• <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 ?

    On the Boundary Condition and Related Instability in the Smoothed Particle Hydrodynamics

    2019-11-07 02:58:20ChongYe葉翀PhilipeMotaJinLi李瑾KaiLin林愷andWeiLiangQian錢衛(wèi)良
    Communications in Theoretical Physics 2019年11期

    Chong Ye (葉翀), Philipe Mota, Jin Li (李瑾),Kai Lin (林愷), and Wei-Liang Qian (錢衛(wèi)良)

    1College of Physics, Chongqing University, Chongqing 401331, China

    2Centro Brasileiro de Pesquisas Físicas, 22290-180, Rio de Janeiro, RJ, Brazil

    3Institute of Geophysics and Geoinformatics, China University of Geosciences, Wuhan 430074, China

    4Escola de Engenharia de Lorena, Universidade de S?o Paulo, 12602-810, Lorena, SP, Brazil

    5Faculdade de Engenharia de Guaratinguetá, Universidade Estadual Paulista, 12516-410, Guaratinguetá, SP, Brazil

    6School of Physical Science and Technology, Yangzhou University, Yangzhou 225002, China

    Abstract In this work,we explore various relevant aspects of the Smoothed Particle Hydrodynamics regarding Burger’s equation.The stability, precision, and efficiency of the algorithm are investigated in terms of different implementations.In particular, we argue that the boundary condition plays an essential role in the stability of numerical implementation.Besides, the issue is shown to be closely associated with the initial particle distribution and the interpolation scheme.Among others, we introduce an interpolation scheme termed symmetrized finite particle method.The main advantage of the scheme is that its implementation does not involve any derivative of the kernel function.Concerning the equation of motion,the calculations are carried out using two distinct scenarios,where the particles are chosen to be either stationary or dynamically evolved.The obtained results are compared with those obtained by using the standard finite difference method for spatial derivatives.Our numerical results indicate subtle differences between different schemes regarding the choice of boundary condition.In particular, a novel type of instability is observed where the regular distribution is compromised as the particles start to traverse each other.Implications and further discussions of the present study are also addressed.

    Key words:hydrodynamic model, SPH algorithm, pair instability

    1 Introduction

    The smoothed particle hydrodynamics (SPH)[1?9]is a meshfree, Lagrangian method for the partial different equations.Distinct from traditional approaches such as the finite difference method (FDM) or finite element method,the SPH approximates a partial differential equation in terms of ordinary differential equations of freely moving interpolation points, known as SPH particles.Subsequently, the infinite number of degrees of freedom of the original system is represented by a finite number of SPH particles, aiming to describe any arbitrary configuration of the continuum medium.As one of the most important applications of the partial differential equation is to describe the dynamics of fluid, SPH method naturally finds extensive applications in the context of hydrodynamics, as inferred by the terminology.Indeed,due to the significance of hydrodynamics, as well as recent advance in computing power, the SPH method is widely employed in many distinct areas of science and engineering.Among traditional applications to fluid dynamics,the SPH method has been employed to investigate a great variety of topics.In astrophysics, the SPH is implemented for the study of galaxy formation, supernova explosion.[7,9?11]In solid state mechanics,[12]the method has been applied to problems such as metal forming,[13]fracture and fragmentation,[14]and shock wave propagation in solids.[15]Moreover, concerning the high energy nuclear physics, the hydrodynamic approach of heavy-ion nuclear collisions plays an essential role associated with RHIC and LHC.[16?18]The latter further inspires onging investigations of fluid/gravity duality,[19?22]one realization of the holographic principle.[23,24]Numerical simulations based on implementations of SPH algorithm[25?29]have shown to give reasonable description on various experimental observables.The algorithm is also employed in chemistry where a variety of chemical evolutions simulations has been implemented in terms of the SPH method.

    The SPH was initially suggested to investigate astrophysical problems by Gingold and Monaghan[2]and by Lucy.[1]In the standard implementation, each SPH particle is assigned a finite size,h, over the extension of which the relevant physical properties are smoothed in terms of a kernel function.By kernel function, the contribution of each particle is weighted regarding their respective distance to the coordinate position of interest.As a consequence,a physical quantity at a given position is evaluated by summing over all the contributions from all the relevant SPH particles in the neighborhood within reach of the kernel function.In terms of the above interpolation regarding SPH particles, a partial differential equation is transformed into a system of ordinary differential equations of the SPH degrees of freedom.

    The SPH algorithm is a meshfree approach, which implies several essential features.First,the method is suited to handle problems with sophisticated boundary conditions, inclusively, scenarios with free boundary.The algorithm can achieve exact advection mostly in a trivial manner, which is a rather difficult task for any Eulerian grid-based approaches.Meanwhile, the absence of the structured computational mesh, by and large, simplifies model implementation.In particular, the spatial derivatives are implemented in a flexible and straightforward fashion, independent of any grid-based configuration.Besides, one of the most remarkable features of the SPH method is that it naturally guarantees the conservation law.In other words, independent of any specific application, the algorithm rigorously conserves quantities such as mass, energy, momentum, and angular momentum by construction.Also, a certain degree of re-meshing takes place automatically as time evolves, as a consequence of the kernel function.It is distinct from other Lagrangian grid-based methods where the grid can sometimes become arbitrarily disordered in time.[30]Moreover, when appropriately configured, it is understood that the computational cost of numerical simulations per SPH particle is usually less than that per cell of grid-based simulations.This is because the resolution of the method follows naturally the mass flow rather than the volume as in the case of most grid-based techniques.It happens as each SPH particle carries a fixed amount of mass and dislocate itself according to the flow velocity, and subsequently, the density only goes up related to an increase in the local matter concentration.[30]To be more specific,it is the case especially when the metric of interest is associated with the relevant density, and thus an automatical refinement on density is achieved.Furthermore, the SPH method is highly flexible and can be combined with other algorithms straight forwardly.[10]

