National Taiwan University

Size: px
Start display at page:

Download "National Taiwan University"

Transcription

1 National Taiwan University Meshless Methods for Scientific Computing (Advisor: C.S. Chen, D.L. oung) Final Project Department: Mechanical Engineering Student: Kai-Nung Cheng SID: D9956 Date: Jan. 8

2 The Time-Marching MFS and Time-Marching LMAPS for Solving Klein-Gordon Equations

3 . Introduction In this report, we consider a partial differential equation (PDE) problem called the Klein-Gordon equation, and which is a relativistic version of the Schrodinger equation. It was first found by Schrodinger. The Klein-Gordon equation is the motion equation of a quantum scalar or pseudo scalar field, a field whose quanta are spinless particle. It is second order in time and does not admit a positive definite conserved probability density. With the appropriate interpretation, it does describe the quantum amplitude for finding a point particle in various places, the relativistic wave function, but the particle propagates both forwards and backwards in time []. For solving the Klein-Gordon equation, we first introduce a meshless numerical method called the time-marching method of fundamental solutions (TM-MFS) to obtain the approximate solution of Klein-Gordon equation [, 3]. It combines the Houbolt finite difference (FD) scheme with the method of the particular solutions (MPS) and the method of fundamental solutions (MFS). Implementing the Houbolt FD with the MFS+MPS is easy to eliminate the difficulties of dealing with time variable and the initial conditions in the PDE to form the linear system of equations. In the other hand, we also consider a large time domain problem described by the Klein-Gordon equation. In order to efficiently solve the large problem, we introduce a localized meshless method combined with the Houbolt FD to obtain the approximate solution of Klein-Gordon equation [4], and we call this method the time-marching localized method of approximate particular solutions (TM-LMAPS). In this report, we consider two kinds of -D Klein-Gordon equations which are the linear homogenous and the nonhomogeneous problems, and both of them are solved by using the TM-MFS and TM-LMAPS, respectively.. Definitions of Klein-Gordon equation The -D linear homogenous and the -D nonhomogeneous Klein-Gordon equations are described as below [5]: () The general form of -D linear homogenous Klein-Gordon equation can be written as: u t = a u bu (.) and one of its exact solutions is expressed as follows: u(x, y, t) = e ±μt (Acosλx + Bsinλy), b = a λ μ (.) where a, b, A and B are arbitrary constants. () The general form of -D linear nonhomogeneous Klein-Gordon equation with a source term is expressed as follows:

4 u t = a u bu + v(x, y, t) (.3) and its exact solution can be expressed as follows: u(x, y, t) = l b sin(t b) + sin (t a λ n + b) l cos(λ nx)cos(λ n y) n= a λ n + b (.4) where b > and λ n = πn l. Based on Eq. (.) and (.3), we assume two Klein-Gordon problems, the homogenous and the nonhomogeneous cases, corresponding to the definitions of Klein-Gordon equation above, and then solve them to obtain the approximate solutions by using the global method (TM-MFS) and local method (TM-LMAPS), respectively. Therefore, four examples have been constructed with giving the different requirements which are described as below: Example : the linear homogenous problem solved by the TM-MFS From Eq. (.), we choose a = and b =, and then the linear homogenous Klein-Gordon equation can be written as follows: u t = u + u, (x, y) Ω (.5) The initial conditions are: u(x, y, t) t= = cosx + 3siny (.6) u t (x, y, t) t= = (cosx + 3siny) (.7) and the boundary condition is given by u = e t (cosx + 3siny), (x, y) Ω (.8) The boundary profile is the star shape domain with the Dirichlet boundary condition only and shown in Figure -. Furthermore, the exact solution of Eq. (.5) can be obtained from Eq. (.), and it is expressed as follows: u = e t (cosx + 3siny) (.9) 3

5 Figure - Star shape domain with the Dirichlet boundary condition Example : the linear nonhomogeneous problem solved by the TM-MFS From Eq. (.3), we choose a =, b =, l = and n =, and then the linear nonhomogeneous Klein-Gordon equation can be written as follows: u t = u u + π sint cos πx cos πy, (x, y) Ω (.) The initial conditions are: u(x, y, t) t= = (.) u t (x, y, t) t= = + cos πx cos πy (.) and the boundary conditions are given by u = sint ( + cos πx cos πy ) (x, y) ΩD (.3) u πx πy = *sint ( + cos cos n )+ n (x, y) Ω N (.4) The boundary profile is the Cassini domain with the mixed boundary condition and shown in Figure -. In the figure, Ω D is Dirichlet boundary above the x-axis and Ω N is Neumann boundary below the x-axis. Furthermore, the exact solution of Eq. (.) can be obtained from Eq. (.4), and it is expressed as follows: u = sint ( + cos πx cos πy ) (.5) 4

6 .5.5 Ω D Ω N Figure - Cassini domain with the mixed boundary condition Example 3: the linear homogenous problem solved by the TM-LMAPS We consider a large problem which can be described by Eq. (.5), and its two initial conditions are the same as Eq. (.6) and (.7). We give this problem the mixed boundary condition expressed as follows: u = e t (cosx + 3siny) (x, y) Ω D (.6) u n = [e t (cosx + 3siny)] n (x, y) Ω N (.7) The boundary profile is the Amoeba shape domain and shown in Figure -3. In the figure, Ω D is Dirichlet boundary above the x-axis and Ω N is Neumann boundary below the x-axis. Example 4: the linear nonhomogeneous problem solved by the TM-LMAPS Similarly, we use Eq. (.) to describe the large Klein-Gordon problem, and its two initial conditions and the mixed boundary condition are the same as the description of Example. Its boundary profile we choose is also the Amoeba shape domain shown in Figure -3. Ω D - Ω N - 3 Figure -3 Amoeba shape domain with the mixed boundary condition 5

7 3. Numerical Algorithm In order to simplify time domain problem, both of meshless methods involving the MFS+MPS and LMAPS, which combine with the Houbolt FD method, respectively, are introduced in this section. The Houbolt FD method is used to discretize the time domain and to eliminate the time variable in the PDE. The TM-MFS is used for solving general problems and TM-LMAPS is used for solving large problems. The numerical algorithms of them are described in the following sub-sections. 3. Time-marching method of fundamental solutions (TM-MFS) For example, a time domain Poisson equation and its boundary condition can be expressed as follows: u c t = u x Ω (3.) B(u) = b(x, t) x Ω (3.) where B is the boundary operator. In order to simplify the time domain problem, the Houbolt FD method is used to discretize the time domain in the PDE. It is an implicit and unconditionally stable time integration method to obtain the first and second order differential terms in time within a small time interval from (n ) t to (n + ) t, and both of terms can be expressed as follows: u n+ t 6 t (un+ 8u n + 9u n u n ) (3.3) u n+ t t (un+ 5u n + 4u n u n ) (3.4) where t is the time interval and u n is u in the n-th time level. Therefore, Eq. (3.) can be re-written by Eq. (3.4) as follow: u n+ = c t (un+ 5u n + 4u n u n ) (3.5) Based on the MFS+MFS, the solution of Eq. (3.5) can be expressed as follows: u n+ = u H n+ + u P n+ (3.6) where u H and u P are the homogenous solution and particular solution, respectively. The homogenous solution can be obtained by the linear combination of fundamental solutions as follows: 6

8 n b u n+ H = α n+ j G x ξ j j= (3.7) where n b is the number of the boundary points, x are the boundary points, ξ are the source points and α is the coefficient of the fundamental solution. In -D problem, G x ξ j can be obtained by: G x ξ j = π ln x ξ j (3.8) On the other hand, the particular solution can be approximated by the radial basis function as follows: n i u P n+ = β j n+ F(r j ) j= (3.9) where n i is the number of the collocation points, F is the radial basis function and β is the coefficient of the basis function. In this report, we choose the compactly supported RBF (CS-RBF) as the basis function (f(r)), and F(r) is the particular solution of f(r) [6]. Both f(r) and F(r) can be expressed as follows: ΔF(r) = f(r) (3.) f(r) = { ( r λ ), r λ, r λ (3.) r 4 6λ F(r) = r3 9λ + r, r λ 4 3λ (3.) { 44 + λ ln (r λ ), r λ Then we can re-write Eq. (3.) and (3.) by the equations mentioned above as the following forms: n i β j n+ [f(r j ) CF(r j )] j= n b α n+ j CG x ξ j j= = C( 5un + 4u n u n ) (3.3) n i β j n+ B[F(r j )] j= n b α n+ j B[G x ξ ] j j= = b(x, t n+ ) (3.4) where C = c Δt and the system of equations is easy to be written as follows: 7

