fig.7(d),RAstill changes from zero to non-zero continuously aspincreases,andRAdoes not jump because the tangent(intersection)point of the two curves is precisely on theRAaxis.Thus,network A undergoes the second-order phase transition.

    Fig.5.Plots of SA and SB versus p for the ER-ER EINDDL at different β and〈k〉.

    Fig.6.Plots of SA versus p for the EINDDL when β >0.8.

    Fig.7.Plots of curves represented by Eqs.(19)and(20).

    4.3.2.Robustness analysis of network B

    From Figs.4 and 5, it can be found thatSBin network B gradually changes from a more significant non-zero value to zero whenp <asβincreases,because network A fails completely whenp <and network B is equivalent to the failure of a single network with the initial removalβfraction of connectivity links.

    Next,we investigate the conditions,on which network B completely collapses.WhenSBis non-zero, Eq.(20) intersects theRBaxis in the plane (RA,RB) as shown in Fig.9(a),where the red line indicates Eq.(19) and the blue line indicates Eq.(20).Asβgradually increases to a certain value,Eq.(20)gradually goes from intersecting theRBaxis to passing the(0,0)point, when the order parameterSBis exactly 0 at a certainp, as shown in Fig.9(b), we define theβandpat this case as the critical valuesandwhen network B collapses completely.Subsequently, atβ ≥andp ≤,the value of the order parameterSBis always 0.Equation(20)no longer intersects with theRBaxis,as shown in Fig.9(c).

    Fig.8.Plots of phase transition threshold pAc versus β for network A.

    Fig.9.Plots of Eqs.(19)and(20)with identical p and different β.

    Fig.10.Plots of SB as a function of p-pc for the EINDDL when β >0.8.

    Then we study what type of phase transition occurs when network B is able to completely collapses.The plots ofSBas a function ofp-pcare made whenβ ≥, as shown in Fig.10.It can be seen thatSBturns from discontinuous to continuous with the increase ofβfrom Fig.10.Hence, asβincreases,network B turns from the first-order to second-order phase transition whenβ ≥.

    The process is similarly explained graphically by labeling the solution of Eq.(20) (i.e., the value ofRB) in Figs.7(b)–7(d).Asβincreases,the value ofRBchanges from discontinuously to a non-zero value to continuously to a non-zero value.Hence,network B changes from the first-order to the secondorder phase transition.Also,from Fig.7(d),it can be seen that when the tangent(intersection)point of the two curves is exactly on theRAaxis,the values ofRAandRBno longer jump,and the second-order phase transition occurs in both network A and network B at this time.

    It is clear that the first-order phase transition occurs in network B whenand the second-order phase transition occurs in network B whenfrom the above study.

    4.3.3.The prevention strategy

    From Subsections 4.3.1 and 4.3.2, it is clear that the robustness of both network A and network B become weaker and then stronger asβincreases from 0 to 1 with the phase transition point of view.The mechanism of this phenomenon is the way of spreading the damages influenced byβ.To illustrate this,we defineFas the ratio of failed links of networkA(B) spreading to networkB(A) to the total number of links of cascading failure in the second and third steps,and make a plot ofFversusβfor EINDDLk=4,p=0.3, as shown in Fig.11.

    Fig.11.Ratio of failure links of A(B)spreading to B(A)with different β.

    The number of failed links of network A (B) cascade to network B(A)increases and decreases asβincreases from 0 to 1.The more failed links there are in each cascading failure step,the more fragile the network is,that is the reason why the robustness of networks A and B decreases and then increases.Therefore,the value ofβcan directly affect the robustness of the EINDDL.Thus the prevention strategy is to find the value ofβwhen the EINDDL is most robust.

    We useSABsimilar to the order parameterSto denote the size of MCGC.Figure 12 shows the plots ofSABas a function ofp.The robustness of the EINDDL can be measured by the phase transition thresholdpcand the jump valuePJat the transition threshold.In order to find the prevention methods to optimize the network design,we study the relationship between robustness measured by both phase transition thresholdpcand the jump valuePJandβ.As we know,the smaller thepcandPJare,the more robust the system is.rising-falling-rising trend with the increase ofβ,and there exists peaks which reflect the strongest cases.Therefore,to prevent mass cascading failures,the value ofβshould be chosen around the high peaks and away from the low peak.