    However, the SPH formulation also faces a few challenges.Several notable problems potentially hamper further exploiation of the method, such as accuracy, tensile instability,[31?33]pairing instability,[34]the kernel as well as particle consistency,[8,35?38]and artificial viscosity.[39]From a physical viewpoint, tensile instability is triggered when SPH particles become mutually attractive due to,by and large,the presence of negative pressure.The eventual instability manifested regarding particle distribution,known as tensile instability,can be viewed as equivalent to the mesh distortion in a mesh-based scenario, is typically found in simulations of elastic material under tension.One possible approach to deal with such instability associated with irregular particle distribution is to sacrifice exact conservation of the algorithm by a slim amount.[40]Besides,other approaches were proposed to handle the instability by improving the estimation of spatial gradients, such as the corrected derivatives method.[41]As the number of neighbors of an individual SPH particle increases,the ratio of the particle spacing to the smoothing length decreases.However, for standard “Bell-shaped” or “Gaussian-like”kernels, their first derivative, and equivalently the force between SPH particles, vanishes at the origin.Consequently,particle pairing occurs.This problem is relatively benign, as it usually does not lead to any severe consequence.It can mostly be avoided by just using an adequate initial particle distribution with a smaller number of neighbors or adopting a short-range kernel function.Concerning kernel and particle consistency, much effort has been devoted to restore the consistency and subsequently improve accuracy.Monaghan derived a symmetrization formulation for the spatial gradient term[3]consistent with the variational approach,which is shown to provide better accuracy.Randles and Libersky proposed a normalization scheme for the density approximation.[35]The concept is further developed by Chenet al.[33,36,42]in terms of a corrective smoothed particle method (CSPM) as well as by Liuet al.[37]in terms of a finite particle method (FPM).For some physical phenomena such as shockwave, the entropy is increased as a physical consequence.In fact, unlike in a Eulerian scheme where numerical dissipation is an intrinsic feature related to the discretization process and mostly unavoidable, diffusion terms in SPH must be inserted explicitly, and as a result, their physical content is transparent.In this regard, it has been argued that it is essential to supply an adequate amount of artificial viscosity which works for not only scenarios where artificial viscosity is required but also for other cases where its presence ought to be minimized.[43,44]

    As a matter of fact, there does not exist a single numerical method which performs satisfactorily well at all different aspects.Therefore, in practice, the choice of the best numerical approach can often depend entirely on which feature of the problem is more relevant to the present study.Nonetheless, stability is usually the primary concern of a numerical scheme, while precision and efficiency are two essential performance metrics in most cases.Our present study thus aims to provide a quantitative analysis of these aspects of the SPH method.In particular, we explore the role of boundary condition, as well as its relationship with initial particle distribution and interpolation scheme.Regarding the specific problem we choose to investigate, the equation of motion is implemented using two distinct scenarios, where the SPH particles are selected to be either stationary or allowed to evolve dynamically in time.The first scenario is shown to be an excellent indicator of the efficiency and accuracy of the individual interpolation algorithm.The second scenario, on the other hand, resides on a more general basis and is subsequently found to be sensitively associated with the stability of different implementations.

    The paper is organized as follows.In the following section, we briefly discussed the main features of different implementations of the SPH algorithms involves in the present study, namely, the standard SPH, CSPM, FPM,and the symmetrized finite particle method (SFPM).The calculations are carried out in Secs.3 and 4 where Burger’s equation is chosen to be the object of the numerical simulations.The results are then compared against the analytic solutions.Two specific scenarios for the equation of motion are made use of with either stationary and dynamical SPH particles.In Sec.3 we show the results concerning the first scenario, where we mainly focus on the precision and efficiency of different interpolation schemes.The second scenario is studied in Sec.4, which is mostly related to the stability of the algorithm regarding the boundary conditions.We discuss how the boundary condition affects the stability of the numerical calculations for different schemes.Also, a type of novel instability is observed where SPH particles are found to traverse each other,which, in turn, undermine the regular particle distribution.The last section is devoted to further discussions and concluding remarks.

    2 The SPH Method

    In the present study, we mainly focus on four different SPH interpolation schemes, namely, standard SPH,CSPM, FPM, and SFPM.In what follows, what briefly review these schemes concerning their derivations.For the spatial distribution of a given physical quantity,f(x),all the above schemes can be viewed as specific implementations to interpolate the distribution by using a finite number of SPH particles as well as the kernel functionW.The latter is assumed to be an even function and its normalization reads

    The Taylor series expansion off(x) reads

    We proceed by multiplying both sides by the kernel functionW(x ?xa) and integrating overx, we have

    As one keeps the first term on the r.h.s.of the above equation and discard the higher order ones, one obtains

    By substituting the normalization condition Eq.(1),we obtain the standard SPH formula, namely,

    We note that the particle approximation is utilized,where a summation enumerating all SPH particle is used to approximate the integral.Hereνbis the “mass” of theb-th particle, usually it is a given value assign to the particle at the beginning of the algorithm.ρbis the density of the SPH particles, iffitself is the density, one has

    In practice, the ratiofb/ρbcan be determined by the physical properties of the system in question.For instance, iffrepresents pressure andρis the number density, their ratio might be determined by the equation of state.[17]Otherwise,fbandρbcan be that evaluated for theb-th particle at the last time step.The first and second order derivatives can be obtained by carrying out spatial derivative of Eq.(5), and are found to be

    where symmetric formulism[3]has been employed.

    One might be aware of the fact that in the particle approximation of Eq.(1),the kernel function is not precisely identical to the Diracδ-function, and instead employ the following modified particle approximation in the place of Eq.(4),

    One finds

    Equation (10) is known as CSPM in literature, which can be shown to preserve the zeroth-order kernel and particle consistency.[8]The first-order and second-order derivatives can be derived by retaining until the second order term on the r.h.s.of Eq.(2), multiplying both sides byW′andW′′, and making use of the symmetry of these functions.One finds

    In fact, one may proceed further by generalizing the above arguments to higher order in a slightly more consistent fashion.This is achieved by consistently expanding the r.h.s.of Eq.(2) to the second order, namely,

    The present case is different from that of CSPM scheme, since now Eqs.(12)–(13) furnishes a system of coupled equations, as follows

    By inverting the above matrix equation, one obtains for the desired expressions for the FPM scheme, which read

    with

    The FPM scheme is understood to possess the zeroth and first-order particle consistency.[8]

    It is worth mentioning that in deriving Eq.(13), one has the freedom to utilize any basis function.Another particular choice, which is meaningful by its own right, is to make use of(x?xa)W(x?xa)and(x?xa)2W(x?xa)in the place ofW′(x ?xa) andW′(x ?xa).The only difference is that the definitions in Eq.(15) are replaced by

    We note that the alternative basis functions maintain the symmetries (even or odd with respect to the origin) of the integral kernel.Meanwhile, the formalism has avoided using the derivatives of the kernel function,which turns out to be quite advantageous as shown below.This last scheme was initially proposed in Ref.[45]and will be referred to as the symmetrized finite particle method(SFPM) in the present work.