9 [ A A ] [ βn+ A 3 A 4 α n+] = *S b + (3.5) where all sub-matrices of Eq. (3.5) are: A = f(r) CF(r), A = CG x ξ, j A 3 = B[F(r)], A 4 = B[G x ξ ] j S = C( 5un + 4u n u n ) (3.6) In Eq. (3.6), u n and u n can be obtained by using the Euler explicit scheme and they are expressed as follows: u n = I I (x ) Δt I II (x ) u n = I I (x ) Δt I II (x ), n (3.7) After obtaining the coefficients, α n+ and β n+, by solving Eq. (3.5) and (3.6), the approximate solution of Eq. (3.) and (3.) can be calculated by: n i u n+ = β j n+ F(r j ) j= n b + α n+ j G x ξ j j= (3.8) 3. Time-marching localized method of approximate particular solutions (TM-LMAPS) For large problems, they can be solved by implementing the localized method of approximate particular solutions (LMAPS). Using the LMAPS to solve the PDE, we can use any collocation point as the center point to construct the local domain which only involves a few n s nearest neighbor points (i.e. Ω i = {x i k }, k ns) to efficiently calculate the approximate solution u (x i ) with low computational interpolation. By the collocation method, we can assume that the approximate solutions u (x n i ) of n collocation points in the local domains can be expressed as the following linear system: u (x i ) Φ x i x i u (x i ) Φ x i x i : : [ u (x i n )] [ Φ x i n x i Φ x i x i Φ x i x i : Φ x n i x i.... :.. Φ x i x n i Φ x i x n i : Φ x n i x n i ] = [ α α : ] (3.9) α n where α is the unknown coefficient, and we can re-write Eq. (3.9) as follows: 8

10 α = Φ u n (3.) From Eq. (3.9) and (3.), we can re-formulate u (x i ) in each local domain as follows: u (x i ) = Φ (x i )α = Φ (x i )Φ u n = Ψ n (x i )u n (3.) where Ψ n = Φ (x i )Φ = [φ, φ,, φ n ] (3.) From Eq. (3.), we can use the global u N instead of local u n by padding vector Ψ n (x) with zero entries based on the mapping between u n and u N. It follows that u (x i ) = Ψ N (x i )u N (3.3) where Ψ N (x i ) is a sparse vector with N components which is obtained by inserting N n zero into Ψ n (x i ) at the proper places. For solving the time domain problem, i.e. Eq. (3.) and (3.), by using the local domains mentioned above, the domain equation, Eq. (3.), can be re-defined with the Houbolt FD method, i.e. Eq. (3.5), as follows: n f(x i ) = α k ΔΦ n(x i ) k= = c t (un+ 5u n + 4u n u n ) (3.4) and the boundary equations, i.e. Eq. (3.), can be re-defined as follows: n g(x i ) = α k BΦ n(x i ) = b(x i, t n+ ) (3.5) k= where α is defined by Eq. (3.). Therefore, we can extend u n of Eq. (3.4) and (3.5) to u N by: f(x i ) = Ξ N (x i )u N i N i (3.6) g(x i ) = Υ N (x i )u N N i i N (3.7) Eventually, we can use Eq. (3.6) and (3.7) to easily construct the sparse system of equations as follows: 9

11 Ξ N (x ) : Ξ N (x Ni ) Υ N (x Ni+ ) [ : Υ N (x N ) ] [ u (x ) : u (x Ni ) u (x Ni+ ) : u (x N ) ] = f(x ) : f(x Ni ) g(x Ni+ ) : [ g(x N ) ] (3.8) By solving Eq. (3.8), we can obtain the approximate solution u at all given points. In the TM-LMAPS, the MQ is chosen as the basis function so that we have φ(r) = r + (cr ) (3.9) with normalized shape parameter c, and r is the maximum distance from the center point to all neighbor points in the local domain. The particular solutions Φ(r) of φ(r) is Φ(r) = 9 (4(cr ) + r ) r + (cr ) (cr ) ln (cr 3 + r + (cr ) ) (3.3) Furthermore, Eq. (3.9) and (3.3) are used to substitute ΔΦ n(x i ) and Φ n(x i ) in Eq. (3.4) and (3.5). 3.3 Error estimation In this report, three kinds of error estimation are used to measure the accuracy of the approximate solution which is obtained from either the TM-MFS or TM-LMAPS. They are the, AME and ARE, and are described as follows: The (Root Mean Square Error) is defined by: N = N u k u k k= (3.3) the AME (Absolute Maximum Error) is defined by: AME = max( u k u k ) k N (3.3) and the ARE (Absolute Relative Error) is defined by: ARE = max ( u k u k u k ) k N (3.33)

12 where u k is the approximate solution, u k is the exact solution and N is the number of the test points. To measure the accuracy of approximate solution in each example described in Section, the AME is used for the linear homogenous Klein-Gordon problem and the ARE is used for the linear nonhomogeneous Klein-Gordon problem. Furthermore, the is used for both of problems. 4. Numerical Results In Section, four examples have been constructed to describe the linear homogenous and the linear nonhomogeneous Klein-Gordon problems with the different domain and boundary conditions, respectively. In this section, we use the global and the local meshless methods, the TM-MFS and TM-LMAPS described in Sections 3. and 3., to solve the Klein-Gordon equations and to efficiently obtain the accurate approximate solution from each example. Example : the linear homogenous Klein-Gordon problem solved by the TM-MFS In order to solve Eq. (.5) to (.8) with the star shape domain shown in Figure - by the TM-MFS, we can generate the interpolation points distributing inside and outside the star shape domain like the pattern of Figure 4- (left) and the test points shown in Figure 4- (right) for the error estimation of the approximate solution. In Example, several cases with the different input data have been considered and listed in Table 4-. To implement the MFS+MPS, we choose, 3 and 4 quasi-random interior points inside the domain, respectively, boundary points on the boundary and source points around the circle with R = 3.5, 4. and 4.5, respectively. The total,48 uniformly distributed test points are used to test the error of the approximate solution. Furthermore, the time interval including E-, 5E-3 and E-3 are used to discretize the time domain in the TM-MFS, respectively. As shown in Figure 4-, the results of the versus time in each case have been obtained by the TM-MFS. All results show that the keeps linearly descending when the time increases and the accurate approximate solution in each case can be obtained by using a small time interval (i.e. t = E-3). According to the wave motion during the time history shown in Figure 4-3, the initial transient wave will tend to be motionless after t = 5. The exact solution of the linear homogenous Klein-Gordon equation, i.e. Eq. (.9), completely represents the trend of the wave motion shown in Figure 4-3. Due to the wave motion to be motionless in the end, it can be found that the will approach to zero if the time is infinite. Furthermore, we also estimate the results of the AME and in the specific time using 4 interior points, boundary points, source points with R = 4. and t = E-3, and they are shown in Figure 4-4. For example, we obtain that the AME is 4.6 E- and is.69 E- at t =., and AME is 3.67 E-4 and is 8.84 E-5 at t = 5..

13 Table 4- Points and time interval for the TM-MFS in Example Case MPS MFS Interior points Boundary points Source points Source location a R = 4.5 b 3 R = 3.5 c 4 R = 4. Houbolt FD t = E-, 5E-3 and E-3 Test points,4 uniformly distributed points and 6 boundary points Figure 4- Point distribution (blue: interior points, green: boundary points and red: source points) for the TM-MFS in Example (left) and test points (right)

14 - t = E- t = 5E-3 t = E-3 - t = E- t = 5E-3 t = E t (a) interior points, boundary points and source points with R = t (b) 3 interior points, boundary points and source points with R = 3.5 t = E- t = 5E-3 t = E I.Pts + B.Pts + S.Pts 3 I.Pts + B.Pts + S.Pts 4 I.Pts + B.Pts + S.Pts t (c) 4 interior points, boundary points and source points with R = t (d) Comparison results of three cases with t = E-3 Figure 4- Results of vs. time in Example 3

15 U U (a) t =. (b) t = U U (c) t =. (d) t = 5. Figure 4-3 Results of wave motion vs. time in Example using 4 interior points, boundary points, source points with R = 4. and t = E-3 4

16 4.6e- 4.6e-.6e-.85e- 3.7e-.65e- 3.3e-.44e-.77e-.4e-.3e-.3e-.85e- 8.3e e e-3 9.4e-3 4.e e-3.93e e-3.88e-7 (a) t =., AME: 4.6 E- :.69 E- (b) t =., AME:.583 E- : E-3 7.4e e-3 3.6e-4 3.4e e-3.88e e-3.5e-4 4.e-3.6e-4 3.5e-3.8e-4.8e-3.44e-4 -.e-3 -.8e-4.4e-3 7.e e-4 3.e e-5.47e-8 (c) t =., AME: E-3 :.6348 E-3 (d) t = 5., AME: 3.67 E-4 : 8.84 E-5 Figure 4-4 Results of absolute error distribution vs. time in Example using 4 interior points, boundary points, source points with R = 4. and t = E-3 5

