Abstract
To facilitate the numerical analysis of particle methods, we derive truncation error estimates for the approximate operators in a generalized particle method. Here, a generalized particle method is defined as a meshfree numerical method that typically includes other conventional particle methods, such as smoothed particle hydrodynamics or moving particle semi-implicit methods. A new regularity of discrete parameters is proposed via two new indicators based on the Voronoi decomposition of the domain along with two hypotheses of reference weight functions. Then, truncation error estimates are derived for an interpolant, approximate gradient operator, and approximate Laplace operator in the generalized particle method. The convergence rates for these estimates are determined based on the frequency with which they appear in the regularity and hypotheses. Finally, the estimates are computed numerically, and the results are shown to be in good agreement with the theoretical results.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Particle methods, such as the smoothed particle hydrodynamics (SPH) [10, 18, 19] and moving particle semi-implicit (MPS) methods [15, 16, 29], are numerical methods for solving partial differential equations that are based on points called particles distributed in a domain. In such methods, an interpolant and several approximate differential operators are defined in terms of linear combinations of weighted interactions between neighboring particles. When such methods are applied to partial differential equations, the equations are effectively discretized in space. As the discretization procedure does not require mesh generation in the domain, particle methods can be applied to moving boundary problems, such as the deformation and destruction of structures [5, 22] and flow problems associated with free surfaces [21, 23].
The accuracy of particle methods has been widely researched. From an engineering perspective, many studies have been conducted into the convergence of such methods in practical applications, such as Amicarelli [1, 2], Fulk [9], and Quinlan et al. [25]. On the other hand, few studies in the literature have presented numerical analyses of these methods from a mathematical perspective. In the 1980s, Mas-Gallic and Raviart [20] and Raviart [26] provided error estimates for particle methods when applied to parabolic and hyperbolic partial differential equations on unbounded domains. In the 2000s, Ben Moussa and Via [4] and Ben Moussa [3] provided error estimates of nonlinear conservation laws on bounded domains. In their work, the time integrations of the particle positions and volumes were obtained by solving the differential equations with respect to advection fields. However, as their method is only applicable to problems described by solvable differential equations, it cannot be used with other problems, such as those involving the Navier–Stokes equations.
Sometime later, Ishijima and Kimura [13] developed a truncation error estimate for an approximate gradient operator in the MPS method. By introducing a regularity for particle distributions based on an indicator called the equivolume partition radius, they determined the conditions that depend solely on the space distributions of the particles. However, a practical limitation is that the indicator cannot be computed.
In previous works, we established truncation error estimates for an interpolant, approximate gradient operator, and approximate Laplace operator of a generalized particle method in which the particle volumes were given as Voronoi volumes [11, 12]. A generalized particle method is a numerical method that typically includes conventional particle methods, such as the SPH and MPS methods. In previous studies, we derived truncation error estimates by introducing a regularity using an indicator known as the covering radius, which is used in the numerical analysis of meshfree methods based on moving least-square methods and radial basis functions [17, 27, 30]. Although the formulations and conditions in those works are computable, they are difficult to deploy in practical computations as the computational costs associated with particle volumes based on Voronoi decomposition are high.
The focus of the current work was to analyze particle methods under more practical conditions by extending our results to cases with commonly used particle volumes. We also introduce another indicator of particle volumes, which we refer to as a Voronoi deviation, that represents the deviation between particle volumes and Voronoi volumes. Then, utilizing the Voronoi deviation, we extend the regularity and introduce two hypotheses of reference weight functions. Using the regularity and hypotheses, we derive truncation error estimates of the interpolant, approximate gradient operator, and approximate gradient operator of the generalized particle method. Finally, we numerically analyze our estimates and compare the results to those from the theory.
The remainder of this paper is organized as follows. The interpolant and approximate operators of the generalized particle method are introduced in Sect. 2. A regularity describing the family of discrete parameters is discussed in Sect. 3, after which we propose our primary theorem with respect to the truncation error estimates and provide some corollaries. Then, the primary theorem is proven in Sect. 4, numerical results are detailed in Sect. 5, and some concluding remarks are outlined in Sect. 6.
