W.F. Ames, Numerical Methods for Partial Differential Equations, Academic Press, San Diego, 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....we analyzed simulated data from a dynamical system of the CGLE type [5,12] The PDE of this system reads ;1 : Z2, 2 : ix Z)Z2 Zl aZ Z q OxxZl q KOxxZ2 with parameter values ix = 0.2, a = 2.08 and x = 1. Data were simulated with random initial conditions with the method of lines [15] with sufficiently accurate temporal and spatial discretizations so that the a posteriori error was below 0.01 [16] The system length is 100.0 (Xb = 0.0, Xrb = 100.0) while the integration time is 60.0 (to = 0.0, tf = 60.0) The number of observed spatial data points per time point was 32 and ....

W.F. Ames, Numerical Methods for Partial Differential Equations, Academic Press, San Diego, 1992.

....element efflE vol e volume of element efflE The terms under the sum are position dependent and their value is zero outside of the electrical circuit elements. They can be considered as heat sources within certain volume parts. Let us now apply a standard discretization technique (see e.g. [1]) of finite differences resulting in a space grid. At each grid point we find a difference equation derived from (1) where the space derivatives of the Delta operator are replaced by weighted mean values. It can be assumed that this grid has been chosen fine enough to yield the desired accuracy ....

W.F. Ames, "Numerical Methods for Partial Differential Equations", 2nd ed., Academic Press, New York, 1977.

....to different categories of partial differential equations and their numerical solution. To perceive the general view, it may be helpful to make oneself familiar with the connection between classes of hyperbolic, parabolic and elliptic equations. This theme is illustrated clearly in Ames [2]. Being familiar with the theory of functional analysis may give a superior view to the thematic at hand. An excellent introduction to functional analysis is provided by Brezis [3] Some very important issues like errors and convergence of a numerical solution or its existence are not considered ....

....with the theory of functional analysis may give a superior view to the thematic at hand. An excellent introduction to functional analysis is provided by Brezis [3] Some very important issues like errors and convergence of a numerical solution or its existence are not considered seriously. Ames [2] and Mitchell [4] includes a mild survey to these issues and furthermore they provide several references for further studying. 1. Potential equation and its discretization Let D be a bounded area and B its boundary. Let us consider a problem B x g u R D x f au u # = # # = # where a 0 ....

[Article contains additional citation context not shown here]

W.F.Ames, Numerical methods for partial differential equations.

.... (x) and (x) These two equations form the basis of our simulation. Since these two equations involve mostly trigonometric functions, they are two coupled second order partial differencial equations with strong nonlinearity. We use iterational relaxation method to find the equilibrium solutions [8, 9]. In each iteration, NewtonRaphson s method using derivatives is used together with a line search algorithm to find the root of each equation. Equilibrium state is assumed when both changes of and are smaller than 1:0 Theta 10 Gamma6 radian sec. With a SiliconGraphis IRIS workstation, the ....

Ames, W.F. 1977, Numerical Methods for Partial Differential Equations,

....ffl a is the anisotropy of dielectric constant. The derived torque equations can be solved numerically for a given electric field distribution, and the electric field can be solved with Maxwell s equations, provided the dielectric tensor for each point is known. We use the iterational method [3,4] to solve these coupled equations. One of the simulation results for the director field is shown in Figure 4(a) The voltage on the left is 2 volts, and on the right is 0. Material constants are taken for 6CB, which is manufactured by BDH company. In Figure 4(b) we show the corresponding electric ....

Ames, W.F. 1977, Numerical Methods for Partial Differential Equations, 2nd ed. (New York, Academic Press).

....Semi implicit integration [28] which has a smooth result and numerical accuracy can be considered as an ideal scheme. However, it requires longer run times and results in more complex algorithms. Implicit Euler integration has proven to be better adapted to solve such deformation problems [29]. The idea is to replace the forces at time step t by the forces at time step t 1. Based on the implicit method, we obtain an increment of the velocity vector: 11 1 1 2 where 1 (3) nnn nnnn Vrrt t VWFKVtCV m tt WKC mm = If F n ....