    To sum up, Eqs.(5)–(7), Eq.(10), Eqs.(15)–(16) and Eqs.(16)–(17) give the specific formalism for standard SPH, CSPM, FPM and SFPM.In the remainder of the paper, these differences will be put into practice in the following two sections.

    3 Burgers’ Equation with Stationary SPH Particles:on Precision and Efficiency of Different Schemes

    In this section, we focus ourselves on the static aspect of the algorithm regarding the precision as well as the efficiency of different schemes.By and large, the precision of a given scheme can be measured in terms of kernel consistency and particle consistency.The former is determined by then-th moment of the kernel functions, namely,

    forn=0,1,2,...The above conditions are rigorous when the interpolation is perfect.When it is satisfied for a given ordern, the corresponding SPH approximation is said to possessn-th order orCnconsistency.

    The particle consistency, on the other hand, is related to the fact that in practice, the integral in Eq.(18) is implemented by a summation.As a result, one requires

    As mentioned above, the present section is primarily dedicated to investigating to what degree the SPH algorithm is capable of reproducing a well-defined distribution of certain physical quantity in terms of the SPH particles.In this context, the precision and efficiency are affected by various factors, such as the specific form of the kernel function, the number of particles, and the interpolation scheme.In what follows, we carry out numerical simulations while tuning parameters regarding these aspects.We note that, for the time being, we will temporarily discard another vital factor, the stability of the algorithm.In other words, we only concentrate on the precision and efficiency of specific problems where the stability is guaranteed or a minor issue.In this regard, we study the following one-dimensional Burgers’ equation:

    whereγis the viscosity coefficient.The reason for our choice of Burgers’ equation is as follows.First, the equation possesses an analytic solution, corresponding to smooth initial conditions whereuvanishes on the boundary.Secondly, usually, the physical interpretation of the equation is the same as the Navier-Stokes equation.In other words,the Burgers’equation determines the temporal evolution of,u,the velocity of the fluid.Therefore,the physical system is governed by two equations for the fluid velocityu(x,t) and matter densityρ(x,t), respectively.The latter is nothing but the continuity equation of the fluid, which takes the following form in one-dimensional space,

    However, it is interesting to point out that one may simply viewuas an intensive physical quantity, and solve Eq.(20) alone independent of Eq.(21).This is possible since, for the specific case of the Burgers’ equation, the equation ofudoes not depend onρ.Therefore, when implementing the SPH scheme, we choose to assign the SPH particles to fixed spatial coordinates.Here,uaof thea-th particle is a function of time, determined by the r.h.s.of Eq.(20), while the coordinates of the SPH particles remain unchanged during the simulation.Lastly,the boundary condition of the present problem is that the distribution attains zero at both end points, namely,

    But since the SPH particle do not dislocate in time, for the present section, one does not need to explicitly implement Eq.(22) into the code.As shown in the next section, the latter is closely related to the issue of stability.For simplicity,throughout our calculations,we employ the fifth order spline kernel function.[46]We also note that the present application involves fixed boundary condition,while on the other hand, SPH algorithm is most advantageous for problems with generic free boundary condition,the resultant precision and efficiency of SPH is not expected to be superior to the outcome of traditional FDM.In what follows,the different schemes discussed in the previous section will be applied by using the above setup.

    In Fig.1, we show the calculated temporal evolutions of the distribution in comparison with those obtained by using the FDM as well as the analytic solutions.It is found that most methods give consistent and reasonable results for different viscosity coefficients.Among all the interpolations schemes, the standard SPH shows the most substantial deviations from the analytic results.To precisely identify the precision and efficiency of different schemes,we numerically evaluate the deviations from analytic results.In Table 1, the averaged relative precision and efficiency for different numbers of SPH particles or grid numbers (for FDM) are summarized.In our calculations, the particles are distributed uniformly, and the size of SPH particles,h, varies according to the total number of SPH particles.To be specific, we chooseh=?xwhere ?xis the distance between neighbor particles.In the calculations, the time interval is taken to be ?t=h/500, and we considerγ=1.0.The precision is estimated for the instantt=1.0 while regarding an average of the relative deviation

    whereNis the total number of selected spatial points where the quality of the interpolation is evaluated by comparing to analytic result.In our calculations,we have used seven evenly distributed spatial points for the evaluation.As the efficiency, one additionally takes into account the CPU time consumption, ?t.The efficiency of a given method is defined as follows

    which is measured with respect to that of the standard SPH method.

    Fig.1 (Color online) The numerical solutions of the Burgers’ equation in comparison with the analytic ones.The calculations are carried out by using stationary SPH particles for standard SPH, CSPM, FPM, FDM, and SFPM,respectively.The results are presented for different values of the viscosity coefficients γ.

    Table 1 The relative precision and efficiency of different methods.The first row indicates the number of SPH particles or grids used in the calculations.

    It is observed in Table 1,for all the methods,the calculatedPmonotonically decrease with increasing SPH particles.This is as expected, which indicates that the precision improves as the particle number increases.Also, the efficiency,which is defined as an overall evaluation of speed and accuracy, improves as the SPH particle increase.The modified SPH methods included in this study; namely,CSPM,FPM,and SFPM show significantly better performance than that of the standard SPH method.It is found that as the number of SPH particle further increases, the improvement of efficiency gradually slows down.This happens due to that the precision of the interpolation starts to be saturated while the computational time increases linearly.It is further expected for the higher dimensional case since the computation cost increases much faster with increasing particle number, the efficiency of the methods may even start to decrease as the particle number becomes significantly large.