In the remainder of this section, we describe some notation and define some relevant function spaces. Let \({\mathbb {R}}^{+}\), \({\mathbb {R}}^{+}_0\), and \({\mathbb {N}}_0\) be the set of positive real numbers, the set of nonnegative real numbers, and the set of nonnegative integers, respectively. Let \(d\) be the dimension of a space. Let \({\mathbb {A}}^{d}\) be the set of all \(d\)-dimensional multi-indices. For \(x =(x_1,x_2,\ldots ,x_d)^{\mathrm{T}}\in {\mathbb {R}}^{d}\) and \(\alpha = (\alpha _1,\alpha _2,\ldots ,\alpha _d)^{\mathrm{T}}\in {\mathbb {A}}^{d}\), \(x^\alpha \) is defined as \(x^\alpha =x_1^{\alpha _1}x_2^{\alpha _2}\cdots x_d^{\alpha _d}\). If there is no ambiguity, the symbol \(|\cdot |\) is used to denote the following: |x| denotes the Euclidean norm for \(x\in {\mathbb {R}}^{d}\); |S| denotes the volume of S for \(S\subset {\mathbb {R}}^{d}\); \(|\alpha |\) denotes \(|\alpha |:=\alpha _1+\alpha _2+\cdots +\alpha _d\) for \(\alpha \in {\mathbb {A}}^{d}\). For \(S\subset {\mathbb {R}}^{d}\), let \(\text {diam}(S)\) be \(\text {diam}(S):=\sup \left\{ |x-y|;\,x,y\in S\right\} \). For \(S\subset {\mathbb {R}}^{d}\), let \(C(\overline{S})\) be the space of real continuous functions defined in \(\overline{S}\) with the norm \(\left\| \,\cdot \,\right\| _{C(\overline{S})}\) defined as
For \(S\subset {\mathbb {R}}^{d}\) and \(\ell \in {\mathbb {N}}\), let \(C^{\ell }(\overline{S})\) be the space of functions in \(C(\overline{S})\) with derivatives up to the \(\ell \)th order with its seminorm \(\left| \,\cdot \,\right| _{C^{\ell }(\overline{S})}\) and norm \(\left\| \,\cdot \,\right\| _{C^{\ell }(\overline{S})}\) defined as
respectively. Here \(D^\alpha v:=\partial _1^{\alpha _1} \partial _2^{\alpha _2} \dots \partial _d^{\alpha _d}v\) with multi-index \(\alpha =(\alpha _1,\alpha _2,\ldots ,\alpha _d)\).
2 Approximate operators in a generalized particle method
Let \({\varOmega }\) be a bounded domain in \({\mathbb {R}}^{d}\). Let \(H\) be a fixed positive number. For \({\varOmega }\) and \(H\), we define extended domain \({\varOmega }_H\) as
For \(N\in {\mathbb {N}}\), we define a particle distribution \({\mathcal {X}}_{N}\) and particle volume set \({{\mathcal {V}}}_{N}\) as
respectively. We refer to \(x_{i}\in {\mathcal {X}}_{N}\) and \(V_{i}\in {{\mathcal {V}}}_{N}\) as a particle and particle volume, respectively. An example of the particle distribution \({\mathcal {X}}_{N}\) in \({\varOmega }_H\,(\subset {\mathbb {R}}^2)\) is shown in Fig. 1.
We define an admissible reference weight function set \({\mathcal {W}}\) as
we refer to \(w\in {\mathcal {W}}\) as a reference weight function, and we define the influence radius \(h_{N}\in {\mathbb {R}}\) as satisfying \(0<h_{N}<H\) and \(h_{N}\rightarrow 0\,(N\rightarrow \infty )\). If there is no ambiguity, we denote \(h_{N}\) as \(h\). For reference weight function \(w\) and influence radius \(h\), we define the weight function \(w_{h}\in C({\mathbb {R}}^{+}_0)\) as
Note that the weight function \(w_{h}\) satisfies
and is absolutely continuous.
For \(v\in C(\overline{{\varOmega }}_{H})\), we define interpolant \({\varPi }_h{}\), approximate gradient operator \(\nabla _h{}\), and approximate Laplace operator \({\varDelta }_h{}\) as
respectively. Here, for \(x\in {\mathbb {R}}^{d}\) and \(r\in {\mathbb {R}}^{+}\cup \{\infty \}\), \({\varLambda }_0(x,r)\) and \({\varLambda }(x,r)\) are index sets of particles defined as
respectively.
As discussed later in Appendix 1, the approximate operators (2), (3), and (4) indicate a wider class of approximate operators of particle methods than those in the SPH and MPS methods. Therefore, we refer to the approximate operators (2), (3), and (4) as generalized approximate operators and to a particle method that uses them as a generalized particle method.
3 Truncation error estimates of approximate operators
We first introduce a regularity of discrete parameters. Let \(\{\sigma _{i}\}\) be the Voronoi decomposition of \({\varOmega }_H\) associated with the particle distribution \({\mathcal {X}}_{N}\), where \(\sigma _{i}\) is the Voronoi region defined as
We define a particle volume decomposition \({\varXi }=\{\xi _{i}\}\) as a decomposition of \({\varOmega }_H\) satisfying
An example of the Voronoi decomposition of \({\varOmega }_H\) associated with the particle distribution \({\mathcal {X}}_{N}\) is shown in Fig. 2. We define a covering radius \(r_{N}\) for particle distribution \({\mathcal {X}}_{N}\) as
Moreover, we define a Voronoi deviation \(d_{N}\) for the particle distribution \({\mathcal {X}}_{N}\) and the particle volume set \({{\mathcal {V}}}_{N}\) as
with
Then, we define a regularity for a family consisting of a particle distribution \({\mathcal {X}}_{N}\), particle volume set \({{\mathcal {V}}}_{N}\), and influence radius \(h\) as follows:
Definition 1
A family \(\{({\mathcal {X}}_{N}, {{\mathcal {V}}}_{N}, h_{N})\}_{N\rightarrow \infty }\) is said to be regular with order \(m\,(m\ge 1)\) if there exists a positive constant \(c_0\) such that
Remark 1
As shown in Fig. 3, the covering radius \(r_{N}\) becomes large in the case of a particle distribution with both dense and sparse regions. Therefore, the covering radius \(r_{N}\) can be considered as an indicator representing the uniformness of particle distribution \({\mathcal {X}}_{N}\).