    Fig.12.Plots of SAB as a function of p for the EINDDL.

    5.Conclusions

    In summary, we have established a model for edgecoupled interdependent networks with directed dependency links (EINDDLs), and the directed dependency strength between two networks is determined byβ.The theoretical analysis framework of an EINDDL is developed based on the selfconsistent probability method,and the theoretical analysis and simulation of RR-RR and ER-ER networks are carried out to prove the theoretical analysis.The results show that under random attacks,network A exhibits different types of phase transitions(first-order,second-order and hybrid)with the increase ofβ, and the critical values ofβfor the whole phase transition type change process of network A are obtained.Network B is rarely affected by cascading failure whenβis small,and asβincreases, network B will gradually converge from the first-order to the second-order phase transition.The robustness of both network A and network B becomes weaker and then stronger asβincreases from 0 to 1 with the phase transition point of view.Finally,by measuring the robustness of the EINDDL with the phase transition thresholdpcand the jump valuePJ,we obtain the value ofβwhere the robustness of the EINDDL is best,and propose the prevention strategy.

    Fig.13.Plots of SR versus β for the EINDDL.

    For differentβ,we get two vectorsand.To evaluate the robustness of networks,we forward transit and normalize the vectorsndto getSpcandSPJ,so the comprehensive robustness evaluation metricSRof the EINDDL is expressed in the following,where the weightsωcandωJof the two metrics are received by the entropy method.

    Due to the forward transition and normalized vectorsp′candthe larger theSRis, the stronger the networks are.The plots ofSRversusβfor differentkis shown in Fig.13.It can be seen that the robustness of the EINDDL shows an

    Acknowledgments

    Project supported by the National Natural Science Foundation of China (Grant Nos.61973118, 51741902,11761033, 12075088, and 11835003), Project in JiangXi Province Department of Science and Technology (Grant Nos.20212BBE51010 and 20182BCB22009), and the Natural Science Foundation of Zhejiang Province (Grant No.Y22F035316).