    4 Burgers’ Equation with Dynamical SPH Particles:on the Stability of the Algorithm

    This section is devoted to the stability analysis of different schemes regarding the boundary condition and initial particle distribution.When not appropriately configured, a numerical algorithm might be plagued by instability.Here we investigate particularly how the boundary condition may affect the numerical instability.As discussed in the introduction, there are at least three different types of characteristics of the SPH algorithm closely related to stability issue, namely, self-regularization, pairing instability, and tensile instability.All of them are related to the particle distribution and the specific form of the kernel function.SPH can automatically maintain a reasonable particle arrangement[10,30]during the course of the temporal evolution.Owing to the form of the kernel function, mutual repelling forces are exerted between neighboring SPH particles,which automatically drives the system towards a regular particle distribution.Intuitively,it corresponds to “re-meshing” procedure that is utilized in other Lagrangian mesh methods.As the SPH particles come closer, paring instability takes place as soon as the repelling force start to decrease with decreasing neighbor distance.Furthermore, under certain conditions[31,32]tensile instability takes place and disrrupts the particles distribution.In what follows, we attempt to configure an initial SPH distribution where the distance of neighboring SPH particles is randomized but within the range where particles repelling each other and the strength decreases with increasing distance.However,our calculation does not reveal any automatical re-meshing.Moreover,we show that boundary condition may introduce small perturbation which is accumulated over time and turned into oscillations.Such instability is observed to be accompanied by particle crossing.In order to carry out the analysis, one allows the SPH particle to move according to the continuity equation, namely, Eq.(21).In other words, we will solve Eqs.(20) and (21) simultaneously, as in most SPH implementations.The numerical results are presented in Figs.2?5.

    Fig.2 (Color online) The numerical solutions of the Burgers’ equation in comparison with the analytic ones.The calculations are carried out by using different values of the viscosity coefficients for standard SPH, CSPM, FPM, FDM,and SFPM, respectively.The boundary condition is implemented by using fixed SPH particles.

    Fig.3 (Color online) The numerical solutions of the Burgers’ equation in comparison with the analytic ones.The calculations are carried out by using different values of the viscosity coefficients for standard SPH, CSPM, FPM, FDM,and SFPM, respectively.The boundary condition is implemented by adding extra SPH particles beyond the physical boundary.

    Fig.4 (Color online) The numerical solutions of the Burgers’ equation in comparison with the analytic ones.The calculations are carried out by using different values of the viscosity coefficients for standard SPH, CSPM, FPM, FDM,and SFPM, respectively.The boundary condition is implemented by adding mirrored imaginary SPH particles beyond the boundary.

    A critical aspect of the present study is to show that the boundary condition plays an essential role in the SPH algorithm.From its appearance, the boundary condition Eq.(22)of the problem looks“benign”.However,it turns out that the numerical stability sensitively depends on its implementation.We study three types of implementation for the boundary condition.First, we freeze the coordinates of the SPH particles on the boundary.Secondly, we extend the system beyond the boundary by adding more SPH particles and let those additional particles evolve freely.To be specific, we do not explicitly exert any additional condition for the most outward particles.Lastly,we implement mirroring boundary condition, by adding imaginary particles outside of the boundary to ensure the validity of Eq.(22).To be more precise, the imaginary particles are mirror images; namely, their positions are mirrored with respect to the location of the boundary,and their velocities are inverted regarding their counterparts.The obtained results are shown in Figs.2?4.

    The calculated results indicate that it is essential to appropriately implement the boundary condition, determined by Eq.(22).As shown in Fig.2, the calculations with fixed boundary SPH particles start to fail as the coefficient of viscosity becomes significantly small.Asγdecreases, it takes a longer time for the system to evolve,and subsequently,small deviations are accumulated for an extended period,and the results demonstrate more significant discrepancy.It is observed that most schemes show significant deviations forγ <0.1 and fort >0.75, where the standard SPH fails dramatically.While one adds additional SPH particles beyond the boundary of the physical problem and apply free boundary condition for those particles,the results show a very similar feature asγbecomes sufficiently small.This can be observed in Fig.3.In Fig.4,the results are shown by employing mirrored boundary condition.We did not draw the mirrored SPH particles beyond the boundary as they are symmetrically positioned regarding their counterparts.In general, the results are significantly improved.It is found that the implemented boundary condition gives a significant impact on the resulting temporal evolution.For CSPM, the results are more stable and agree reasonably with the analytic ones for all different viscosity coefficients.Although, in some region,the numerical results are still found to deviate from the analytic ones.For FPM and SFPM, small oscillations are observed when the time becomes more substantial as SPH particles start to pile up at the right boundaryx=1.For the standard SPH method,the results are less satisfactory compared to the other approaches.We understand that it is mostly related to the fact that the standard SPH method suffers a severe problem with kernel and particle consistency.