Remark 2
A Voronoi deviation \(d_{N}\) equals zero if and only if the particle volumes are given as the Voronoi volume (\(V_{i}=\left| \sigma _{i}\right| \)). Moreover, the Voronoi deviation \(d_{N}\) becomes large if the particle volumes are given as values far from the Voronoi volumes. Therefore, the Voronoi deviation \(d_{N}\) can be regarded as an indicator of the deviation between the particle volume set and the Voronoi volume set.
Remark 3
For a given family \(\{({\mathcal {X}}_{N}, {{\mathcal {V}}}_{N}, h_{N})\}_{N\rightarrow \infty }\) and given constant \(m\, (m\ge 1)\), it is possible to determine whether or not the family is regular with order \(m\) as the covering radius \(r_{N}\) and Voronoi deviation \(d_{N}\) are absolutely computable, as shown in Appendix 2.
Next, we introduce two hypotheses of reference weight function \(w\):
Hypothesis 1
For \(n\in {\mathbb {N}}\), the reference weight function \(w\) satisfies for all \(\alpha \in {\mathbb {A}}^{d}\) with \(1\le |\alpha |\le n\),
Hypothesis 2
For \(k\in {\mathbb {N}}_0\), the reference weight function \(w\) satisfies
where for \(j\in {\mathbb {N}}_0\), \(w^{(j)}(r): (0,\infty )\rightarrow {\mathbb {R}}\) is defined as
and \((w^{(k)})^\prime \) is \(\mathrm{d}w^{(k)}/\mathrm{d}r\).
Remark 4
All reference functions \(w\in {\mathcal {W}}\) satisfy Hypothesis 1 with \(n=1\). Moreover, for all \(n\in {\mathbb {N}}\) and \(k\in {\mathbb {N}}\), reference weight functions satisfying Hypothesis 1 with \(n\) and Hypothesis 2 with \(k\) can be constructed as shown in Appendix 3.
We now state a theorem that defines truncation error estimates of approximate operators in the generalized particle method with a continuous norm:
Theorem 3
Suppose that a family\(\{({\mathcal {X}}_{N},{{\mathcal {V}}}_{N},h_{N})\}_{N\rightarrow \infty }\)is regular with order\(m\,(m\ge 1)\)and that reference weight function\(w\)satisfies Hypothesis 1with\(n\). Then, there exists a positive constant\(c\)independent of\(N\)such that
In addition, if\(w\in {\mathcal {W}}\)satisfies Hypothesis 2with\(k=0\), then we have
and if\(w\in {\mathcal {W}}\)satisfies Hypothesis 2with\(k=1\), then we have
The proof of Theorem 3 is presented in the next section. As shown in the corollaries in Appendix 1, the approximate operators commonly used in the SPH and MPS methods are valid for Theorem 3 under appropriate settings.
4 Proof of truncation error estimates
The following notation will be used in the subsequent proof of Theorem 3. Hereafter, let \(c\) be a generic positive constant independent of \(N\) (allowed dependence on the fixed positive parameter \(H\)). For \(\alpha \in {\mathbb {A}}^{d}\), set \(I_{\alpha }\) as
For \(\alpha \in {\mathbb {A}}^{d}\) and \(\ell \in {\mathbb {N}}\), set \(I_{\alpha ,\ell }\) as
For \(\ell \in {\mathbb {N}}\), set \(J_{\ell }\) as
We now present the following lemma.
Lemma 1
Suppose that\(w\in {\mathcal {W}}\)satisfiesHypothesis 1with\(n\). Then, there exists a positive constant\(c\)independent of\(N\)such that
Proof
First, we prove (12). We fix \(x\in \overline{{\varOmega }}\). Then, let \(B(x,r)\) be the open ball in \({\mathbb {R}}^{d}\) with center \(x\) and radius r, i.e.,
From \(h<H\), we have \(B(x,h)\subset {\varOmega }_H\). Then, for all \(v\in C^{\ell +1}(\overline{{\varOmega }}_{H}) \, (\ell \in {\mathbb {N}})\) and \(x_{i}\in B(x,h)\), we obtain the Taylor expansion of \(v\) as
From (2) and (15) with \(\ell =n\), we have
Moreover, by Hypothesis 1, we have
Because
we have
Moreover, we have
Therefore, from (16), (18), and (19), we obtain (12).
Next, we prove (13). From (3) and (15) with \(\ell =n+1\), we have
Because for \(\beta \in {\mathbb {A}}^{d}\) with \(|\beta |=2\),
we have
Hypothesis 1 with \(n\) yields
From (17), we have
Moreover, we have
Therefore, from (23), (24), and (25), we obtain (13).
Finally, we prove (14). From (4) and (15) with \(\ell =n+2\), we have
From (20), we have
Hypothesis 1 with \(n\) yields
Therefore, we have
From (17), we have
Moreover, we have
Therefore, from (26), (27), and (28), we obtain (14).
Next, we show estimates of \(I_{\alpha }\), \(I_{\alpha ,\ell }\), and \(J_{\ell }\).