17 Example : the linear nonhomogeneous Klein-Gordon problem solved by the TM-MFS To solve Eq. (.) to (.4) with the Cassini domain and mixed boundary shown in Figure -, we can follow the way described in Example to generate the interpolation points distributing inside and outside the Cassini domain like the pattern of Figure 4-5 (left) and the test points shown in Figure 4-5 (right) for the error estimation of the approximate solution. In Example, we consider several cases with the different input data and list them in Table 4-. To implement the MFS+MPS, we choose, 3 and 4 quasi-random interior points inside the domain, respectively, Dirichlet boundary points and Neumann boundary points on the mixed boundary and source points around the circle with R = 3. and 3.5, respectively. The total,44 uniformly distributed test points are used to test the error of the approximate solution. Furthermore, the time interval including E-, 5E-3 and E-3 are used to discretize the time domain in the TM-MFS, respectively. As shown in Figure 4-6, the results of the versus time in each case have been obtained by the TM-MFS. From these results, it can be found that all s are the periodic values within E- to E-3 order error during the time history, and the s never keep ascending or descending when the time increases. In addition, the accurate approximate solution can also be obtained by using a small time interval (i.e. t = E-3). After observing the wave motion during the time history in Figure 4-7, we find that the wave repeatedly vibrates during the time history and it never decays. Therefore, the wave is a periodic and non-decaying motion and it can be completely represented by the exact solution of the linear nonhomogeneous Klein-Gordon equation, i.e. Eq. (.5). Due to the periodic wave motion, the trend of the results during the time history is the same as the wave motion results in Figure 4-7. Furthermore, we also estimate the results of the ARE and in the specific time using 4 interior points, boundary points, source points with R = 3.5 and t = E-3, and they are shown in Figure 4-8. For example, we obtain that the ARE is.4 E- and is.6733 E-3 at t =., and ARE is.649 E- and is.458 E-3 at t = 5. Table 4- Points and time interval for the TM-MFS in Example Case MPS MFS Interior points Boundary points Source points Source location a (D) + (N) R = 3. b 3 (D) + (N) R = 3. c 4 (D) + (N) R = 3.5 Houbolt FD t = E-, 5E-3 and E-3 Test points,84 uniformly distributed points and 6 boundary points 6

18 Figure 4-5 Point distribution (blue: interior points, green: Dirichlet boundary points, magenta: Neumann boundary points and red: source points) for the TM-MFS in Example (left) and test points (right) 7

19 t = E- t = 5E-3 t = E-3 t = E- t = 5E-3 t = E t t (a) interior points, boundary points and source points with R = 3. (b) 3 interior points, boundary points and source points with R = 3. t = E- t = 5E-3 t = E-3 - I.Pts + B.Pts + S.Pts 3 I.Pts + B.Pts + S.Pts 4 I.Pts + B.Pts + S.Pts t (c) 4 interior points, boundary points and source points with R = t (d) Comparison results of three cases with t = E-3 Figure 4-6 Results of vs. time in Example 8

20 U U (a) t =. (b) t = U - U (c) t = (d) t = 5 Figure 4-7 Results of wave motion vs. time in Example using 4 interior points, boundary points, source points with R = 3.5 and t = E-3 9

21 .5.e e-3 9.9e e-3 8.8e-3.96e e-3 6.6e e-3.e-3 5.5e-3.85e-3 4.4e-3.48e e e-3 -.e-3.e-3-7.4e-4 3.7e e e-6 (a) t =., max. ARE:.4 E- :.6733 E-3 (b) t = 5., max. ARE: E-3 : E-4.5.6e-.5.64e-.3e-.8e-.48e-.3e-.5.58e-.35e-.5.5e- 9.86e-3.3e- 8.e e e e e-3-4.5e-3.6e-3-3.9e-3.65e e e-6 (c) t =, max. ARE:.57 E- :.49 E-3 (d) t = 5, max. ARE:.649 E- :.458 E-3 Figure 4-8 Results of ARE distribution vs. time in Example using 4 interior points, boundary points, source points with R = 3.5 and t = E-3

22 Example 3: the linear homogenous Klein-Gordon problem solved by the TM-LMAPS We consider a large time domain problem which can be described by Eq. (.5) to (.7) and satisfies the boundary profile of the Amoeba shape domain with the mixed boundary condition, i.e. Eq. (.6) and (.7). To solve this problem by the TM-LMAPS, we can generate a large number of interpolation points distributing inside the Amoeba shape domain and on its boundary like the pattern of Figure 4-9 and use the same interpolation points as the test points for error estimation of the approximate solution. For the LMAPS, it is important to determine the number of the nearest neighbor points (n s ) around each center point for the interpolation in the local domain. Therefore, we have constructed several cases with the different input data and listed them in Table 4-3. In this example, we choose 3,496 uniformly distributed interior points inside the domain, 46 Dirichlet boundary points and 54 Neumann boundary points on the mixed boundary. Furthermore, the local nearest neighbor points, n s = 5,,, and the shape parameter of the MQ basis function, c = 5,, 3 are used for each cases, respectively. The total 3,896 interpolation points are also used to test the error of the approximate solution. The time interval including E-, 5E-3 and E-3 are used to discretize the time domain in the TM-LMAPS, respectively. As shown in Figure 4-, the results of the versus time in each case have been obtained by the TM-LMAPS. These results show that the keeps linearly descending when the time increases and the accurate approximate solution in each case can be obtained by using a small time interval (i.e. t = E-3). Checking the results in Figure 4- (d), it is clear to know that the acceptable approximate solution can be obtained when we only choose n s = 5 and t = E-3 since the s of three cases are very close. According to the wave motion results in Figure 4-, we can find that these results are very similar to the results in Figure 4-3. The initial transient wave will tend to be motionless after t = 5. Eq. (.9) completely represents this trend of the wave motion and the will approach to zero if the time is infinite. Furthermore, we also estimate the results of the AME and in the specific time using n s = 5 and t = E-3, and they are shown in Figure 4-. From this figure, we find that the worse AMEs almost occur on the Neumann boundary and no error occurs on the Dirichlet boundary. For example, we obtain that the AME is E- and is.4368 E-3 at t =., and AME is.88 E-4 and is.55 E-5 E-5 at t = 5.. Table 4-3 Points, shape parameter and time interval for the TM-LMAPS in Example 3 Case LMAPS Interior points Boundary points n s c a 3, (Dirichlet) + 54 (Neumann) 5 3 b 3, (Dirichlet) + 54 (Neumann) c 3, (Dirichlet) + 54 (Neumann) 5 Houbolt FD t = E-, 5E-3 and E-3 Test points 3,496 uniformly distributed points and 4 boundary points

23 Figure 4-9 Point distribution (blue: interior points, green: Dirichlet boundary points and magenta: Neumann boundary points) for the TM-LMAPS in Example 3

24 - t = E- t = 5E-3 t = E-3 - t = E- t = 5E-3 t = E t t (a) 3,496 interior points, 4 boundary points, ns = 5 and c = 3 (b) 3,496 interior points, 4 boundary points, ns = and c = - t = E- t = 5E-3 t = E ns = 5, c = 3 ns =, c = ns =, c = t t (c) 3,496 interior points, 4 boundary points, ns = and c = 5 (d) Comparison results of three cases with t = E-3 Figure 4- Results of vs. time in Example 3 3

25 U U (a) t =. (b) t = U U (c) t =. (d) t = 5. Figure 4- Results of wave motion vs. time in Example 3 using 3,496 interior points, 4 boundary points, ns = 5, c = 3 and t = E-3 4

26 (a) t =., AME: E- :.4368 E-3 (b) t =., AME:.547 E- :.949 E-3 (c) t =., AME: E-3 : 4.8 E-4 (d) t = 5., AME:.88 E-4 :.55 E-5 Figure 4- Results of absolute error distribution vs. time in Example 3 using 3,496 interior points, 4 boundary points, ns = 5, c = 3 and t = E-3 5