W.F. Ames, 1977, Numerical Methods for Partial Differential Equations, 2 nd ed., Academic Press, 1977.

....equation is written as a difference equation. Estimates of errors between the solution of the problem and that obtained depend on the boundedness of the partial derivatives of some order. A major difficulty is that these estimates cannot be valid at or near a singular point of the domain [3]. In the finite element method local estimates are used when handling singular points [10] but these are only available for some specific problems. There is no systematic way to obtain such estimates. A method which has been known to give more accurate results is the Green s function approach ....

W. F. Ames, "Numerical Methods for Partial Differential Equations," 2nd edition, Academic Press, New York, 1977.

....(n) m where we have used r = c= 2h) The time stepping schemes of (3.6) and (3.7) are known respectively as explicit Euler and implicit Euler, regardless of the particular spatial discretization used. More details on finite difference schemes and their properties can be found in works such as [6, 63, 75, 85]. Before leaving the subject, however, it seems fitting 30 to remark here that explicit schemes are subject to restrictions on the size of the timestep relative to the space step h. To see this in the present context of the advection equation, note that this equation has analytic solution f(x ....

....a method using collocation in both space and time. Given an approximation U(x; t) to the exact solution u(x; t) of a partial differential equation, there are a number of important associated properties. Here we simply introduce the ideas and for more details refer the reader to the works of Ames [6], Gottlieb and Orszag [43] Hall and Porsching [47] Richtmyer and Morton [75] Smith [85] and references therein. It will be convenient to write the PDE in the form Ku = 0, where K is an evolutionary partial differential operator, and to represent the numerical method by KU = 0. The ....

W. F. Ames. Numerical Methods for Partial Differential Equations. Acedemic Press, second edition, 1977. 97

....stable finite difference algorithms. This is due to their necessarily wide stencil which degrades diagonal dominance for implicit schemes. 6] For an interesting discussion on how crossderivatives affect weather prediction, for example, and the methods used to circumvent problems, confer Ames. [7] In the author s estimation, the reduction in terms, and in complexity, from Eq. 32 to that of Eqs. 35 and 36 is surprising. Just as surprising is the fact that the resultant equations have an exact solution form. 8] The solution is non trivial and is being studied by the author. Finite difference ....

W. F. Ames, Numerical Methods for Partial Differential Equations, 2 nd edition, Academic Press, New York, 1977.

....(11) seems to be exaggerated such that we consider the solution obtained with (12) to be better. The computational superiority of the deformation tensor to the gradient formulation of the viscous terms and turbulent diffusion terms is consistent with an early observation of Schultz cited in Ames [Ame77] of the superiority of tensor artificial viscosity in multidimensional finite difference calculations. It is interesting that nonlinear artificial viscosities were used already in the 1950 s by von Neumann for compressible flow with shocks, see e.g. Ames [Ame77] Richtmyer and Morton [RM67] 990 ....

....observation of Schultz cited in Ames [Ame77] of the superiority of tensor artificial viscosity in multidimensional finite difference calculations. It is interesting that nonlinear artificial viscosities were used already in the 1950 s by von Neumann for compressible flow with shocks, see e.g. Ames [Ame77], Richtmyer and Morton [RM67] 990 991 992 993 994 995 996 997 998 999 1000 10.15 10.2 10.25 10.3 10.35 10.4 10.45 10.5 10.55 10.6 10.65 time 1 Galerkin FEM, deformation tensor Galerkin FEM, gradient Figure 13: H 1 semi norm for Galerkin finite element method of the Navier Stokes equations, ....

W. Ames. Numerical Methods for Partial Differential Equations. Academic Press, New York, 1977.

.... Then the required partial derivatives can be approximated as Substituting these approximations into (15) gives our iterative solution to GVF as follows: 16a) 16b) where (17) Convergence of the above iterative process is guaranteed by a standard result in the theory of numerical methods (cf. [25]) Provided that , and are bounded, 16) is stable whenever the Courant Friedrichs Lewy step size restriction is maintained. Since normally , and are fixed, using the definition of in (17) we find that the following restriction on the time step must be maintained in order to guarantee ....