    In order to investigate the instability occurs near the rightmost boundary, we carry out further calculations focusing the region near the vicinity ofx=1.To be specific, the numerical calculations are carried out for the range ofx ∈(0,2) with the boundary beingx=0 andx=2.The imaginary SPH particles beyond the original boundary are now treated as real particles and evolve according to the Burgers’ equation.In this context, the particles on both sides are moving toward the center atx=1.The numerical results are shown in Fig.5.The two plots on the first row are the trajectories the positions and velocities of SPH particles, shown by curves in different colors.A total of 125 SPH particles are initially distributed uniformly, whereh=0.2 and the size of the time step dt=0.02.The plots on the second as well as third rows are the same as the first row, except for the following difference.For the second row,the SPH particle distribution is also uniform initially, but a total number of 250 particles,withh=0.1 and dt=0.01.For the third row,regarding the initial condition for the case of the second row, the positions and velocity of SPH particles are perturbed by a small amount at the initial time.While the calculations are carried out for the rangex ∈(0,2)with the boundary are defined by the pointsx=0 and 2,the results are only shown in the vicinity region ofx ~1.It is found that on the first row of Fig.5, particles simple travel across each other after they come close.This is a troublesome result, especially when comparing to the analytic results, the fluid on either side of the boundary should never traversex=1.Therefore,some fundamental problem regarding interpolation is implied.As will be discussed below, the stability mentioned above in Figs.2?4 is likely related to the present phenomenon.On the second and third rows, particles bound back from each other due to the repelling force between them, as intended.As discussed before, if particles maintain at an appropriate distance between each other, particle auto-remeshing is expected.In the plots on the second row of Fig.5, this is manifested, to a certain degree, in terms of the fact that all the SPH particles gradually decelerate and come to a full stop simultaneously att ~0.6.Further, in the plots on the third row,it is shown that the particles also bounce back from each other, owing to the auto-remeshing mechanism.However, our results indicate that auto-remeshing does not always work in the present context, in particular, where the fluid becomes piled up atx=1 and is forced to squeeze into a small region, as governed by the Burgers’ equation.In this case, the results become quite sensitive to the specific parameter used in the calculations,as demonstrated by the plots on the first and second rows.When the auto-remeshing fails, as shown in the first row,the SPH particles are found to traverse each other when they stay too close to each other.Judging from the form of the kernel function, when ?x < h, the repelling between two neighboring SPH particles decrease with decreasing distance.Therefore, as the neighboring particles become too close,as it inevitably happens in the case of the Burgers’ equation, the mechanism of particle auto-remeshing may fail.In fact, we observe that the resultant particle traversing occurs for a broad choice of parameters in the present scenario of the Burgers’ equation.We note that such a phenomenon is observed to accompany the instability discussed previously concerning Figs.2?4 closely.

    5 Concluding Remarks

    To summarize, in this paper, we carry out an analysis regarding the stability, precision, and efficiency of the SPH method.In particular, various aspects of the algorithm are explored such as the boundary condition,initial particle distribution,and interpolation scheme.Besides,a generalized FPM scheme, the symmetrized finite particle method,is employed in our study.The main advantage of the SFPM is that its implementation does not explicitly involve the derivatives of the kernel function.We studied two different scenarios for the implementation of the equation of motion, where the SPH particles are either stationary or dynamically evolved in time.In the first scenario, the obtained results are also compared with those obtained by using the FDM as well as analytic ones.All numerical schemes provide reasonable results in the case.Among others, it is indicated that CSPM and SFPM provides an overall better performance regarding precision as well as stability.

    For the second scenario, the positions and velocity of the SPH particles are determined by the equations of motion.Though it is more meaningful to apply this scheme to topics with dynamical boundary conditions,in the present study, it is adopted to the same problem of the Burgers’equation.It is subsequently found that the boundary condition implementation plays a crucial role in the success of the algorithm.Among the three implementations, the mirrored boundary condition leads to mostly satisfactory results.Still,instability in terms of oscillations is observed when particle crossing starts to take place.Even for the case of mirrored boundary condition,increasing oscillation is observed for FPM and SFPM methods in regions with more significant particle density.Since the SFPM does not explicitly involve any derivative of the kernel function, the observed instability is likely of a distinct nature.This is because the sufficient condition of tensile instability involves the second order of the kernel function when derived regarding small perturbations.[31]One crucial feature of observed instability is that the oscillations take place when SPH particles start to cross each other.We will try to address the issue of particle crossing and the related instability in future work.

    Acknowledgment

    We gratefully acknowledge the financial support from Brazilian funding agencies Funda?c?o de Amparo à Pesquisa do Estado de S?o Paulo (FAPESP), Conselho Nacional de Desenvolvimento Científico e Tecnológico(CNPq), Coordena?c?o de Aperfei?coamento de Pessoal de Nível Superior (CAPES), National Natural Science Foundation of China(NNSFC)under Grant Nos.11805166 and 11873001, and Nature Science Fund of Chongqing under Grant No.cstc2018jcyjAX0767.