27 Example 4: the linear nonhomogeneous Klein-Gordon problem solved by the TM-LMAPS In this example, a large time domain problem is considered and described by Eq. (.) to (.4), and it satisfies the boundary profile of the Amoeba shape domain with the mixed boundary condition. For solving this problem by the TM-LMAPS, we use the interpolation points and test points which are the same as the point distribution of Example 3 (shown in Figure 4-9). Similarly, we also calculate the approximate solution and test the error from several cases listed in Table 4-3. Therefore, the linear nonhomogeneous Klein-Gordon problem, i.e. Eq. (.), is solved by using 3,496 uniformly distributed interior points, 4 boundary points, n s = 5,,, and c = 5,, 3, respectively. The time interval including E-, 5E-3 and E-3 are used to discretize the time domain in the TM-LMAPS, respectively. As shown in Figure 4-3, the results of the versus time in each case have been obtained by the TM-LMAPS. From these results, it can be found that all s are the periodic values within E- to E-4 order error during the time history, and the s never keep ascending or descending when time increases. In addition, the accurate approximate solution can also be obtained by using a small time interval (i.e. t = E-3). Checking the results in Figure 4-3 (d), we can find that the obtained by using n s = 5 and t = E-3 is less unstable than that obtained by using other parameters. Since using a few nearest neighbor points ( n s = 5) can provide a stable and acceptable approximate solution, we should try to choose the nearest neighbor points as few as possible for the interpolation in local domain when the TM-LMAPS is implemented for solving the time domain problem. From Figure 4-4, the wave motion is a periodic vibration during the time history and it never decays. This motion is similar to the result of Figure 4-7 in Example. Furthermore, we also estimate the results of the ARE and in the specific time using n s = 5 and t = E-3, and they are shown in Figure 4-5. This figure shows that the worse AMEs almost occur on the Neumann boundary and no error occurs on the Dirichlet boundary. For example, we obtain that the ARE is.96 E- and is E-4 at t =., and ARE is.87 E- and is E-4 at t = 5. 6

28 - t = E- t = 5E-3 t = E-3 - t = E- t = 5E-3 t = E t t (a) 3,496 interior points, 4 boundary points, ns = 5 and c = 3 (b) 3,496 interior points, 4 boundary points, ns = and c = - - t = E- t = 5E-3 t = E-3 - ns = 5, c = 3 ns =, c = ns =, c = t t (c) 3,496 interior points, 4 boundary points, ns = and c = 5 (d) Comparison results of three cases with t = E-3 Figure 4-3 Results of vs. time in Example 4 7

29 U U (a) t =. (b) t = 5. U - U (c) t = (d) t = 5 Figure 4-4 Results of wave motion vs. time in Example 4 using 3,496 interior points, 4 boundary points, ns = 5, c = 3 and t = E-3 8

30 (a) t =., max. ARE:.96 E- : E-4 (b) t = 5., max. ARE:.887 E- : E-4 (c) t =, max. ARE:.335 E- : E-4 (d) t = 5, max. ARE:.87 E- : E-4 Figure 4-5 Results of ARE distribution vs. time in Example 4 using 3,496 interior points, 4 boundary points, ns = 5, c = 3 and t = E-3 9

31 5. Discussions In Section 4, we have successfully solved four Klein-Gordon examples by using both global method (TM-MFS) and local method (TM-LMAPS). As shown in Table 5-, the results of the in the specific conditions obtained from each example have been listed in this table. According to these results, we obtain: () For obtaining the more accurate approximate solutions of Klein-Gordon equations, a small time interval (i.e. t = E-3) should be used for the Houbolt FD in both of methods. () Both of methods can obtain the quite accurate approximate solutions from each Klein-Gordon example, and all s of them are around E-3 to E-4 order error. (3) Comparing the results of all Klein-Gordon examples, it is found that the s obtained from the TM-LMPAS are superior to that obtained from the TM-MFS. This reason is that the dynamic results (e.g. wave motion) can be much clearly represented by using a large number of interpolation points. Since there is no ill-condition problem in the TM-LMAPS, the more accurate approximate solution can be obtained from the TM-LMAPS. (4) From the results of the linear nonhomogeneous Klein-Gordon problems (i.e. Examples and 4), the TM-MFS obtains around E-3 order and the TM-LMAPS obtains around E-4 order. (5) Figure 5- shows the results of the versus shaper parameter c of the MQ in Example 4. From these results, it is clear to know that the good solutions of Klein-Gordon problem can be obtained when we choose n s = 5 and c = 3 for the TM-LMAPS in Example 4. Furthermore, this figure also shows that n s = is still a good choices for the TM-LMAPS when c is less than 9. (6) In Example 4, the stable results of the versus time can be obtained by using a few nearest neighbor points (e.g. n s = 5) in the local domains. Table 5- Comparison results of in all Klein-Gordon examples Method Ex. Klein-Gordon problem type Domain Boundary Interpolation points Time..69 E- TM-MFS Linear homogenous Star Dirichlet 4~ E E E-5 (Global) Linear nonhomogen eous Cassini Dirichlet + Neumann 4~ E E-4.49 E E-3 TM-LMAPS 3 Linear homogenous Amoeba Dirichlet + Neumann 3, E E E E-5 (Local) 4 Linear nonhomogen eous Amoeba Dirichlet + Neumann 3, E E E E-4 3

32 ns = 5 ns = ns = ns = 5 ns = ns = c (a) 3,496 interior points, 4 boundary points, t = c (b) 3,496 interior points, 4 boundary points, t = 5. - ns = 5 ns = ns = - ns = 5 ns = ns = c (c) 3,496 interior points, 4 boundary points, t = c (d) 3,496 interior points, 4 boundary points, t = 5 Figure 5- Results of vs. shaper parameter c in Example 4 3

33 6. Conclusions In this report, we introduce both of meshless methods, the TM-MFS and TM-MAPS, to solve the Klein-Gordon equations. For numerical method verification, we have constructed four examples including the D linear homogenous and the -D nonhomogeneous Klein-Gordon problems and then solved them by using both TM-MFS and TM-MAPS, respectively. As a result, we are confident to show that our two proposed methods are quite useful for solving the Klein-Gordon equations. According to the results in Sections 4 and 5, some conclusions have been summarized as follows: () For solving the Klein-Gordon equations, we do not need to be an expert of the Klein-Gordon problems since we can easily solve them by our proposed methods. () Both TM-MFS and TM-LMAPS are useful for solving the time domain PDE problem, and which can obtain the quite accurate approximate solutions on Klein-Gordon problems. (3) All s of the approximate solutions obtained from the TM-MFS and TM-LMAPS are around E-3 to E-4 order error on Klein-Gordon problems. (4) For solving large and dynamic problems, we strongly recommend that the TM-LMAPS should be chosen as the numerical solver since it can provide the better results and save a lot of time on numerical interpolation. (5) To implement the TM-LMAPS, we recommend that the number of the nearest neighbor points, n s = 5, is a good choice to obtain the stable and good s on periodic problems and n s = is also good for the accurate results. 7. Reference [] Klein-Gordon equation, from Wikipedia, the free encyclopedia. [] D.L. oung, M.H. Gu, C.M. Fan, The time-marching method of fundamental solutions for wave equations, Engineering Analysis with Boundary Elements, p.4-45, 9. [3] M.H. Gu, C.M. Fan, D.L. oung, The method of fundamental solutions for the multi-dimensional wave equations, Journal of Marine Science and Technology, Vol. 9, No. 6, pp (). [4] C.S. Chen, Lecture Note, A localized MAPS. [5] The world of mathematical equation, Eqworld. [6] C.S. Chen, C.A. Brebbia, and H. Power, Dual reciprocity method using compactly supported radial basis functions, Commun. Numer. Meth. Engng, 5, 37-5 (999). 3

1 Comparison of Kansa s method versus Method of Fundamental Solution (MFS) and Dual Reciprocity Method of Fundamental Solution (MFS- DRM)

1 Comparison of Kansa s method versus Method of Fundamental Solution (MFS) and Dual Reciprocity Method of Fundamental Solution (MFS- DRM) 1 Comparison of Kansa s method versus Method of Fundamental Solution (MFS) and Dual Reciprocity Method of Fundamental Solution (MFS- DRM) 1.1 Introduction In this work, performances of two most widely

More information

Finite Difference Methods for Boundary Value Problems

Finite Difference Methods for Boundary Value Problems Finite Difference Methods for Boundary Value Problems October 2, 2013 () Finite Differences October 2, 2013 1 / 52 Goals Learn steps to approximate BVPs using the Finite Difference Method Start with two-point

More information

RBF-FD Approximation to Solve Poisson Equation in 3D

RBF-FD Approximation to Solve Poisson Equation in 3D RBF-FD Approximation to Solve Poisson Equation in 3D Jagadeeswaran.R March 14, 2014 1 / 28 Overview Problem Setup Generalized finite difference method. Uses numerical differentiations generated by Gaussian