W. F. Ames, Numerical Methods for Partial Differential Equations, 3rd ed. New York: Academic, 1992.

....in much more computation. One can see from Tables 3.1 and 3.2 that Scheme 2 gives better approximate solutions than Scheme 1, so it is reasonable to prefer Scheme 2 even when stability is not an issue. Since Scheme 2 is also more stable than Scheme 1, it is preferred from this point of view too [1]. 3.2 Tree Methods 3.2.1 Basic Ideas The tree method is also called the Lattice method. It can be used to value quite general derivative securities, and to obtain an exact formula by taking the limit, in the case that the lattice solution converges to a continuous time solution [8] 10] The ....

Ames, W. F. 1977. Numerical Methods for Partial Differential Equations. Academic Press.

....ph and qh, and to have known values b on the perimeter of S. If S is sub divided into a network of squares of side h, then the mesh points are defined by: x = ih; i = 0; 1; p) y = jh; j = 0; 1; q) 4. 2) Approximating the equation by the five point difference scheme [Smi65, Ame69] yields: u i Gamma1;j u i 1;j u i;j Gamma1 u i;j 1 Gamma 4u i;j Gamma h 2 f i;j = 0: 4.3) u i;j is then given by: u i;j = 1 4 (u i Gamma1;j u i 1;j u i;j Gamma1 u i;j 1 Gamma h 2 f i;j ) 4.4) If the nth iterative value of u i;j is denoted by u n i;j , then an ....

William Ames. Numerical Methods for Partial Differential Equations. Thomas Nelson and Sons Ltd., 1969.

....of one factor interest rate models, which in general do not have closed form solutions and thus require numerical methods. In this chapter we will consider numerical methods for one factor models. 4. 1 Introduction There are three kinds of numerical methods for option models (see, for example, [Ames], Hull] GKO] 1. Monte Carlo methods, 2. PDE based numerical methods that include finite difference methods, finite element methods and finite volume methods, though only finite difference methods will be considered here, 3. tree methods. Monte Carlo methods are easy to implement, but are ....

.... Delta f c 2 1 2 Delta( Deltay) 2 g pm = Deltat Delta f b 1 (i;j) Deltay Gamma c 2 1 ( Deltay) 2 g; 4.12) p d = 1 Deltat Delta f Gamma b 1 (i;j) Deltay c 2 1 2 Delta( Deltay) 2 g The second finite difference approximation in (4. 11) is a up wind difference scheme ([Ames]) Similarly, in the case of branching down (Figure 4.4.B) we have p u = 1 Deltat Delta f b 1 (i;j) Deltay c 2 1 2 Delta( Deltay) 2 g pm = Deltat Delta f Gamma b 1 (i;j) Deltay Gamma c 2 1 ( Deltay) 2 g (4.13) p d = Deltat Delta f c 2 1 2 Delta( Deltay) 2 g How does ....

[Article contains additional citation context not shown here]

Ames, W., Numerical Methods for Partial Differential Equations, 3rd Ed., Academic Press, Boston, 1993.

.... Gamma OE i;j;n Gamma i;i 1 ) sin(OE i Gamma1;j;n Gamma OE i;j;n Gamma i;i Gamma1 ) sin(OE i;j 1;n Gamma OE i;j;n Gamma j;j 1 ) sin(OE i;j Gamma1;n Gamma OE i;j;n Gamma j;j Gamma1 )g: A.11) We approximate the solutions of Eqs. 5 and 6) by using the leapfrog algorithm[23]. We probed several possibilities for the choices of the fictitious time t = N t Delta t , the number of pseudo time steps N t , and the size of the pseudo time step Delta t . Because of the errors associated with any numerical integration scheme, the pseudo energy will begin to deviate from its ....