    猜你喜歡
    周杰銀座
    Experimental realization of quantum controlled teleportation of arbitrary two-qubit state via a five-qubit entangled state
    銀座集團-山東商院商貿(mào)流通虛擬仿真實訓(xùn)基地建設(shè)研討會在我校召開
    “一根筋”周杰,特立獨行的娛樂“圈外人”
    May You Always
    山東銀座集團的財務(wù)戰(zhàn)略分析
    新營銷(2019年14期)2019-12-18 01:34:40
    銀座東急購物廣場日本東京
    中鐵·銀座尚筑 迭代公寓 國匠品質(zhì)
    齊魯周刊(2017年30期)2017-08-09 00:39:46
    中鐵·銀座尚筑新品發(fā)布會暨銀座簽約儀式
    齊魯周刊(2017年30期)2017-08-09 00:36:55
    Characteristics of Meteorological Factors over Different Landscape Types During Dust Storm Events in Cele,Xinjiang,China
    臨沭縣玉山鎮(zhèn):小小紫薯“入駐”銀座,變身市民“新寵”
    長江蔬菜(2013年23期)2013-03-11 15:09:39
    国产黄片美女视频| 国产永久视频网站| 国产亚洲最大av| 最新中文字幕久久久久| 日本色播在线视频| 日韩一本色道免费dvd| 97人妻精品一区二区三区麻豆| 免费看a级黄色片| 久久99精品国语久久久| 日韩人妻高清精品专区| 夜夜看夜夜爽夜夜摸| 免费大片黄手机在线观看| 一个人看视频在线观看www免费| 国产日韩欧美在线精品| 亚洲国产成人一精品久久久| 中文字幕av在线有码专区| 免费观看精品视频网站| 久久久精品免费免费高清| 日韩欧美一区视频在线观看 | 亚洲不卡免费看| 亚洲av在线观看美女高潮| av在线老鸭窝| 成人一区二区视频在线观看| 日韩电影二区| 久久这里有精品视频免费| 80岁老熟妇乱子伦牲交| 久久久久免费精品人妻一区二区| 欧美人与善性xxx| 国内精品一区二区在线观看| 亚洲色图av天堂| 天天一区二区日本电影三级| 女人久久www免费人成看片| 欧美最新免费一区二区三区| 精品不卡国产一区二区三区| eeuss影院久久| 成人漫画全彩无遮挡| 能在线免费观看的黄片| 免费观看精品视频网站| 只有这里有精品99| 成人特级av手机在线观看| 日本欧美国产在线视频| 啦啦啦韩国在线观看视频| 国产熟女欧美一区二区| 五月玫瑰六月丁香| 在线 av 中文字幕| 联通29元200g的流量卡| 波野结衣二区三区在线| 高清午夜精品一区二区三区| 在线天堂最新版资源| 纵有疾风起免费观看全集完整版 | 2022亚洲国产成人精品| 肉色欧美久久久久久久蜜桃 | 成人亚洲欧美一区二区av| 夜夜看夜夜爽夜夜摸| 久久久久久九九精品二区国产| 亚洲精品日韩在线中文字幕| 国产精品无大码| 亚洲一级一片aⅴ在线观看| 一级毛片电影观看| 国产黄片视频在线免费观看| 亚洲国产精品成人综合色| 久久韩国三级中文字幕| 男女视频在线观看网站免费| 欧美性感艳星| 卡戴珊不雅视频在线播放| 日韩欧美精品免费久久| 最近2019中文字幕mv第一页| 三级国产精品片| av线在线观看网站| 久久这里有精品视频免费| 久久久久久久午夜电影| 身体一侧抽搐| 亚洲成人精品中文字幕电影| 在线免费观看的www视频| 国产成年人精品一区二区| 欧美不卡视频在线免费观看| 精品一区二区三区人妻视频| 黄色一级大片看看| 18+在线观看网站| 国产伦一二天堂av在线观看| 