More information

The Method of Fundamental Solutions with Eigenfunction Expansion Method for Nonhomogeneous Diffusion Equation

The Method of Fundamental Solutions with Eigenfunction Expansion Method for Nonhomogeneous Diffusion Equation The Method of Fundamental Solutions with Eigenfunction Expansion Method for Nonhomogeneous Diffusion Equation D. L. Young, 1 C. W. Chen, 1 C. M. Fan, 1 C. C. Tsai 2 1 Department of Civil Engineering and

More information

In this chapter we study elliptical PDEs. That is, PDEs of the form. 2 u = lots,

In this chapter we study elliptical PDEs. That is, PDEs of the form. 2 u = lots, Chapter 8 Elliptic PDEs In this chapter we study elliptical PDEs. That is, PDEs of the form 2 u = lots, where lots means lower-order terms (u x, u y,..., u, f). Here are some ways to think about the physical

More information

Solving Boundary Value Problems (with Gaussians)

Solving Boundary Value Problems (with Gaussians) What is a boundary value problem? Solving Boundary Value Problems (with Gaussians) Definition A differential equation with constraints on the boundary Michael McCourt Division Argonne National Laboratory

More information

Introduction to numerical schemes

Introduction to numerical schemes 236861 Numerical Geometry of Images Tutorial 2 Introduction to numerical schemes Heat equation The simple parabolic PDE with the initial values u t = K 2 u 2 x u(0, x) = u 0 (x) and some boundary conditions

More information

Chapter 3 Second Order Linear Equations

Chapter 3 Second Order Linear Equations Partial Differential Equations (Math 3303) A Ë@ Õæ Aë áöß @. X. @ 2015-2014 ú GA JË@ É Ë@ Chapter 3 Second Order Linear Equations Second-order partial differential equations for an known function u(x,

More information

SOLVING INHOMOGENEOUS PROBLEMS BY SINGULAR BOUNDARY METHOD

SOLVING INHOMOGENEOUS PROBLEMS BY SINGULAR BOUNDARY METHOD 8 Journal of Marine Science and Technology Vol. o. pp. 8-4 03) DOI: 0.69/JMST-0-0704- SOLVIG IHOMOGEEOUS PROBLEMS BY SIGULAR BOUDARY METHOD Xing Wei Wen Chen and Zhuo-Jia Fu Key words: singular boundary

More information

2.29 Numerical Fluid Mechanics Spring 2015 Lecture 13

2.29 Numerical Fluid Mechanics Spring 2015 Lecture 13 REVIEW Lecture 12: Spring 2015 Lecture 13 Grid-Refinement and Error estimation Estimation of the order of convergence and of the discretization error Richardson s extrapolation and Iterative improvements

More information

Recent developments in the dual receiprocity method using compactly supported radial basis functions

Recent developments in the dual receiprocity method using compactly supported radial basis functions Recent developments in the dual receiprocity method using compactly supported radial basis functions C.S. Chen, M.A. Golberg and R.A. Schaback 3 Department of Mathematical Sciences University of Nevada,

More information

Numerical solution of nonlinear sine-gordon equation with local RBF-based finite difference collocation method

Numerical solution of nonlinear sine-gordon equation with local RBF-based finite difference collocation method Numerical solution of nonlinear sine-gordon equation with local RBF-based finite difference collocation method Y. Azari Keywords: Local RBF-based finite difference (LRBF-FD), Global RBF collocation, sine-gordon

More information

Positive Definite Kernels: Opportunities and Challenges

Positive Definite Kernels: Opportunities and Challenges Positive Definite Kernels: Opportunities and Challenges Michael McCourt Department of Mathematical and Statistical Sciences University of Colorado, Denver CUNY Mathematics Seminar CUNY Graduate College

More information

MIT (Spring 2014)

MIT (Spring 2014) 18.311 MIT (Spring 014) Rodolfo R. Rosales May 6, 014. Problem Set # 08. Due: Last day of lectures. IMPORTANT: Turn in the regular and the special problems stapled in two SEPARATE packages. Print your

More information

Chapter Two: Numerical Methods for Elliptic PDEs. 1 Finite Difference Methods for Elliptic PDEs

Chapter Two: Numerical Methods for Elliptic PDEs. 1 Finite Difference Methods for Elliptic PDEs Chapter Two: Numerical Methods for Elliptic PDEs Finite Difference Methods for Elliptic PDEs.. Finite difference scheme. We consider a simple example u := subject to Dirichlet boundary conditions ( ) u

More information

Compactly supported radial basis functions for solving certain high order partial differential equations in 3D

Compactly supported radial basis functions for solving certain high order partial differential equations in 3D Compactly supported radial basis functions for solving certain high order partial differential equations in 3D Wen Li a, Ming Li a,, C.S. Chen a,b, Xiaofeng Liu a a College of Mathematics, Taiyuan University

More information

arxiv: v1 [physics.comp-ph] 28 Jan 2008

arxiv: v1 [physics.comp-ph] 28 Jan 2008 A new method for studying the vibration of non-homogeneous membranes arxiv:0801.4369v1 [physics.comp-ph] 28 Jan 2008 Abstract Paolo Amore Facultad de Ciencias, Universidad de Colima, Bernal Díaz del Castillo

More information

The purpose of this lecture is to present a few applications of conformal mappings in problems which arise in physics and engineering.

The purpose of this lecture is to present a few applications of conformal mappings in problems which arise in physics and engineering. Lecture 16 Applications of Conformal Mapping MATH-GA 451.001 Complex Variables The purpose of this lecture is to present a few applications of conformal mappings in problems which arise in physics and

More information

Numerical Solutions to Partial Differential Equations

Numerical Solutions to Partial Differential Equations Numerical Solutions to Partial Differential Equations Zhiping Li LMAM and School of Mathematical Sciences Peking University The Implicit Schemes for the Model Problem The Crank-Nicolson scheme and θ-scheme

More information

Numerical solution of surface PDEs with Radial Basis Functions

Numerical solution of surface PDEs with Radial Basis Functions Numerical solution of surface PDEs with Radial Basis Functions Andriy Sokolov Institut für Angewandte Mathematik (LS3) TU Dortmund andriy.sokolov@math.tu-dortmund.de TU Dortmund June 1, 2017 Outline 1

More information

LECTURE # 0 BASIC NOTATIONS AND CONCEPTS IN THE THEORY OF PARTIAL DIFFERENTIAL EQUATIONS (PDES)

LECTURE # 0 BASIC NOTATIONS AND CONCEPTS IN THE THEORY OF PARTIAL DIFFERENTIAL EQUATIONS (PDES) LECTURE # 0 BASIC NOTATIONS AND CONCEPTS IN THE THEORY OF PARTIAL DIFFERENTIAL EQUATIONS (PDES) RAYTCHO LAZAROV 1 Notations and Basic Functional Spaces Scalar function in R d, d 1 will be denoted by u,

More information

MATH 590: Meshfree Methods

MATH 590: Meshfree Methods MATH 590: Meshfree Methods Chapter 34: Improving the Condition Number of the Interpolation Matrix Greg Fasshauer Department of Applied Mathematics Illinois Institute of Technology Fall 2010 fasshauer@iit.edu

More information

Linear differential equations with constant coefficients Method of undetermined coefficients

Linear differential equations with constant coefficients Method of undetermined coefficients Linear differential equations with constant coefficients Method of undetermined coefficients e u+vi = e u (cos vx + i sin vx), u, v R, i 2 = -1 Quasi-polynomial: Q α+βi,k (x) = e αx [cos βx ( f 0 + f 1

More information

Numerical Analysis of Differential Equations Numerical Solution of Elliptic Boundary Value

Numerical Analysis of Differential Equations Numerical Solution of Elliptic Boundary Value Numerical Analysis of Differential Equations 188 5 Numerical Solution of Elliptic Boundary Value Problems 5 Numerical Solution of Elliptic Boundary Value Problems TU Bergakademie Freiberg, SS 2012 Numerical

More information

FDM for parabolic equations

FDM for parabolic equations FDM for parabolic equations Consider the heat equation where Well-posed problem Existence & Uniqueness Mass & Energy decreasing FDM for parabolic equations CNFD Crank-Nicolson + 2 nd order finite difference

More information

Finite Difference Methods for

Finite Difference Methods for CE 601: Numerical Methods Lecture 33 Finite Difference Methods for PDEs Course Coordinator: Course Coordinator: Dr. Suresh A. Kartha, Associate Professor, Department of Civil Engineering, IIT Guwahati.