Lemma 2
There exists a positive constant \(c\) independent of \(N\) such that
Proof
We arbitrarily fix \(x\in \overline{{\varOmega }}\), \(\alpha \in {\mathbb {A}}^{d}\), and particle volume decomposition \({\varXi }=\{\xi _{i}\mid i=1,2,\ldots ,N\}\) and split \(I_{\alpha }\) into
with
Then, we estimate \(E_{1}\), \(E_{2}\), and \(E_{3}\).
First, we estimate \(E_{1}\). Because
we can rewrite \(E_{1}\) as
From
we obtain
From
we have
Because \(w_{h}\) is absolutely continuous, we have
for all \(y, z\in {\mathbb {R}}^{d}\). Here, \(w^\prime \) and \(w_{h}^\prime \) are \(\mathrm{d}w/\mathrm{d}r\) and \(\mathrm{d}w_{h}/\mathrm{d}r\), respectively. Moreover, we have
From (33), (34), and (35), we have
Therefore, from (32) and (36), we obtain
Next, we estimate \(E_{2}\). Because \(\text{ supp }(w_{h}) = [0,h]\) and \(\sigma _{j}\subset B(x_{j},r_{N})\), we have
From (37), we have
Moreover, from (34) and (35), we have
Therefore, from (31) and (38), we obtain
Finally, we estimate \(E_{3}\). Because
we can rewrite \(E_{3}\) as
Because \(E_{3}=0\) when \(|\alpha |=0\), we estimate when \(|\alpha |\ge 1\). Let \(\beta _k\,(k=1,2,\ldots ,|\alpha |)\) be \(d\)-dimensional multi-indices with satisfying
Then, we have, for all \(y,z\in {\mathbb {R}}^{d}\),
By \(\text{ supp }(w_{h}) = [0,h]\) and \(\sigma _{j}\subset B(x_{j},r_{N})\), if \(j\not \in {\varLambda }_0(x,h+r_{N})\), then
Moreover, from \(w\in {\mathcal {W}}\subset C({\mathbb {R}}^{+}_0)\), we have
From (35), (41), and (42), we have
Therefore, from (40), (43), and \(h\le H\), we obtain
From the estimates of \(E_{1}\), \(E_{2}\), and \(E_{3}\), we obtain
Because \({\varXi }\) is arbitrary, we establish (29).
Lemma 3
Suppose that a reference weight function\(w\)satisfiesHypothesis 2with\(k\). Then, there exists a positive constant\(c\)independent of\(N\)such that for all\(\alpha \in {\mathbb {A}}^{d}\)and\(\ell \in {\mathbb {N}}\)with\(1\le \ell -k\le |\alpha |\),
Proof
We arbitrarily fix \(x\in \overline{{\varOmega }}\), \(\alpha \in {\mathbb {A}}^{d}\), particle volume decomposition \({\varXi }=\{\xi _{i}\mid i=1,2,\ldots ,N\}\), and \(\ell \in {\mathbb {N}}\) with \(1\le \ell -k\le |\alpha |\) and split \(I_{\alpha ,\ell }\) into
with
Then, we estimate \(E_{4}\), \(E_{5}\), and \(E_{6}\).
First, we estimate \(E_{4}\) and set \(w^{(k)}\) as (8) and \(w^{(k)}_h\) as
Then, from (30), we can rewrite \(E_{4}\) as
Because
we obtain
From \(\text{ supp }(w^{(k)}_h) = [0,h]\), we have
Thus, we obtain
Using an argument similar to (34), if \(w\) satisfies Hypothesis 2 with \(k\), then for all \(y, z\in {\mathbb {R}}^{d}\),
Next, we estimate \(E_{5}\). By using \(w^{(k)}_h\), we can rewrite \(E_{5}\) as
From (45), we obtain
By \(\text{ supp }(w^{(k)}_h) = [0,h]\) and \(\sigma _{j}\subset B(x_{j},r_{N})\), we have
Finally, we estimate \(E_{6}\). Using \(w^{(k)}_h\), we can rewrite \(E_{6}\) as
where \(\xi ^{*}_{i}(x)\) is
For \(\alpha \in {\mathbb {A}}^{d}\), let \(\beta _j\,(j=1,2,\ldots ,|\alpha |)\) be \(d\)-dimensional multi-indices satisfying
Let \(\beta _j^*\,(j=0,1,\ldots ,|\alpha |)\) be \(d\)-dimensional multi-indices defined as
For all \(y, z\in {\mathbb {R}}^{d}{\setminus }\{0\}\), when \(|\alpha |=\ell -k\), we have
Moreover, from (39) and (49), when \(|\alpha |>\ell -k\), we have
Therefore, when \(|\alpha |\ge \ell -k\), we have for all \(y\in {\varOmega }_H{\setminus }\{x\}\) and \(i\in {\varLambda }(x,\infty )\),
From (50), we obtain
Because \(|\alpha |\ge \ell -k\), we have
Therefore, we have
Because for all \(y\in {\varOmega }_H\),
by the same procedure as (43), we have
Therefore, we obtain
From the estimates of \(E_{4}\), \(E_{5}\), and \(E_{6}\), we obtain
Because \({\varXi }\) is arbitrary, we establish (44).