不卡视频在线观看欧美| 国产精品一区二区在线观看99 | 亚洲精品乱码久久久久久按摩| 亚洲自拍偷在线| 国产精品久久久久久精品电影| 床上黄色一级片| 欧美激情在线99| 99久国产av精品国产电影| 成人鲁丝片一二三区免费| 观看免费一级毛片| 亚洲国产日韩欧美精品在线观看| 久久久久久久久久人人人人人人| 男人舔奶头视频| 亚洲国产av新网站| 欧美性猛交╳xxx乱大交人| 国语对白做爰xxxⅹ性视频网站| 午夜激情福利司机影院| 成人鲁丝片一二三区免费| 亚洲精品国产av蜜桃| 精品亚洲乱码少妇综合久久| 国产亚洲精品av在线| h日本视频在线播放| 美女脱内裤让男人舔精品视频| 三级国产精品欧美在线观看| 免费看a级黄色片| 精华霜和精华液先用哪个| 国产精品人妻久久久影院| 一夜夜www| 最近2019中文字幕mv第一页| 午夜久久久久精精品| 久久99热这里只频精品6学生| 毛片女人毛片| 亚洲av电影不卡..在线观看| 看黄色毛片网站| 成人亚洲欧美一区二区av| 深夜a级毛片| 有码 亚洲区| 欧美zozozo另类| 99热这里只有是精品在线观看| 狠狠精品人妻久久久久久综合| 亚洲精品乱码久久久v下载方式| 啦啦啦中文免费视频观看日本| 哪个播放器可以免费观看大片| 3wmmmm亚洲av在线观看| 亚洲精品日韩在线中文字幕| 少妇人妻一区二区三区视频| 国产爱豆传媒在线观看| 一级黄片播放器| 韩国av在线不卡| 丝袜美腿在线中文| 日韩亚洲欧美综合| 99久国产av精品国产电影| 精品人妻熟女av久视频| 亚洲综合精品二区| 99热全是精品| 国产亚洲一区二区精品| 免费看av在线观看网站| 精品一区二区免费观看| 草草在线视频免费看| 亚洲av电影在线观看一区二区三区 | 在线观看免费高清a一片| 中文乱码字字幕精品一区二区三区 | 亚洲欧洲国产日韩| 久久久久久久久大av| 神马国产精品三级电影在线观看| 国国产精品蜜臀av免费| 午夜福利视频1000在线观看| 免费av观看视频| 九九在线视频观看精品| 国产亚洲5aaaaa淫片| 精品久久久久久电影网| 亚洲内射少妇av| 一本一本综合久久| 99热这里只有是精品在线观看| 日韩欧美国产在线观看| 午夜日本视频在线| 色哟哟·www| 少妇熟女欧美另类| 国产精品不卡视频一区二区| 日韩中字成人| 最近中文字幕高清免费大全6| 日本三级黄在线观看| 亚洲,欧美,日韩| 中文字幕av成人在线电影| 亚洲av一区综合| 18禁在线无遮挡免费观看视频| 国产大屁股一区二区在线视频| 岛国毛片在线播放| 精品久久国产蜜桃| 韩国高清视频一区二区三区| 国产伦一二天堂av在线观看| 男的添女的下面高潮视频| 一级片'在线观看视频| 精品欧美国产一区二区三| 亚洲精品成人av观看孕妇| 欧美高清成人免费视频www| 午夜福利高清视频| 2018国产大陆天天弄谢| 卡戴珊不雅视频在线播放| 嘟嘟电影网在线观看| 亚洲不卡免费看| 亚洲精品一二三| 婷婷六月久久综合丁香| 能在线免费看毛片的网站| av专区在线播放| 寂寞人妻少妇视频99o| 噜噜噜噜噜久久久久久91| 国产精品久久久久久精品电影小说 | 啦啦啦啦在线视频资源| 国产精品久久久久久久电影| 汤姆久久久久久久影院中文字幕 | 夫妻午夜视频| 18禁裸乳无遮挡免费网站照片| 成人av在线播放网站| 啦啦啦啦在线视频资源| 亚洲av二区三区四区| 国内揄拍国产精品人妻在线| 免费电影在线观看免费观看| 国产一区亚洲一区在线观看| av在线播放精品| 三级国产精品片| 欧美日本视频| 国产欧美另类精品又又久久亚洲欧美| 精品久久久精品久久久| 如何舔出高潮| 国产精品.