    国产成人a∨麻豆精品| 欧美97在线视频| 一边摸一边抽搐一进一小说| 日韩成人av中文字幕在线观看| 午夜福利在线在线| 麻豆一二三区av精品| 国产在视频线在精品| 免费观看性生交大片5| 国产国拍精品亚洲av在线观看| 最近中文字幕2019免费版| 大香蕉久久网| 尤物成人国产欧美一区二区三区| 亚洲人与动物交配视频| 午夜精品在线福利| 亚洲综合精品二区| 久久韩国三级中文字幕| 深爱激情五月婷婷| 亚洲精品色激情综合| 热99在线观看视频| 国产极品精品免费视频能看的| 国产精品一区二区在线观看99 | 秋霞伦理黄片| 午夜久久久久精精品| 如何舔出高潮| 亚洲伊人久久精品综合 | 一边亲一边摸免费视频| 色综合亚洲欧美另类图片| 哪个播放器可以免费观看大片| 亚洲精品久久久久久婷婷小说 | 亚洲综合色惰| 亚洲精品aⅴ在线观看| 三级国产精品欧美在线观看| 精品国产一区二区三区久久久樱花 | 2021少妇久久久久久久久久久| 草草在线视频免费看| 91aial.com中文字幕在线观看| 欧美又色又爽又黄视频| 日韩一本色道免费dvd| 欧美日韩一区二区视频在线观看视频在线 | 国产亚洲精品久久久com| 国产一区二区亚洲精品在线观看| 亚洲精品久久久久久婷婷小说 | 国产熟女欧美一区二区| 国产白丝娇喘喷水9色精品| av在线亚洲专区| 中国美白少妇内射xxxbb| 国产欧美日韩精品一区二区| 丝袜喷水一区| 国产综合懂色| 久久草成人影院| 免费看av在线观看网站| 欧美丝袜亚洲另类| 99在线视频只有这里精品首页| 国产亚洲精品久久久com| 99久久精品国产国产毛片| 亚洲国产精品成人综合色| 国国产精品蜜臀av免费| 亚洲18禁久久av| 日本wwww免费看| 人人妻人人澡欧美一区二区| 亚洲av日韩在线播放| 色网站视频免费| 国产探花极品一区二区| 欧美变态另类bdsm刘玥| videos熟女内射| 免费黄网站久久成人精品| 啦啦啦啦在线视频资源| 日韩欧美精品免费久久| 91久久精品国产一区二区成人| 狂野欧美激情性xxxx在线观看| 日韩三级伦理在线观看| 久久99热这里只有精品18| 91av网一区二区| 蜜臀久久99精品久久宅男| 大话2 男鬼变身卡| 亚洲成色77777| 黄色配什么色好看| 亚洲性久久影院| 亚洲国产日韩欧美精品在线观看| 99久久精品一区二区三区| 久久精品国产鲁丝片午夜精品| 国产精品蜜桃在线观看| 啦啦啦啦在线视频资源| 99热全是精品| 国产人妻一区二区三区在| 18禁裸乳无遮挡免费网站照片| 成人午夜高清在线视频| 精品免费久久久久久久清纯| 又爽又黄a免费视频| 久久人人爽人人片av| 国产片特级美女逼逼视频| 美女cb高潮喷水在线观看| 一级毛片我不卡| 一个人观看的视频www高清免费观看| 嫩草影院新地址| or卡值多少钱| 国产高清国产精品国产三级 | 午夜久久久久精精品| 精品久久久久久久久久久久久| 又爽又黄a免费视频| 国产成人a区在线观看| 成人一区二区视频在线观看| 少妇的逼水好多| 国产午夜精品论理片| 一个人看视频在线观看www免费| 国产精品.久久久| 成人亚洲欧美一区二区av| 人体艺术视频欧美日本| 欧美日韩一区二区视频在线观看视频在线 | 女人久久www免费人成看片 | 日日干狠狠操夜夜爽| 毛片一级片免费看久久久久| 成人无遮挡网站| 国内精品宾馆在线| 简卡轻食公司| 建设人人有责人人尽责人人享有的 | 国内少妇人妻偷人精品xxx网站| 国产精品乱码一区二三区的特点| 天天一区二区日本电影三级| 久久久久久久久久久免费av| 亚洲电影在线观看av| 22中文网久久字幕| 日韩人妻高清精品专区| 在线观看av片永久免费下载| 国产白丝娇喘喷水9色精品| av福利片在线观看| 两个人的视频大全免费| 日韩欧美精品v在线| 99久久精品热视频| 国产精品一区二区三区四区久久| 日本猛色少妇xxxxx猛交久久| 国产高清国产精品国产三级 | 嫩草影院新地址| 欧美xxxx黑人xx丫x性爽| 老司机影院成人| 欧美丝袜亚洲另类| 欧美高清性xxxxhd video| 日韩欧美国产在线观看| 啦啦啦韩国在线观看视频| 亚洲国产日韩欧美精品在线观看| 精品酒店卫生间| 你懂的网址亚洲精品在线观看 | 日本一二三区视频观看| 精品无人区乱码1区二区| 午夜精品国产一区二区电影 | 大话2 男鬼变身卡| 国产午夜精品久久久久久一区二区三区| 亚洲欧美日韩无卡精品| 最近最新中文字幕免费大全7| 色综合站精品国产| 欧美性感艳星| 天天躁日日操中文字幕| 亚洲四区av| 99热全是精品| 九九在线视频观看精品| av在线蜜桃| 2021天堂中文幕一二区在线观| 亚洲欧洲国产日韩| 高清日韩中文字幕在线| 激情 狠狠 欧美| 亚洲婷婷狠狠爱综合网| 国产麻豆成人av免费视频| 极品教师在线视频| 黄色欧美视频在线观看| 国产成人freesex在线| 日本免费在线观看一区| 亚洲精品乱码久久久久久按摩| 亚洲精品日韩av片在线观看| 免费不卡的大黄色大毛片视频在线观看 | 看十八女毛片水多多多| 欧美精品一区二区大全| 91av网一区二区| 亚洲欧洲日产国产| 69人妻影院| 大香蕉97超碰在线| 校园人妻丝袜中文字幕| 亚洲国产欧美人成| 卡戴珊不雅视频在线播放| 老司机影院成人| 久热久热在线精品观看| 国产日韩欧美在线精品| 日韩,欧美,国产一区二区三区 | 亚洲国产精品成人综合色| 能在线免费观看的黄片| 欧美成人免费av一区二区三区| 免费人成在线观看视频色| 欧美不卡视频在线免费观看| 中文字幕av成人在线电影| 色噜噜av男人的天堂激情| 九色成人免费人妻av| 综合色丁香网| 久久久久网色| 亚洲美女搞黄在线观看| 午夜免费男女啪啪视频观看| 国语自产精品视频在线第100页| 欧美精品一区二区大全| 婷婷六月久久综合丁香| 亚洲最大成人av| 春色校园在线视频观看| 免费看日本二区| 国产淫语在线视频| 亚洲经典国产精华液单| 