Lemma 4
There exists a positive constant \(c\) independent of \(N\) such that
Proof
We arbitrarily fix \(x\in \overline{{\varOmega }}\) and particle volume decomposition \({\varXi }=\{\xi _{i}\mid i=1,2,\ldots ,N\}\), and split \(J_{\ell }\) into
with
Then, we estimate \(E_{7}\), \(E_{8}\), \(E_{9}\), and \(E_{10}\).
From (30), we can rewrite \(E_{7}\) as
For all \(y\in {\varOmega }_H\), we have
For all \(x_{i}\in {\mathcal {X}}_{N}\) and \(y\in {\varOmega }_H\), we have
From (43), (53), and \(h<H\), we obtain
From (1), we obtain
From the estimates of \(E_{7}\), \(E_{8}\), \(E_{9}\), and \(E_{10}\), we obtain
Because \({\varXi }\) is arbitrary, we establish (51).
Using the lemmas defined above, we now prove Theorem 3.
Proof of Theorem 3
By Lemmas 1, 2, and 4, we have for all \(v\in C^{n+1}(\overline{{\varOmega }}_{H})\)
Moreover, by Lemmas 1, 3, and 4, when \(w\) satisfies Hypothesis 2 with \(k=0\), we have for all \(v\in C^{n+2}(\overline{{\varOmega }}_{H})\)
and when \(w\) satisfies Hypothesis 2 with \(k=1\) for all \(v\in C^{n+3}(\overline{{\varOmega }}_{H})\),
Because the family \(\{({\mathcal {X}}_{N}, {{\mathcal {V}}}_{N}, h_{N})\}_{N\rightarrow \infty }\) is regular, by applying (7) to (54), (55), and (56), we obtain (9), (10), and (11), respectively. We now conclude the proof of Theorem 3.
5 Numerical results
Set \({\varOmega }=(0,1)^2\) and \(H=0.1\). Then, \({\varOmega }_H=(-0.1,1.1)^2\). We now compute the truncation errors of \(v:{\varOmega }_H\rightarrow {\mathbb {R}}\), which are defined as \(v(x,y)=\sin (2\pi (x+y))\). Particle distribution \({\mathcal {X}}_{N}\) is set as
where \(\Delta x\) is taken by \(2^{-5}, 2^{-6}, \ldots , 2^{-12}\) and \(\eta _{i j}^{(k)}\,(i,j\in {\mathbb {Z}}, k=1,2)\) are random numbers satisfying \(|\eta _{i j}^{(k)}|< 1/4\). Particle distribution \({\mathcal {X}}_{N}\) with \(\Delta x=2^{-5}\) is shown in Fig. 4. Particle volume set \({{\mathcal {V}}}_{N}\) is defined as
For \(m=1,3,5\), the influence radius \(h_{N}\) is set as
Note that if \(\Delta x=2^{-5}\), then \(h=2.6\times 2^{-5}\) for all m. Using the discrete parameters above, the covering radius \(r_{N}\) satisfies \(r_{N}\le \sqrt{2}(1+1/4)\Delta x/2\). Moreover, the Voronoi deviation \(d_{N}\) satisfies \(d_{N}\le 64(1+\sqrt{2})\Delta x/\pi \). Therefore, the family \(\{({\mathcal {X}}_{N},{{\mathcal {V}}}_{N},h_{N})\}\) is regular with order m.
For the interpolant, we consider the following three cases of reference weight functions:
\(({\Pi 1})\) is the lowest-order polynomial function belonging to \({\mathcal {W}}\). \(({\Pi 2})\) is the cubic B-spline commonly used in the SPH method and belonging to \({\mathcal {W}}\). \(({\Pi 3})\) is the lowest-order polynomial function belonging to \({\mathcal {W}}\) that satisfies Hypothesis 1 with \(n=3\).
For the approximate gradient operator, we consider the following three cases of reference weight functions:
\(({\nabla 1})\) is the lowest-order polynomial function belonging to \({\mathcal {W}}\) that satisfies Hypothesis 2 with \(k=0\). \(({\nabla 2})\) is chosen so that the approximate gradient operator (3) with \(({\nabla 2})\) coincides with that in the SPH method with the cubic B-spline (see Appendix 1). \(({\nabla 3})\) is the lowest-order polynomial function belonging to \({\mathcal {W}}\) that satisfies Hypothesis 1 with \(n=3\) and Hypothesis 2 with \(k=0\).
For the approximate Laplace operator, we consider the following three cases of reference weight functions:
\(({\Delta 1})\) is the lowest-order polynomial function belonging to \({\mathcal {W}}\) that satisfies Hypothesis 2 with \(k=1\). \(({\Delta 2})\) is chosen so that approximate Laplace operator (4) with \(({\Delta 2})\) coincides with that in the SPH method with the cubic B-spline (see Appendix 1). \(({\Delta 3})\) is the lowest-order polynomial function belonging to \({\mathcal {W}}\) that satisfies Hypothesis 1 with \(n=3\) and Hypothesis 2 with \(k=1\).