久久久| 国产伦精品一区二区三区视频9| 国产淫片久久久久久久久| 精品久久久久久久人妻蜜臀av| 最新中文字幕久久久久| 欧美日韩国产mv在线观看视频 | 99久久中文字幕三级久久日本| 六月丁香七月| 99热这里只有是精品50| 晚上一个人看的免费电影| 非洲黑人性xxxx精品又粗又长| 午夜久久久久精精品| 高清视频免费观看一区二区 | 精品午夜福利在线看| 舔av片在线| 午夜精品在线福利| 狠狠精品人妻久久久久久综合| 日韩一区二区视频免费看| 卡戴珊不雅视频在线播放| 亚洲欧洲日产国产| 国国产精品蜜臀av免费| 亚洲熟女精品中文字幕| 午夜福利在线在线| 少妇人妻一区二区三区视频| 日韩精品有码人妻一区| 一级a做视频免费观看| 免费观看无遮挡的男女| 日日干狠狠操夜夜爽| 在线观看美女被高潮喷水网站| 97在线视频观看| 两个人视频免费观看高清| 日本猛色少妇xxxxx猛交久久| 我要看日韩黄色一级片| 最后的刺客免费高清国语| 精品欧美国产一区二区三| 最近2019中文字幕mv第一页| 欧美日韩亚洲高清精品| 亚洲av成人精品一区久久| 99久久精品一区二区三区| 午夜视频国产福利| 不卡视频在线观看欧美| 寂寞人妻少妇视频99o| 九九在线视频观看精品| 人人妻人人看人人澡| 国产欧美日韩精品一区二区| 97超碰精品成人国产| 国产av码专区亚洲av| 国产极品天堂在线| 亚洲精品亚洲一区二区| 午夜福利在线在线| 最新中文字幕久久久久| 久久人人爽人人爽人人片va| 偷拍熟女少妇极品色| 毛片女人毛片| 久久99热6这里只有精品| 尤物成人国产欧美一区二区三区| videos熟女内射| 午夜精品一区二区三区免费看| 不卡视频在线观看欧美| 美女脱内裤让男人舔精品视频| 22中文网久久字幕| 深夜a级毛片| 日韩大片免费观看网站| 狂野欧美激情性xxxx在线观看| 免费观看性生交大片5| 国产高清有码在线观看视频| 高清欧美精品videossex| 国产视频内射| 亚洲天堂国产精品一区在线| 干丝袜人妻中文字幕| 高清毛片免费看| 精品久久久噜噜| 成年版毛片免费区| 亚洲欧美精品专区久久| 搡老妇女老女人老熟妇| 99久国产av精品| 中文字幕av在线有码专区| 最近中文字幕2019免费版| 欧美bdsm另类| 日本爱情动作片www.在线观看| 亚洲四区av| 人妻系列 视频| 男女那种视频在线观看| 一个人看的www免费观看视频| 国产av在哪里看| 国产国拍精品亚洲av在线观看| 听说在线观看完整版免费高清| 国产黄片视频在线免费观看| 春色校园在线视频观看| 免费看av在线观看网站| 国产欧美日韩精品一区二区| 国产黄片美女视频| 精华霜和精华液先用哪个| 亚洲欧美成人综合另类久久久| 日韩三级伦理在线观看| 69人妻影院| 天堂俺去俺来也www色官网 | 亚洲av男天堂| 婷婷六月久久综合丁香| 国产黄a三级三级三级人| 一级二级三级毛片免费看| 亚洲最大成人av| 亚洲精品影视一区二区三区av| 成年免费大片在线观看| 午夜免费男女啪啪视频观看| 日韩国内少妇激情av| 22中文网久久字幕| 久久久久精品久久久久真实原创| 久久99热这里只有精品18| 我要看日韩黄色一级片| 亚洲国产精品成人久久小说| 欧美激情在线99| 三级毛片av免费| 国产乱人偷精品视频| av专区在线播放| 你懂的网址亚洲精品在线观看| 五月伊人婷婷丁香| 欧美一区二区亚洲| 午夜激情福利司机影院| 丝瓜视频免费看黄片| 赤兔流量卡办理| 精品不卡国产一区二区三区| 免费播放大片免费观看视频在线观看| 男女边吃奶边做爰视频| 插阴视频在线观看视频| 亚洲国产成人一精品久久久| 日韩在线高清观看一区二区三区| 日本-黄色视频高清免费观看| 少妇被粗大猛烈的视频| 一级二级三级毛片免费看| 简卡轻食公司| 国产高清三级在线| 黄色日韩在线| 在线天堂最新版资源| 国产精品无大码| 欧美性感艳星| 麻豆精品久久久久久蜜桃| 久久人人爽人人片av| 一级爰片在线观看| 91狼人影院| 国产爱豆传媒在线观看| 2018国产大陆天天弄谢| 在线免费十八禁| 