自拍偷自拍亚洲精品老妇| 亚洲av免费在线观看| 91在线精品国自产拍蜜月| 国产淫片久久久久久久久| 亚洲美女搞黄在线观看| 国产激情偷乱视频一区二区| 国语自产精品视频在线第100页| 神马国产精品三级电影在线观看| 高清日韩中文字幕在线| 免费大片18禁| 亚洲欧美一区二区三区国产| 亚洲美女搞黄在线观看| 在线播放无遮挡| 久久精品91蜜桃| 26uuu在线亚洲综合色| or卡值多少钱| 久久久色成人| 中文乱码字字幕精品一区二区三区 | 久久人人爽人人爽人人片va| 69av精品久久久久久| 欧美xxxx性猛交bbbb| 国产高清三级在线| 国产色婷婷99| 免费一级毛片在线播放高清视频| 九草在线视频观看| 日本黄大片高清| 中文亚洲av片在线观看爽| 男的添女的下面高潮视频| 永久免费av网站大全| 亚洲国产精品专区欧美| 欧美一区二区亚洲| 2021少妇久久久久久久久久久| 如何舔出高潮| 老司机影院毛片| 中文天堂在线官网| 热99在线观看视频| 青青草视频在线视频观看| 国产淫片久久久久久久久| 国产乱人偷精品视频| 亚洲,欧美,日韩| 日本免费一区二区三区高清不卡| 亚洲欧洲国产日韩| 一夜夜www| 午夜激情欧美在线| 国产一区有黄有色的免费视频 | 国产精品国产三级国产专区5o | 免费av毛片视频| 久久6这里有精品| 国产亚洲一区二区精品| av专区在线播放| 日本-黄色视频高清免费观看| 91av网一区二区| 18禁在线播放成人免费| 免费观看的影片在线观看| 国产久久久一区二区三区| 国产精品不卡视频一区二区| 国产精品综合久久久久久久免费| 久久久久久伊人网av| 最近中文字幕高清免费大全6| 狂野欧美白嫩少妇大欣赏| 亚洲天堂国产精品一区在线| 免费看光身美女| 国产亚洲5aaaaa淫片| 一边亲一边摸免费视频| 一级毛片我不卡| 亚洲不卡免费看| 天天一区二区日本电影三级| 久久久午夜欧美精品| 一边亲一边摸免费视频| 一区二区三区乱码不卡18| 久久久久久久久久久丰满| 亚洲最大成人中文| 2022亚洲国产成人精品| 国产成人福利小说| 日韩亚洲欧美综合| 级片在线观看| 国产精品1区2区在线观看.| 国产不卡一卡二| 我要搜黄色片| 国产精品国产三级国产专区5o | 日韩三级伦理在线观看| 午夜免费激情av| 99久久人妻综合| 国产精品国产高清国产av| 日本午夜av视频| 国产黄色小视频在线观看| 成年免费大片在线观看| 波多野结衣巨乳人妻| 国产乱人视频| 久久精品影院6| 久久久色成人| 日本一本二区三区精品| 美女脱内裤让男人舔精品视频| 少妇丰满av| 亚洲aⅴ乱码一区二区在线播放| 国产一级毛片在线| 国产伦一二天堂av在线观看| 男人舔奶头视频| 久久久久久久午夜电影| 狂野欧美白嫩少妇大欣赏| av专区在线播放| 91久久精品电影网| 国产一区亚洲一区在线观看| 能在线免费看毛片的网站| 久久精品国产亚洲av天美| 国产毛片a区久久久久| 男人舔女人下体高潮全视频| 两个人视频免费观看高清| 亚洲欧美精品综合久久99| 亚洲无线观看免费| 亚洲欧美精品专区久久| 我要看日韩黄色一级片| 少妇熟女欧美另类| 免费一级毛片在线播放高清视频| 午夜福利高清视频| 久久人人爽人人爽人人片va| 久久久国产成人免费| 亚洲丝袜综合中文字幕| 亚洲欧美精品自产自拍| 黑人高潮一二区| 好男人在线观看高清免费视频| 直男gayav资源| 高清午夜精品一区二区三区| 日韩欧美三级三区| 国产老妇伦熟女老妇高清| 99久久精品一区二区三区| 日本免费一区二区三区高清不卡| 久久久久久国产a免费观看| 国产美女午夜福利| 日韩一区二区视频免费看| 国内少妇人妻偷人精品xxx网站| 色哟哟·www| 免费黄色在线免费观看| 中国美白少妇内射xxxbb| 日本色播在线视频| 插逼视频在线观看| 看非洲黑人一级黄片| 一二三四中文在线观看免费高清| 欧美成人午夜免费资源| 老司机影院毛片| 国产精品,欧美在线| 美女cb高潮喷水在线观看| 成人美女网站在线观看视频| 国产免费一级a男人的天堂| 亚洲欧洲日产国产| 非洲黑人性xxxx精品又粗又长| 中国美白少妇内射xxxbb| 国产真实伦视频高清在线观看| 免费观看性生交大片5| 亚洲人成网站在线播| h日本视频在线播放| 国产精品久久久久久精品电影| 国产精华一区二区三区| 亚洲av成人精品一区久久| 我的老师免费观看完整版| 国产av不卡久久| 在线免费观看不下载黄p国产| 国产成人午夜福利电影在线观看| 欧美日韩精品成人综合77777| 男女视频在线观看网站免费| 老师上课跳d突然被开到最大视频| 日本黄大片高清| 成年女人永久免费观看视频| 在线免费观看不下载黄p国产| 尾随美女入室| 男人舔女人下体高潮全视频| 日韩高清综合在线| 可以在线观看毛片的网站| 日日啪夜夜撸| 日本av手机在线免费观看| 免费黄网站久久成人精品| 全区人妻精品视频| 麻豆国产97在线/欧美| 日韩一区二区三区影片| 少妇高潮的动态图| 色综合亚洲欧美另类图片| 99久国产av精品国产电影| 可以在线观看毛片的网站| 2022亚洲国产成人精品| 日本av手机在线免费观看| 日本三级黄在线观看| 国产成人精品久久久久久| 成年女人永久免费观看视频| 国产精品国产三级国产专区5o | 日韩强制内射视频| 免费看日本二区| 波野结衣二区三区在线| 久久久久久久久久黄片| 亚洲欧美中文字幕日韩二区| 久久久久免费精品人妻一区二区| 免费观看a级毛片全部| 国模一区二区三区四区视频| 国产欧美日韩精品一区二区| 亚洲国产精品sss在线观看| 天天躁日日操中文字幕| av在线老鸭窝| 蜜臀久久99精品久久宅男| 美女内射精品一级片tv| 精品久久久噜噜| 自拍偷自拍亚洲精品老妇| 欧美激情在线99| 真实男女啪啪啪动态图| 午夜福利高清视频| 国产黄片视频在线免费观看| 九九爱精品视频在线观看| 国产女主播在线喷水免费视频网站 | 精品不卡国产一区二区三区| 亚洲av福利一区| 