More information

Lucio Demeio Dipartimento di Ingegneria Industriale e delle Scienze Matematiche

Lucio Demeio Dipartimento di Ingegneria Industriale e delle Scienze Matematiche Scuola di Dottorato THE WAVE EQUATION Lucio Demeio Dipartimento di Ingegneria Industriale e delle Scienze Matematiche Lucio Demeio - DIISM wave equation 1 / 44 1 The Vibrating String Equation 2 Second

More information

PARTIAL DIFFERENTIAL EQUATIONS. Lecturer: D.M.A. Stuart MT 2007

PARTIAL DIFFERENTIAL EQUATIONS. Lecturer: D.M.A. Stuart MT 2007 PARTIAL DIFFERENTIAL EQUATIONS Lecturer: D.M.A. Stuart MT 2007 In addition to the sets of lecture notes written by previous lecturers ([1, 2]) the books [4, 7] are very good for the PDE topics in the course.

More information

Electrodynamics PHY712. Lecture 3 Electrostatic potentials and fields. Reference: Chap. 1 in J. D. Jackson s textbook.

Electrodynamics PHY712. Lecture 3 Electrostatic potentials and fields. Reference: Chap. 1 in J. D. Jackson s textbook. Electrodynamics PHY712 Lecture 3 Electrostatic potentials and fields Reference: Chap. 1 in J. D. Jackson s textbook. 1. Poisson and Laplace Equations 2. Green s Theorem 3. One-dimensional examples 1 Poisson

More information

Transient heat conduction by different versions of the Method of Fundamental Solutions a comparison study

Transient heat conduction by different versions of the Method of Fundamental Solutions a comparison study Computer Assisted Mechanics and Engineering Sciences, 17: 75 88, 2010. Copyright c 2010 by Institute of Fundamental Technological Research, Polish Academy of Sciences Transient heat conduction by different

More information

Introduction to PDEs and Numerical Methods Lecture 7. Solving linear systems

Introduction to PDEs and Numerical Methods Lecture 7. Solving linear systems Platzhalter für Bild, Bild auf Titelfolie hinter das Logo einsetzen Introduction to PDEs and Numerical Methods Lecture 7. Solving linear systems Dr. Noemi Friedman, 09.2.205. Reminder: Instationary heat

More information

Linear Operators and the General Solution of Elementary Linear Ordinary Differential Equations

Linear Operators and the General Solution of Elementary Linear Ordinary Differential Equations CODEE Journal Volume 9 Article 11 5-12-2012 Linear Operators and the General Solution of Elementary Linear Ordinary Differential Equations Norbert Euler Follow this and additional works at: http://scholarship.claremont.edu/codee

More information

Background. Background. C. T. Kelley NC State University tim C. T. Kelley Background NCSU, Spring / 58

Background. Background. C. T. Kelley NC State University tim C. T. Kelley Background NCSU, Spring / 58 Background C. T. Kelley NC State University tim kelley@ncsu.edu C. T. Kelley Background NCSU, Spring 2012 1 / 58 Notation vectors, matrices, norms l 1 : max col sum... spectral radius scaled integral norms

More information

Strauss PDEs 2e: Section Exercise 2 Page 1 of 6. Solve the completely inhomogeneous diffusion problem on the half-line

Strauss PDEs 2e: Section Exercise 2 Page 1 of 6. Solve the completely inhomogeneous diffusion problem on the half-line Strauss PDEs 2e: Section 3.3 - Exercise 2 Page of 6 Exercise 2 Solve the completely inhomogeneous diffusion problem on the half-line v t kv xx = f(x, t) for < x

More information

Intro to Research Computing with Python: Partial Differential Equations

Intro to Research Computing with Python: Partial Differential Equations Intro to Research Computing with Python: Partial Differential Equations Erik Spence SciNet HPC Consortium 28 November 2013 Erik Spence (SciNet HPC Consortium) PDEs 28 November 2013 1 / 23 Today s class

More information

The Method of Fundamental Solutions applied to the numerical calculation of eigenfrequencies and eigenmodes for 3D simply connected domains

The Method of Fundamental Solutions applied to the numerical calculation of eigenfrequencies and eigenmodes for 3D simply connected domains ECCOMAS Thematic Conference on Meshless Methods 2005 C34.1 The Method of Fundamental Solutions applied to the numerical calculation of eigenfrequencies and eigenmodes for 3D simply connected domains Carlos

More information

Integral Equations in Electromagnetics

Integral Equations in Electromagnetics Integral Equations in Electromagnetics Massachusetts Institute of Technology 6.635 lecture notes Most integral equations do not have a closed form solution. However, they can often be discretized and solved

More information

An Introduction to Partial Differential Equations in the Undergraduate Curriculum

An Introduction to Partial Differential Equations in the Undergraduate Curriculum An Introduction to Partial Differential Equations in the Undergraduate Curriculum John C. Polking LECTURE 12 Heat Transfer in the Ball 12.1. Outline of Lecture The problem The problem for radial temperatures

More information

Consistency Estimates for gfd Methods and Selection of Sets of Influence

Consistency Estimates for gfd Methods and Selection of Sets of Influence Consistency Estimates for gfd Methods and Selection of Sets of Influence Oleg Davydov University of Giessen, Germany Localized Kernel-Based Meshless Methods for PDEs ICERM / Brown University 7 11 August

More information

Weak form of Boundary Value Problems. Simulation Methods in Acoustics

Weak form of Boundary Value Problems. Simulation Methods in Acoustics Weak form of Boundary Value Problems Simulation Methods in Acoustics Note on finite dimensional description of functions Approximation: N f (x) ˆf (x) = q j φ j (x) j=1 Residual function: r(x) = f (x)

More information

FDM for wave equations

FDM for wave equations FDM for wave equations Consider the second order wave equation Some properties Existence & Uniqueness Wave speed finite!!! Dependence region Analytical solution in 1D Finite difference discretization Finite

More information

Lecture 9 Approximations of Laplace s Equation, Finite Element Method. Mathématiques appliquées (MATH0504-1) B. Dewals, C.

Lecture 9 Approximations of Laplace s Equation, Finite Element Method. Mathématiques appliquées (MATH0504-1) B. Dewals, C. Lecture 9 Approximations of Laplace s Equation, Finite Element Method Mathématiques appliquées (MATH54-1) B. Dewals, C. Geuzaine V1.2 23/11/218 1 Learning objectives of this lecture Apply the finite difference

More information

A new meshless method for steady-state heat conduction problems in anisotropic and inhomogeneous media

A new meshless method for steady-state heat conduction problems in anisotropic and inhomogeneous media Archive of Applied Mechanics 74 25 563--579 Springer-Verlag 25 DOI 1.17/s419-5-375-8 A new meshless method for steady-state heat conduction problems in anisotropic and inhomogeneous media H. Wang, Q.-H.

More information

Preliminary Examination, Numerical Analysis, August 2016

Preliminary Examination, Numerical Analysis, August 2016 Preliminary Examination, Numerical Analysis, August 2016 Instructions: This exam is closed books and notes. The time allowed is three hours and you need to work on any three out of questions 1-4 and any

More information

MATH 251 Final Examination December 16, 2014 FORM A. Name: Student Number: Section:

MATH 251 Final Examination December 16, 2014 FORM A. Name: Student Number: Section: MATH 2 Final Examination December 6, 204 FORM A Name: Student Number: Section: This exam has 7 questions for a total of 0 points. In order to obtain full credit for partial credit problems, all work must

More information

Recovery of high order accuracy in radial basis function approximation for discontinuous problems

Recovery of high order accuracy in radial basis function approximation for discontinuous problems Recovery of high order accuracy in radial basis function approximation for discontinuous problems Chris L. Bresten, Sigal Gottlieb 1, Daniel Higgs, Jae-Hun Jung* 2 Abstract Radial basis function(rbf) methods

More information

Fast Method of Particular Solutions for Solving Partial Differential Equations

Fast Method of Particular Solutions for Solving Partial Differential Equations The University of Southern Mississippi The Aquila Digital Community Dissertations Fall 12-2016 Fast Method of Particular Solutions for Solving Partial Differential Equations Anup Raja Lamichhane University

More information

Partial Differential Equations

Partial Differential Equations Partial Differential Equations Introduction Deng Li Discretization Methods Chunfang Chen, Danny Thorne, Adam Zornes CS521 Feb.,7, 2006 What do You Stand For? A PDE is a Partial Differential Equation This