国产 一区 欧美 日韩| 大香蕉久久网| 亚洲自拍偷在线| 久久综合国产亚洲精品| 日本黄色片子视频| 欧美人与善性xxx| 国产高潮美女av| 久久99蜜桃精品久久| 99久久精品热视频| 一夜夜www| 成年人午夜在线观看视频 | 97超碰精品成人国产| 免费av不卡在线播放| 亚洲欧美日韩无卡精品| 91久久精品国产一区二区三区| 99热6这里只有精品| 亚洲三级黄色毛片| 精品熟女少妇av免费看| 精品久久久久久久久亚洲| 国产高清有码在线观看视频| 亚洲久久久久久中文字幕| 中文资源天堂在线| 边亲边吃奶的免费视频| 男女下面进入的视频免费午夜| 一夜夜www| 欧美极品一区二区三区四区| 麻豆成人av视频| 国产精品无大码| 91在线精品国自产拍蜜月| 高清毛片免费看| 日本熟妇午夜| 久久99精品国语久久久| 亚洲av成人av| 婷婷色综合www| 美女被艹到高潮喷水动态| 日韩av在线免费看完整版不卡| 欧美精品国产亚洲| 两个人视频免费观看高清| 啦啦啦啦在线视频资源| 欧美变态另类bdsm刘玥| 亚洲精品亚洲一区二区| av在线天堂中文字幕| 韩国高清视频一区二区三区| 亚洲av中文字字幕乱码综合| 2022亚洲国产成人精品| 天美传媒精品一区二区| 久久精品国产亚洲av天美| 最近中文字幕2019免费版| 26uuu在线亚洲综合色| 成人性生交大片免费视频hd| 日韩一区二区视频免费看| 精品熟女少妇av免费看| 2021天堂中文幕一二区在线观| 淫秽高清视频在线观看| 女人被狂操c到高潮| 亚洲精品一二三| 婷婷色麻豆天堂久久| 国产亚洲最大av| 麻豆乱淫一区二区| 精品久久久久久久人妻蜜臀av| 亚洲欧美精品自产自拍| 色吧在线观看| 精品欧美国产一区二区三| 夜夜爽夜夜爽视频| 国产有黄有色有爽视频| 高清日韩中文字幕在线| 国内精品宾馆在线| 成人综合一区亚洲| 男女边摸边吃奶| 免费大片18禁| 精品99又大又爽又粗少妇毛片| 国产精品1区2区在线观看.| 亚洲美女视频黄频| 美女cb高潮喷水在线观看| 成人综合一区亚洲| freevideosex欧美| 啦啦啦中文免费视频观看日本| 国产精品蜜桃在线观看| .国产精品久久| 亚洲国产欧美在线一区| 亚洲精品国产成人久久av| 日韩欧美精品免费久久| 韩国av在线不卡| 精品久久久噜噜| 日韩av不卡免费在线播放| 成人漫画全彩无遮挡| 午夜激情福利司机影院| 国产高清三级在线| 国内少妇人妻偷人精品xxx网站| 国产亚洲5aaaaa淫片| 国产国拍精品亚洲av在线观看| 简卡轻食公司| 成人欧美大片| 久久久久久久亚洲中文字幕| 少妇熟女aⅴ在线视频| 亚洲精品成人久久久久久| 搡老乐熟女国产| 午夜激情福利司机影院| 国产成人一区二区在线| 亚洲精品日韩av片在线观看| 婷婷色综合www| 91狼人影院| 亚洲av男天堂| 国产精品美女特级片免费视频播放器| 欧美高清成人免费视频www| 只有这里有精品99| 男女视频在线观看网站免费| 欧美一级a爱片免费观看看| 中文欧美无线码| 久久99热这里只频精品6学生| 久久久久久久国产电影| 大话2 男鬼变身卡| 欧美日韩视频高清一区二区三区二| 99久久人妻综合| 精品一区二区三区人妻视频| a级毛片免费高清观看在线播放| 晚上一个人看的免费电影| 日本黄大片高清| 嫩草影院精品99| 亚洲美女搞黄在线观看| 国产一级毛片七仙女欲春2| 91久久精品国产一区二区三区| 欧美日韩亚洲高清精品| 亚洲精品日韩av片在线观看| 丰满人妻一区二区三区视频av| 国产美女午夜福利| 少妇高潮的动态图| 亚洲精品色激情综合| 国产精品一二三区在线看| h日本视频在线播放| 亚洲内射少妇av| 蜜桃亚洲精品一区二区三区| 夜夜看夜夜爽夜夜摸| 亚洲最大成人av| 精品久久久久久成人av| 亚洲精品aⅴ在线观看| 日本爱情动作片www.