亚洲一区高清亚洲精品| 国产伦在线观看视频一区| 我的女老师完整版在线观看| 国产精品美女特级片免费视频播放器| 麻豆成人av视频| 麻豆国产97在线/欧美| 晚上一个人看的免费电影| 熟女电影av网| 天堂√8在线中文| 亚洲精品456在线播放app| 欧美区成人在线视频| 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| 国产av码专区亚洲av| 97在线视频观看| 韩国高清视频一区二区三区| 三级经典国产精品| 精品久久国产蜜桃| 亚洲国产成人一精品久久久| 日本免费在线观看一区| 成人漫画全彩无遮挡| 亚洲国产精品国产精品| 国产三级中文精品| 久久午夜福利片| 国产在线男女| 亚洲怡红院男人天堂| 日本一二三区视频观看| 午夜免费男女啪啪视频观看| 久久精品夜夜夜夜夜久久蜜豆| 精品一区二区三区视频在线| 日韩国内少妇激情av| 精品一区二区免费观看| 国产精品一区www在线观看| 国产成人freesex在线| 九色成人免费人妻av| 99热这里只有精品一区| 日本免费一区二区三区高清不卡| 免费观看性生交大片5| 欧美97在线视频| 少妇被粗大猛烈的视频| 免费观看的影片在线观看| 久久久久久久久久成人| 亚洲av福利一区| 极品教师在线视频| 国产人妻一区二区三区在| 久久久欧美国产精品| 日韩中字成人| 最近的中文字幕免费完整| 岛国毛片在线播放| 国产亚洲午夜精品一区二区久久 | 精品人妻偷拍中文字幕| 视频中文字幕在线观看| 黄色欧美视频在线观看| 哪个播放器可以免费观看大片| 乱系列少妇在线播放| 色网站视频免费| 国产精品日韩av在线免费观看| 国产精品久久久久久久久免| 国产精品一区二区性色av| 亚洲欧美日韩卡通动漫| 亚洲丝袜综合中文字幕| 日韩中字成人| 国产成人aa在线观看| 一个人看的www免费观看视频| 久久久久久久久大av| 最后的刺客免费高清国语| 欧美人与善性xxx| 欧美精品国产亚洲| 高清毛片免费看| 久久精品人妻少妇| 久久久久九九精品影院| 特级一级黄色大片| 国产成人福利小说| av黄色大香蕉| 免费观看a级毛片全部| 少妇熟女欧美另类| 亚洲中文字幕一区二区三区有码在线看| 狠狠狠狠99中文字幕| 麻豆一二三区av精品| 色吧在线观看| 欧美成人精品欧美一级黄| 天天躁夜夜躁狠狠久久av| 一区二区三区免费毛片| 国产免费男女视频| 1024手机看黄色片| 国产精品1区2区在线观看.| 超碰av人人做人人爽久久| 联通29元200g的流量卡| 草草在线视频免费看| 亚洲一区高清亚洲精品| 精品一区二区三区人妻视频| 男女视频在线观看网站免费| 国产午夜精品论理片| 精品久久国产蜜桃| 老师上课跳d突然被开到最大视频| 国产 一区精品| 国产中年淑女户外野战色| 成年女人看的毛片在线观看| av卡一久久| 精品少妇黑人巨大在线播放 | 久久精品国产鲁丝片午夜精品| 国产亚洲5aaaaa淫片| 最近2019中文字幕mv第一页| 日本黄色视频三级网站网址| 国产免费视频播放在线视频 | 美女被艹到高潮喷水动态| 国产成年人精品一区二区| 日本三级黄在线观看| 亚洲欧美日韩无卡精品| 欧美xxxx性猛交bbbb| 亚洲第一区二区三区不卡| 性插视频无遮挡在线免费观看| 国产精品乱码一区二三区的特点| 男女国产视频网站| 亚洲在线观看片| 亚洲av男天堂| 欧美成人午夜免费资源| 亚洲乱码一区二区免费版| 精品无人区乱码1区二区| 人妻夜夜爽99麻豆av| 97热精品久久久久久| 人人妻人人澡人人爽人人夜夜 | 亚洲精品,欧美精品| 能在线免费观看的黄片| 亚洲国产高清在线一区二区三| 国模一区二区三区四区视频| 黄色一级大片看看| av在线天堂中文字幕| 日韩欧美在线乱码| 亚洲内射少妇av| 国产亚洲av片在线观看秒播厂 | www.av在线官网国产| 欧美性感艳星| 中文资源天堂在线| 欧美性猛交黑人性爽| 青春草视频在线免费观看| 亚洲一级一片aⅴ在线观看| 欧美成人午夜免费资源| 少妇被粗大猛烈的视频| 国产精品一二三区在线看| 日日啪夜夜撸| 99热6这里只有精品| 亚洲国产高清在线一区二区三| 国产精品久久久久久av不卡| 女人久久www免费人成看片 | 精品久久久久久久人妻蜜臀av| 亚洲国产欧洲综合997久久,| 观看免费一级毛片| 日韩制服骚丝袜av| 欧美另类亚洲清纯唯美| 亚洲精品成人久久久久久| 超碰97精品在线观看| 中文欧美无线码| 亚洲精品456在线播放app| 黄片无遮挡物在线观看| 色视频www国产| 中文欧美无线码| 女人被狂操c到高潮| 只有这里有精品99| 少妇熟女欧美另类| 蜜臀久久99精品久久宅男| 人妻系列 视频| 国产极品天堂在线| 精品无人区乱码1区二区| 日日啪夜夜撸| 桃色一区二区三区在线观看| 亚洲乱码一区二区免费版| 久久精品国产99精品国产亚洲性色| 人人妻人人澡欧美一区二区| 国产精品国产三级国产专区5o | 久久久久网色| 午夜福利网站1000一区二区三区| 好男人在线观看高清免费视频| 麻豆久久精品国产亚洲av| 国产淫片久久久久久久久| 91久久精品电影网| 亚洲欧美一区二区三区国产| 亚洲精品亚洲一区二区| 韩国高清视频一区二区三区| 欧美日韩在线观看h| 嘟嘟电影网在线观看| 国产一区二区在线观看日韩| 精品久久久久久成人av| 亚洲怡红院男人天堂| 日韩欧美国产在线观看| 亚洲精品乱久久久久久| 天堂网av新在线| 国产成人福利小说| 精品久久久久久久人妻蜜臀av| 中文字幕熟女人妻在线| 国产伦在线观看视频一区| 嫩草影院新地址| 少妇高潮的动态图| 国产精品久久视频播放| 美女脱内裤让男人舔精品视频| 春色校园在线视频观看| 国产精品一区二区性色av| 熟女人妻精品中文字幕| 神马国产精品三级电影在线观看| 三级国产精品片| 亚洲欧美日韩无卡精品| 午夜精品国产一区二区电影 | 岛国在线免费视频观看| 人妻制服诱惑在线中文字幕| 亚洲国产最新在线播放|