More information

MATH 590: Meshfree Methods

MATH 590: Meshfree Methods MATH 590: Meshfree Methods Chapter 40: Symmetric RBF Collocation in MATLAB Greg Fasshauer Department of Applied Mathematics Illinois Institute of Technology Fall 2010 fasshauer@iit.edu MATH 590 Chapter

More information

Mathematics Qualifying Exam Study Material

Mathematics Qualifying Exam Study Material Mathematics Qualifying Exam Study Material The candidate is expected to have a thorough understanding of engineering mathematics topics. These topics are listed below for clarification. Not all instructors

More information

2. Determine whether the following pair of functions are linearly dependent, or linearly independent:

2. Determine whether the following pair of functions are linearly dependent, or linearly independent: Topics to be covered on the exam include: Recognizing, and verifying solutions to homogeneous second-order linear differential equations, and their corresponding Initial Value Problems Recognizing and

More information

INTRODUCTION TO PDEs

INTRODUCTION TO PDEs INTRODUCTION TO PDEs In this course we are interested in the numerical approximation of PDEs using finite difference methods (FDM). We will use some simple prototype boundary value problems (BVP) and initial

More information

Final Exam May 4, 2016

Final Exam May 4, 2016 1 Math 425 / AMCS 525 Dr. DeTurck Final Exam May 4, 2016 You may use your book and notes on this exam. Show your work in the exam book. Work only the problems that correspond to the section that you prepared.

More information

MB4018 Differential equations

MB4018 Differential equations MB4018 Differential equations Part II http://www.staff.ul.ie/natalia/mb4018.html Prof. Natalia Kopteva Spring 2015 MB4018 (Spring 2015) Differential equations Part II 0 / 69 Section 1 Second-Order Linear

More information

Time-dependent variational forms

Time-dependent variational forms Time-dependent variational forms Hans Petter Langtangen 1,2 1 Center for Biomedical Computing, Simula Research Laboratory 2 Department of Informatics, University of Oslo Oct 30, 2015 PRELIMINARY VERSION

More information

Localized Meshless Methods With Radial Basis Functions For Eigenvalue Problems

Localized Meshless Methods With Radial Basis Functions For Eigenvalue Problems The University of Southern Mississippi The Aquila Digital Community Honors Theses Honors College 5-2013 Localized Meshless Methods With Radial Basis Functions For Eigenvalue Problems Amy M. Kern Follow

More information

Lecture 17: Initial value problems

Lecture 17: Initial value problems Lecture 17: Initial value problems Let s start with initial value problems, and consider numerical solution to the simplest PDE we can think of u/ t + c u/ x = 0 (with u a scalar) for which the solution

More information

Introduction to PDEs and Numerical Methods Tutorial 5. Finite difference methods equilibrium equation and iterative solvers

Introduction to PDEs and Numerical Methods Tutorial 5. Finite difference methods equilibrium equation and iterative solvers Platzhalter für Bild, Bild auf Titelfolie hinter das Logo einsetzen Introduction to PDEs and Numerical Methods Tutorial 5. Finite difference methods equilibrium equation and iterative solvers Dr. Noemi

More information

Math Practice Exam 3 - solutions

Math Practice Exam 3 - solutions Math 181 - Practice Exam 3 - solutions Problem 1 Consider the function h(x) = (9x 2 33x 25)e 3x+1. a) Find h (x). b) Find all values of x where h (x) is zero ( critical values ). c) Using the sign pattern

More information

Bessel s Equation. MATH 365 Ordinary Differential Equations. J. Robert Buchanan. Fall Department of Mathematics

Bessel s Equation. MATH 365 Ordinary Differential Equations. J. Robert Buchanan. Fall Department of Mathematics Bessel s Equation MATH 365 Ordinary Differential Equations J. Robert Buchanan Department of Mathematics Fall 2018 Background Bessel s equation of order ν has the form where ν is a constant. x 2 y + xy

More information

Polytechnic Institute of NYU MA 2132 Final Practice Answers Fall 2012

Polytechnic Institute of NYU MA 2132 Final Practice Answers Fall 2012 Polytechnic Institute of NYU MA Final Practice Answers Fall Studying from past or sample exams is NOT recommended. If you do, it should be only AFTER you know how to do all of the homework and worksheet

More information

5. FVM discretization and Solution Procedure

5. FVM discretization and Solution Procedure 5. FVM discretization and Solution Procedure 1. The fluid domain is divided into a finite number of control volumes (cells of a computational grid). 2. Integral form of the conservation equations are discretized

More information

Chapter 1: Systems of Linear Equations

Chapter 1: Systems of Linear Equations Chapter : Systems of Linear Equations February, 9 Systems of linear equations Linear systems Lecture A linear equation in variables x, x,, x n is an equation of the form a x + a x + + a n x n = b, where

More information

MATH 251 Final Examination December 16, 2015 FORM A. Name: Student Number: Section:

MATH 251 Final Examination December 16, 2015 FORM A. Name: Student Number: Section: MATH 5 Final Examination December 6, 5 FORM A Name: Student Number: Section: This exam has 7 questions for a total of 5 points. In order to obtain full credit for partial credit problems, all work must

More information

Module 3: BASICS OF CFD. Part A: Finite Difference Methods

Module 3: BASICS OF CFD. Part A: Finite Difference Methods Module 3: BASICS OF CFD Part A: Finite Difference Methods THE CFD APPROACH Assembling the governing equations Identifying flow domain and boundary conditions Geometrical discretization of flow domain Discretization

More information

Weighted Residual Methods

Weighted Residual Methods Weighted Residual Methods Introductory Course on Multiphysics Modelling TOMASZ G. ZIELIŃSKI bluebox.ippt.pan.pl/ tzielins/ Table of Contents Problem definition. oundary-value Problem..................

More information

Problem Set 4 Issued: Wednesday, March 18, 2015 Due: Wednesday, April 8, 2015

Problem Set 4 Issued: Wednesday, March 18, 2015 Due: Wednesday, April 8, 2015 MASSACHUSETTS INSTITUTE OF TECHNOLOGY DEPARTMENT OF MECHANICAL ENGINEERING CAMBRIDGE, MASSACHUSETTS 0139.9 NUMERICAL FLUID MECHANICS SPRING 015 Problem Set 4 Issued: Wednesday, March 18, 015 Due: Wednesday,

More information

6 Non-homogeneous Heat Problems

6 Non-homogeneous Heat Problems 6 Non-homogeneous Heat Problems Up to this point all the problems we have considered for the heat or wave equation we what we call homogeneous problems. This means that for an interval < x < l the problems

More information

1 Finite difference example: 1D implicit heat equation

1 Finite difference example: 1D implicit heat equation 1 Finite difference example: 1D implicit heat equation 1.1 Boundary conditions Neumann and Dirichlet We solve the transient heat equation ρc p t = ( k ) (1) on the domain L/2 x L/2 subject to the following

More information

CLASSIFICATION AND PRINCIPLE OF SUPERPOSITION FOR SECOND ORDER LINEAR PDE

CLASSIFICATION AND PRINCIPLE OF SUPERPOSITION FOR SECOND ORDER LINEAR PDE CLASSIFICATION AND PRINCIPLE OF SUPERPOSITION FOR SECOND ORDER LINEAR PDE 1. Linear Partial Differential Equations A partial differential equation (PDE) is an equation, for an unknown function u, that

More information

Mth Review Problems for Test 2 Stewart 8e Chapter 3. For Test #2 study these problems, the examples in your notes, and the homework.