The above settings were used in the computation of the following relative errors
Here, the discrete norm \(\left\| \cdot \right\| _{\ell ^{\infty }({\varOmega })}\) is defined as
Figure 5 shows graphs of the relative errors of (a) interpolant \({\varPi }_h{}\), (b) approximate gradient operator \(\nabla _h{}\), and (c) approximate Laplace operator \({\varDelta }_h{}\) versus the influence radius \(h_{N}\) with regular orders \(m=1, 3, 5\). In Fig. 5, the slopes of the triangles show the theoretical convergence rates obtained via Theorem 3. Table 1 lists the numerical and theoretical convergence rates obtained from the cases of \(\varDelta x=2^{-11}\) and \(2^{-12}\), where the theoretical convergence rates correspond to Theorem 3. In the case of \(m=1\), as the settings could not be applied to Theorem 3, only numerical results without convergence were obtained. In contrast, the settings in cases \(m=3\) and 5 could be applied Theorem 3; thus, the numerical results with convergence were obtained. Moreover, the approximate operators with reference weight functions satisfying Hypothesis 1 with \(n=3\) became higher convergence orders in the cases where \(m=5\) as per Theorem 3.
6 Conclusions
We analyzed truncation errors in a generalized particle method, which is a wider class of particle methods that includes commonly used methods such as the SPH and MPS methods. In our analysis, we introduced two indicators: the first was the covering radius, which represents the maximum radius of the Voronoi region associated with the particle distribution, while the second was the Voronoi deviation, which indicates the deviation between particle volumes and Voronoi volumes. With the covering radius and Voronoi deviation, we introduced a regularity of a family of discrete parameters, which includes the particle distribution, particle volume set, and influence radius associated with the number of particles. Moreover, we introduced two hypotheses of reference weight functions. With the regularity and hypotheses of reference weight functions, we established truncation error estimates for the continuous norm. The convergence rates are dependent on the regular order and order of the reference weight functions appearing in a hypothesis. Moreover, as it was possible to validate the conditions by calculation, we showed the numerical convergence orders were in good agreement with the theoretical ones.
In a forthcoming paper, we plan to establish error estimates of the generalized particle method for the Poisson and heat equations.
References
Amicarelli, A., Marongiu, J.C., Leboeuf, F., Leduc, J., Caro, J.: SPH truncation error in estimating a 3D function. Comput. Fluids 44(1), 279–296 (2011)
Amicarelli, A., Marongiu, J.C., Leboeuf, F., Leduc, J., Neuhauser, M., Fang, L., Caro, J.: SPH truncation error in estimating a 3D derivative. Int. J. Numer. Methods Eng. 87(7), 677–700 (2011)
Ben Moussa, B.: On the convergence of SPH method for scalar conservation laws with boundary conditions. Methods Appl. Anal. 13(1), 29–62 (2006)
Ben Moussa, B., Vila, J.: Convergence of SPH method for scalar nonlinear conservation laws. SIAM J. Numer. Anal. 37(3), 863–887 (2000)
Benz, W., Asphaug, E.: Simulations of brittle solids using smooth particle hydrodynamics. Comput. Phys. Commun. 87(1), 253–265 (1995)
Boissonnat, J.D., Yvinec, M.: Algorithmic Geometry. Cambridge University Press, Cambridge (1998)
Dantzig, G.B.: Linear Programming and Extensions. Princeton Landmarks in Mathematics. Princeton University Press, Princeton (1998)
Dehnen, W., Aly, H.: Improving convergence in smoothed particle hydrodynamics simulations without pairing instability. Mon. Not. R. Astron. Soc. 425(2), 1068–1082 (2012)
Fulk, D.A., Quinn, D.W.: An analysis of 1-d smoothed particle hydrodynamics kernels. J. Comput. Phys. 126(1), 165–180 (1996)
Gingold, R.A., Monaghan, J.J.: Smoothed particle hydrodynamics-theory and application to non-spherical stars. Mon. Not. R. Astron. Soc. 181, 375–389 (1977)
Imoto, Y., Tagami, D.: A truncation error estimate of the interpolants of a particle method based on the Voronoi decomposition. JSIAM Lett. 8, 29–32 (2016)
Imoto, Y., Tagami, D.: Truncation error estimates of approximate differential operators of a particle method based on the Voronoi decomposition. JSIAM Lett. 9, 69–72 (2017)
Ishijima, K., Kimura, M.: Truncation error analysis of finite difference formulae in meshfree particle methods (in Japanese). Trans. Jpn. Soc. Ind. Appl. Math. 20, 165–182 (2010)
Khayyer, A., Gotoh, H.: Enhancement of stability and accuracy of the moving particle semi-implicit method. J. Comput. Phys. 230(8), 3093–3118 (2011)
Koshizuka, S.: A particle method for incompressible viscous flow with fluid fragmentation. J. Comput. Fluid Dyn. 4, 29–46 (1995)