在线观看| 在线观看人妻少妇| 国产亚洲5aaaaa淫片| 午夜福利在线观看吧| 午夜视频国产福利| av国产免费在线观看| 国产精品嫩草影院av在线观看| 欧美激情国产日韩精品一区| 七月丁香在线播放| 国产乱来视频区| 波野结衣二区三区在线| 一级毛片电影观看| 蜜桃久久精品国产亚洲av| av网站免费在线观看视频 | 欧美zozozo另类| 国产欧美日韩精品一区二区| 看非洲黑人一级黄片| 国产人妻一区二区三区在| 欧美激情久久久久久爽电影| 日韩av免费高清视频| 狠狠精品人妻久久久久久综合| 成人毛片a级毛片在线播放| 啦啦啦中文免费视频观看日本| 精品一区二区三卡| 一本—道久久a久久精品蜜桃钙片 精品乱码久久久久久99久播 | 亚洲国产精品专区欧美| 国产乱来视频区| 国产免费一级a男人的天堂| 91久久精品国产一区二区成人| 午夜免费男女啪啪视频观看| 国产亚洲91精品色在线| 亚洲av电影不卡..在线观看| 国产精品一区二区性色av| 国精品久久久久久国模美| 免费人成在线观看视频色| 少妇丰满av| 久久99热6这里只有精品| 成人毛片60女人毛片免费| 18禁在线无遮挡免费观看视频| 欧美激情国产日韩精品一区| 一级毛片电影观看| 亚洲国产精品国产精品| 欧美日韩在线观看h| 简卡轻食公司| 夜夜看夜夜爽夜夜摸| 午夜福利视频精品| 精品少妇黑人巨大在线播放| 国产精品爽爽va在线观看网站| av线在线观看网站| 91av网一区二区| 午夜激情欧美在线| 亚洲电影在线观看av| av在线播放精品| 亚洲精华国产精华液的使用体验| 草草在线视频免费看| 国产视频内射| 午夜福利网站1000一区二区三区| 91精品伊人久久大香线蕉| 不卡视频在线观看欧美| 国产亚洲最大av| 夜夜爽夜夜爽视频| 美女黄网站色视频| 亚洲精品乱码久久久v下载方式| 国产一区二区三区av在线| 日韩一区二区三区影片| 精品人妻偷拍中文字幕| 中文乱码字字幕精品一区二区三区 | 精品一区二区三区人妻视频| 一二三四中文在线观看免费高清| 国产午夜精品论理片| 成人特级av手机在线观看| 中文字幕久久专区| 丰满人妻一区二区三区视频av| 久久久久久久大尺度免费视频| 国产精品国产三级专区第一集| 乱系列少妇在线播放| 寂寞人妻少妇视频99o| 国精品久久久久久国模美| 小蜜桃在线观看免费完整版高清| 久久人人爽人人爽人人片va| 国产精品麻豆人妻色哟哟久久 | 韩国av在线不卡| 纵有疾风起免费观看全集完整版 | 国产单亲对白刺激| 亚洲激情五月婷婷啪啪| 日本一二三区视频观看| 成年版毛片免费区| 久久97久久精品| 搞女人的毛片| 亚洲av一区综合| 一个人看视频在线观看www免费| 18禁在线播放成人免费| 99久久精品一区二区三区| 亚洲av福利一区| 免费少妇av软件| 免费不卡的大黄色大毛片视频在线观看 | 亚洲不卡免费看| 男人舔女人下体高潮全视频| 男人狂女人下面高潮的视频| 亚洲成人精品中文字幕电影| 久久精品国产亚洲网站| 婷婷色麻豆天堂久久| 欧美xxxx性猛交bbbb| 两个人的视频大全免费| 禁无遮挡网站| 熟女电影av网| 国产在视频线精品| 午夜福利在线观看免费完整高清在| 亚洲国产最新在线播放| 亚洲av二区三区四区| 在线观看免费高清a一片| 精品酒店卫生间| 一二三四中文在线观看免费高清| 日韩,欧美,国产一区二区三区| 观看美女的网站| 色5月婷婷丁香| av在线天堂中文字幕| 亚洲欧美成人精品一区二区| 噜噜噜噜噜久久久久久91| 亚洲人成网站在线观看播放| 97在线视频观看| 午夜视频国产福利| 男女国产视频网站| 亚洲精华国产精华液的使用体验| 淫秽高清视频在线观看| 日韩欧美一区视频在线观看 | 亚洲精品,欧美精品| 亚洲一区高清亚洲精品| 亚洲国产精品sss在线观看| 欧美极品一区二区三区四区| 久久久久精品久久久久真实原创|