Mth Review Problems for Test 2 Stewart 8e Chapter 3. For Test #2 study these problems, the examples in your notes, and the homework. For Test # study these problems, the examples in your notes, and the homework. Derivative Rules D [u n ] = nu n 1 du D [ln u] = du u D [log b u] = du u ln b D [e u ] = e u du D [a u ] = a u ln a du D [sin

More information

Numerical Solutions to Partial Differential Equations

Numerical Solutions to Partial Differential Equations Numerical Solutions to Partial Differential Equations Zhiping Li LMAM and School of Mathematical Sciences Peking University Numerical Methods for Partial Differential Equations Finite Difference Methods

More information

Hermite Interpolation

Hermite Interpolation Jim Lambers MAT 77 Fall Semester 010-11 Lecture Notes These notes correspond to Sections 4 and 5 in the text Hermite Interpolation Suppose that the interpolation points are perturbed so that two neighboring

More information

On the use of multipole methods for domain integration in the BEM

On the use of multipole methods for domain integration in the BEM On the use of multipole methods for domain integration in the BEM A.A. Mammoii, M.S. Ingber & M.J. Brown Department of Mechanical Engineering, University of New Mexico, USA Abstract Often, a single application

More information

Finite difference methods for the diffusion equation

Finite difference methods for the diffusion equation Finite difference methods for the diffusion equation D150, Tillämpade numeriska metoder II Olof Runborg May 0, 003 These notes summarize a part of the material in Chapter 13 of Iserles. They are based

More information

The Klein-Gordon equation

The Klein-Gordon equation Lecture 8 The Klein-Gordon equation WS2010/11: Introduction to Nuclear and Particle Physics The bosons in field theory Bosons with spin 0 scalar (or pseudo-scalar) meson fields canonical field quantization

More information

An improved subspace selection algorithm for meshless collocation methods

An improved subspace selection algorithm for meshless collocation methods An improved subspace selection algorithm for meshless collocation methods Leevan Ling 1 and Robert Schaback 2 1 Department of Mathematics, Hong Kong Baptist University, Kowloon Tong, Hong Kong. 2 Institut

More information

Fundamentals Physics

Fundamentals Physics Fundamentals Physics And Differential Equations 1 Dynamics Dynamics of a material point Ideal case, but often sufficient Dynamics of a solid Including rotation, torques 2 Position, Velocity, Acceleration

More information

CHAPTER III HAAR WAVELET METHOD FOR SOLVING FISHER S EQUATION

CHAPTER III HAAR WAVELET METHOD FOR SOLVING FISHER S EQUATION CHAPTER III HAAR WAVELET METHOD FOR SOLVING FISHER S EQUATION A version of this chapter has been published as Haar Wavelet Method for solving Fisher s equation, Appl. Math.Comput.,(ELSEVIER) 211 (2009)

More information

Tutorial 2. Introduction to numerical schemes

Tutorial 2. Introduction to numerical schemes 236861 Numerical Geometry of Images Tutorial 2 Introduction to numerical schemes c 2012 Classifying PDEs Looking at the PDE Au xx + 2Bu xy + Cu yy + Du x + Eu y + Fu +.. = 0, and its discriminant, B 2

More information

Numerical Solution Techniques in Mechanical and Aerospace Engineering

Numerical Solution Techniques in Mechanical and Aerospace Engineering Numerical Solution Techniques in Mechanical and Aerospace Engineering Chunlei Liang LECTURE 3 Solvers of linear algebraic equations 3.1. Outline of Lecture Finite-difference method for a 2D elliptic PDE

More information

Existence Theory: Green s Functions

Existence Theory: Green s Functions Chapter 5 Existence Theory: Green s Functions In this chapter we describe a method for constructing a Green s Function The method outlined is formal (not rigorous) When we find a solution to a PDE by constructing

More information

A Meshless Method for the Laplace and Biharmonic Equations Subjected to Noisy Boundary Data

A Meshless Method for the Laplace and Biharmonic Equations Subjected to Noisy Boundary Data Copyright c 4 Tech Science Press CMES, vol.6, no.3, pp.3-61, 4 A Meshless Method for the Laplace and Biharmonic Equations Subjected to Noisy Boundary Data B. Jin 1, Abstract: In this paper, we propose

More information

Radial point interpolation based finite difference method for mechanics problems

Radial point interpolation based finite difference method for mechanics problems INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING Int. J. Numer. Meth. Engng 26; 68:728 754 Published online 11 April 26 in Wiley InterScience (www.interscience.wiley.com). DOI: 1.12/nme.1733

More information

Separation of Variables

Separation of Variables Separation of Variables A typical starting point to study differential equations is to guess solutions of a certain form. Since we will deal with linear PDEs, the superposition principle will allow us

More information

Final: Solutions Math 118A, Fall 2013

Final: Solutions Math 118A, Fall 2013 Final: Solutions Math 118A, Fall 2013 1. [20 pts] For each of the following PDEs for u(x, y), give their order and say if they are nonlinear or linear. If they are linear, say if they are homogeneous or

More information

Implicitly Defined High-Order Operator Splittings for Parabolic and Hyperbolic Variable-Coefficient PDE Using Modified Moments

Implicitly Defined High-Order Operator Splittings for Parabolic and Hyperbolic Variable-Coefficient PDE Using Modified Moments Implicitly Defined High-Order Operator Splittings for Parabolic and Hyperbolic Variable-Coefficient PDE Using Modified Moments James V. Lambers September 24, 2008 Abstract This paper presents a reformulation

More information

In this worksheet we will use the eigenfunction expansion to solve nonhomogeneous equation.

In this worksheet we will use the eigenfunction expansion to solve nonhomogeneous equation. Eigen Function Expansion and Applications. In this worksheet we will use the eigenfunction expansion to solve nonhomogeneous equation. a/ The theory. b/ Example: Solving the Euler equation in two ways.

More information

Find the Fourier series of the odd-periodic extension of the function f (x) = 1 for x ( 1, 0). Solution: The Fourier series is.

Find the Fourier series of the odd-periodic extension of the function f (x) = 1 for x ( 1, 0). Solution: The Fourier series is. Review for Final Exam. Monday /09, :45-:45pm in CC-403. Exam is cumulative, -4 problems. 5 grading attempts per problem. Problems similar to homeworks. Integration and LT tables provided. No notes, no

More information

Finite difference method for heat equation

Finite difference method for heat equation Finite difference method for heat equation Praveen. C praveen@math.tifrbng.res.in Tata Institute of Fundamental Research Center for Applicable Mathematics Bangalore 560065 http://math.tifrbng.res.in/~praveen

More information

Answers to Exercises Computational Fluid Dynamics

Answers to Exercises Computational Fluid Dynamics Answers to Exercises Computational Fluid Dynamics Exercise - Artificial diffusion upwind computations.9 k=. exact.8 k=..7 k=..6 k=..5.4.3.2...2.3.4.5.6.7.8.9 x For k =.,. and., and for N = 2, the discrete

More information

Evaluation of Chebyshev pseudospectral methods for third order differential equations

Evaluation of Chebyshev pseudospectral methods for third order differential equations Numerical Algorithms 16 (1997) 255 281 255 Evaluation of Chebyshev pseudospectral methods for third order differential equations Rosemary Renaut a and Yi Su b a Department of Mathematics, Arizona State

More information

Finite Differences: Consistency, Stability and Convergence

Finite Differences: Consistency, Stability and Convergence Finite Differences: Consistency, Stability and Convergence Varun Shankar March, 06 Introduction Now that we have tackled our first space-time PDE, we will take a quick detour from presenting new FD methods,

More information

Lecture Introduction

Lecture Introduction Lecture 1 1.1 Introduction The theory of Partial Differential Equations (PDEs) is central to mathematics, both pure and applied. The main difference between the theory of PDEs and the theory of Ordinary

More information

Separation of Variables in Linear PDE: One-Dimensional Problems

Separation of Variables in Linear PDE: One-Dimensional Problems Separation of Variables in Linear PDE: One-Dimensional Problems Now we apply the theory of Hilbert spaces to linear differential equations with partial derivatives (PDE). We start with a particular example,

More information

A Robust Preconditioner for the Hessian System in Elliptic Optimal Control Problems

A Robust Preconditioner for the Hessian System in Elliptic Optimal Control Problems A Robust Preconditioner for the Hessian System in Elliptic Optimal Control Problems Etereldes Gonçalves 1, Tarek P. Mathew 1, Markus Sarkis 1,2, and Christian E. Schaerer 1 1 Instituto de Matemática Pura

More information

Midterm 2: Sample solutions Math 118A, Fall 2013

Midterm 2: Sample solutions Math 118A, Fall 2013 Midterm 2: Sample solutions Math 118A, Fall 213 1. Find all separated solutions u(r,t = F(rG(t of the radially symmetric heat equation u t = k ( r u. r r r Solve for G(t explicitly. Write down an ODE for

More information

The Riccati transformation method for linear two point boundary problems

The Riccati transformation method for linear two point boundary problems Chapter The Riccati transformation method for linear two point boundary problems The solution algorithm for two point boundary value problems to be employed here has been derived from different points

More information

Applied Numerical Analysis Homework #3

Applied Numerical Analysis Homework #3 Applied Numerical Analysis Homework #3 Interpolation: Splines, Multiple dimensions, Radial Bases, Least-Squares Splines Question Consider a cubic spline interpolation of a set of data points, and derivatives

More information

Lecture Notes on Numerical Schemes for Flow and Transport Problems

Lecture Notes on Numerical Schemes for Flow and Transport Problems Lecture Notes on Numerical Schemes for Flow and Transport Problems by Sri Redeki Pudaprasetya sr pudap@math.itb.ac.id Department of Mathematics Faculty of Mathematics and Natural Sciences Bandung Institute

More information