Koshizuka, S., Oka, Y.: Moving-particle semi-implicit method for fragmentation of incompressible fluid. Nuclear Sci. Eng. 123(3), 421–434 (1996)
Levin, D.: The approximation power of moving least-squares. Math. Comput. 67(224), 1517–1531 (1998)
Liu, M., Liu, G.: Smoothed particle hydrodynamics (SPH): an overview and recent developments. Arch. Comput. Methods Eng. 17(1), 25–76 (2010)
Lucy, L.B.: A numerical approach to the testing of the fission hypothesis. Astron. J. 82, 1013–1024 (1977)
Mas-Gallic, S., Raviart, P.: A particle method for first-order symmetric systems. Numer. Math. 51(3), 323–352 (1987)
Monaghan, J.J.: Simulating free surface flows with SPH. J. Comput. Phys. 110(2), 399–406 (1994)
Monaghan, J.J., Lattanzio, J.C.: A simulation of the collapse and fragmentation of cooling molecular clouds. Astrophys. J. 375, 177–189 (1991)
Murotani, K., Koshizuka, S., Tamai, T., Shibata, K., Mitsume, N., Yoshimura, S., Tanaka, S., Hasegawa, K., Nagai, E., Fujisawa, T.: Development of hierarchical domain decomposition explicit MPS method and application to large-scale tsunami analysis with floating objects. J. Adv. Simul. Sci. Eng. 1(1), 16–35 (2014)
Price, D.J.: Smoothed particle hydrodynamics and magnetohydrodynamics. J. Comput. Phys. 231(3), 759–794 (2012)
Quinlan, N.J., Basa, M., Lastiwka, M.: Truncation error in mesh-free particle methods. Int. J. Numer. Methods Eng. 66(13), 2064–2085 (2006)
Raviart, P.A.: An Analysis of Particle Methods. Numerical Methods in Fluid Dynamics (Como, 1983), Lect Notes Math, vol. 1127, pp. 243–324. Springer, Berlin (1985)
Schaback, R.: Error estimates and condition numbers for radial basis function interpolation. Adv. Comput. Math. 3(3), 251–264 (1995)
Shakibaeinia, A., Jin, Y.C.: A weakly compressible MPS method for modeling of open-boundary free-surface flow. Int. J. Numer. Methods Fluids 63(10), 1208–1232 (2010)
Shakibaeinia, A., Jin, Y.C.: MPS mesh-free particle method for multiphase flows. Comput. Methods Appl. Mech. Eng. 229, 13–26 (2012)
Wendland, H.: Scattered Data Approximation, Cambridge Monographs on Applied and Computational Mathematics, vol. 17. Cambridge University Press, Cambridge (2005)
Acknowledgements
We are grateful to the referee for the constructive comments provided. We would like to thank Drs. Daisuke Tagami and Hayato Waki for their helpful comments over the course of this work. This research was supported by JSPS KAKENHI Grant number 17K17585 and the JSPS A3 Foresight Program.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Appendix 1: Description of conventional particle methods by the generalized particle method
This appendix provides a description of conventional particle methods, such as the smoothed particle hydrodynamics (SPH) [18, 24] and the moving particle semi-implicit (MPS) methods [16], in the context of the generalized particle method. In the SPH method, upon using the reference weight function \(w^\mathrm{SPH}\in {\mathcal {W}}\) and parameters \(m_i, \rho _i\in {\mathbb {R}}^{+}\,(i=1,2,\ldots ,N)\), for \(v\in C(\overline{{\varOmega }}_{H})\), the approximate operators are defined as
By setting \(w=w^\mathrm{SPH}\) and \({{\mathcal {V}}}_{N}=\{V_{i}=m_i/\rho _i;\,i=1,2,\ldots ,N\}\), the generalized interpolant (2) coincides with (57). Moreover, because
by setting
and \({{\mathcal {V}}}_{N}=\{V_{i}=m_i/\rho _i;\,i=1,2,\ldots ,N\}\), (3) and (4) coincide with (58) and (59), respectively.
From Theorem 3, we obtain the following corollary that is a truncation error estimate of approximate operators (58) and (59).
Corollary 1
Suppose that parameters\(\rho _i, m_i\)satisfy
and that\(\{({\mathcal {X}}_{N}, {{\mathcal {V}}}_{N}, h_{N})\}_{N\rightarrow \infty }\)is regular with order\(m\), where\({{\mathcal {V}}}_{N}=\{\rho _i/m_i;\,i=1,2,\ldots ,N\}\). Moreover, suppose that\(w^\mathrm{SPH}\)satisfies the following conditions;
Then, there exists a positive constant\(c\)independent of\(N\)such that
Remark 5
Note that representative reference weight functions employed in the SPH method, such as the cubic B-spline, quintic B-spline, and Wendland function (5-order positive definite function) [8, 18], satisfy (60).
In the MPS method [14], upon using reference weight function \(w^\mathrm{MPS}\in {\mathcal {W}}\) and parameters \({\widehat{n}}, {\widehat{\lambda }}\in {\mathbb {R}}^{+}\) for \(v\in C^{1}(\overline{{\varOmega }}_{H})\), approximate differential operators can be defined as
Note that an interpolant is not defined in the MPS method. By setting \(w=w^\mathrm{MPS}\) and \({{\mathcal {V}}}_{N}=\{V_{i}={\widehat{n}}^{-1};\,i=1,2,\ldots ,N\}\), the approximate gradient operator (3) coincides with (61). Moreover, by setting \(w(r) = {\widehat{\lambda }}^{-1}r^2w^\mathrm{MPS}(r)\) and \({{\mathcal {V}}}_{N}=\{V_{i}={\widehat{n}}^{-1};\,i=1,2,\ldots ,N\}\), approximate Laplace operator (4) coincides with (62).
Corollary 2
Suppose that
Moreover, suppose that\(\{({\mathcal {X}}_{N}, {{\mathcal {V}}}_{N}, h_{N})\}_{N\rightarrow \infty }\)is regular with order\(m\), where\({{\mathcal {V}}}_{N}=\{V_{i}={\widehat{n}}^{-1};\,i=1,2,\ldots ,N\}\). Then, there exists a positive constant\(c\)independent of\(N\)such that
Furthermore, when \(w^\mathrm{MPS}\) satisfies Hypothesis 2 with \(k=0\),
Remark 6
Note that the reference weight function, which is commonly used in the MPS method and defined as
does not satisfy \(w^\mathrm{MPS}\in {\mathcal {W}}\). In contrast, the continuous reference weight function as introduced in [28] satisfies \(w^\mathrm{MPS}\in {\mathcal {W}}\). However, as far as we know, no reference weight functions that also satisfy Hypothesis 2 with \(k=0\) are proposed in the MPS method.
Appendix 2: Computational procedure of the indicators
This appendix introduces the procedures for computing the indicators introduced in this paper, namely, the covering radius (5) and Voronoi deviation (6).
The covering radius \(r_{N}\) can be computed as follows. As per the methods used to construct Voronoi decompositions, such as the increment method [6], we first draw the boundaries of the Voronoi region in \({\varOmega }_H\). Next, for each particle, we compute the maximum distance from particle \(x_{i}\) to the boundary of its Voronoi region \(\sigma _{i}\) (i.e., \(\max _{y\in \overline{\sigma _{i}}}|x_{i}-y|\)). Finally, we obtain the covering radius \(r_{N}\) by computing the maximum of these distances.
Next, we consider the Voronoi deviation \(d_{N}\). Let \(\zeta \in {\mathbb {R}}^{3N}\) be
Using parameters \(q, s_i, a_{ij}\in {\mathbb {R}}^{+}\, (i,j=1,2,\ldots ,N)\), we set \(z\in {\mathbb {R}}^{N^2+N+1}\) as
Moreover, we set \(M\in {\mathbb {R}}^{3N\times (N^2+N+1)}\) so that equation \(Mz=\zeta \) represents
and
Then, by considering \(a_{ij}\) to be \(\left| \sigma _{i}\cap \xi _{j}\right| \), we find that the minimum value of q with condition \(Mz=\zeta \) coincides with the Voronoi deviation \(d_{N}\). We therefore consider the linear problem:
Here, \(b:=(0,0,\ldots ,0,1)^T\in {\mathbb {R}}^{N^2+N+1}\). The solution \(b^Tz\) of (63) is equivalent to the Voronoi deviation \(d_{N}\). Because \(Mz=\zeta \) is unique for \(({\mathcal {X}}_{N}, {{\mathcal {V}}}_{N}, h_{N})\), the linear problem is computable via numerical methods for linear programming problems, such as the simplex method [7].
Appendix 3: Construction of reference weight functions
For all \(n\in {\mathbb {N}}\, (n\ge 2)\), it is possible to construct a reference weight function satisfying Hypothesis 1 with \(n\) as the condition of Hypothesis 1 can be rewritten to include a finite number of conditions
Here, the Gauss symbol \(\lfloor a \rfloor \) denotes the largest integer that is less than or equal to a. For example, function \(w\) is set as the pth polynomial function:
Then, if coefficients \(a_\ell \) satisfy the linear equations
then \(w\) satisfies \(w\in {\mathcal {W}}\) and Hypothesis 1 with \(n\). Therefore, to construct reference functions with Hypothesis 1 with \(n\) represented by polynomial functions, the degree of the polynomial functions must be at least \(\lfloor n/2 \rfloor +2\).
Moreover, for all \(k\in {\mathbb {N}}_0\), reference weight functions satisfying Hypothesis 2 with \(k\) can be constructed based on the following proposition.
Proposition 1
Assume that reference the weight function\(w\)defined in\({\mathbb {R}}^{+}_0\)satisfies\(w\in C^1({\mathbb {R}}^{+}_0)\)and is represented by a polynomial function in [0, s] for \(s\in (0,1]\). Let \(p_{0}\)be the minimum degree of\(w\) in [0, s]. Then, if\(p_{0}-k\ge 1\), \(w\)satisfies Hypothesis 2with\(k\).
Proof
From the assumption, \(w\) can be represented by
where \(p\in {\mathbb {N}}\) and \(a_\ell \in {\mathbb {R}}\, (\ell =p_0, p_0+1, \ldots , p)\). Set \(w^{(k)}\) as (8). Since
and
if \(p_0-k\ge 1\), we have \(w\) satisfies Hypothesis 2 with \(k.\)
This proposition means that the regularity of the reference functions around zero is important.
About this article
Cite this article
Imoto, Y. Truncation error estimates of approximate operators in a generalized particle method. Japan J. Indust. Appl. Math. 37, 565–598 (2020). https://doi.org/10.1007/s13160-020-00407-1
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s13160-020-00407-1
Keywords
- Generalized particle method
- Truncation error estimate
- Approximate operator
- Smoothed particle hydrodynamics method
- Moving particle semi-implicit method