W. F. Ames, Numerical Methods for Partial Differential Equations, 2nd ed. ( Academic Press, New York, 1964).

.... and algebraic equations (DAEs) Numerical solutions of initial and boundary problems of ordinary differential equations are given in many books (see e.g. 30, 31, 33 35] Numerical solutions of partial differential equations (PDEs) are e.g. given in books by Lapidus and Pinder [36] or Ames [37]. A recent compilation of literature on DAEs is given by Unger et al. 38] Models for unit operations are scattered over many journals. Books that refer to special types of unit operations are e.g. for chemical reactors [39] or separation processes [40, 41] Phase distribution at equilibrium is ....

Ames, W.F., 1992, Numerical Methods for Partial Differential Equations, Academic Press, New York (3rd Ed.)

....1 62 G, it would read (Hz) ff = z ff Gamma2d 1 Gamma 2z ff Gammad 1 z ff : These alterations correspond to a discretization of the natural boundary conditions u xxx = u xx = 0 across a vertical boundary. Similar conditions can be posed on V: For further details on such discretizations see [1], p. 137. Given the above discretization of boundary conditions together with mild restrictions on the number and location of the nodes in D; it is shown in [4] that the operator P is symmetric and positive definite on the space of grid functions supported in U , i.e. for any grid functions u and ....

....problems performed in [6] show that the stationary ADI method behaves essentially like the classical successive overrelaxation method. However, the rate of convergence of ADI methods can be considerably improved by using a possibly different iteration parameter in each iteration step, see [6] or [1], pp. 206 212. Following again [12] we call the resulting procedure an instationary ADI method. The algorithm for determining these parameters denoted now by ae i , i = 1; m, for some fixed m 2 IN can be formulated as follows and has to be done only once. Algorithm 4.3 (Determination ....

[Article contains additional citation context not shown here]

W. F. Ames, Numerical Methods for Partial Differential Equations, 3rd ed., Academic Press, 1992.

....e Gamma3t , the coefficient being the square of the discrete mode frequency. Our numerical solution of the gradient flow equation was obtained using a predictorcorrector finite difference scheme. Since this is an implicit scheme, there is no stability restriction on Deltat= Deltax 2 (see [1]) The resulting tridiagonal linear system is solved explicitly using the Thomas algorithm. Our space time domain is defined on Gamma10 x 10 and 0 t 65 in space and time steps of 0.02 and 0.005 respectively. Fig. 1 shows the field OE at the initial time and various subsequent times. The ....

....the field dynamics of the K KK system, given by eq. 2.3) The integration is performed using a second order three level implicit formula. As for the gradient flow case, this scheme is stable for any value of Deltat= Deltax provided that the three level parameter is greater than 1 4 (see [1]) Our space time domain is defined on Gamma20 x 20 and 0 t 60 in space and time steps of 0.02 and 0.015 respectively. We have considered two kinds of initial data. First, we have taken the initial data (3.6) with a = 4 and with OE = 0. The initial potential energy is very close to 4. We ....

Ames W F 1992 Numerical Methods for Partial Differential Equations, 3rd edn, (Academic Press, Boston)

.... b i;j Deltat)v n i;j r(v n i 1;j v n i;j 1 v n i Gamma1;j v n i;j Gamma1 Gamma 4v n i;j ) c 2 i;j Deltat (16b) where r = Deltat Deltax Deltay (17) Convergence of the above iterative process is guaranteed by a standard result in the theory of numerical methods (cf. [23]) Provided that b, c 1 , and c 2 are bounded, 16) is stable whenever the Courant FriedrichsLewy step size restriction r 1=4 is maintained. Since normally Deltax, Deltay, and are fixed, using the definition of r in (17) we find that the following restriction on the time step Deltat must ....

W. F. Ames. Numerical Methods for Partial Differential Equations. Academic Press, Boston, 3rd edition, 1992.

....Formulation. The special case of Algorithm LD using ff ij = ff over all edges (i; j) 2 E can be rewritten as the following iterative scheme: w (t 1) I Gamma ffL) w (t) 2) This iteration is also known as the Extrapolated Richardson Method or the First Order Richardson Method [Ame77]. This iteration converges to solution of the equation L w = 0. The solution is the load balanced vector (w; w; w) T . Drawback of Previous Approaches. In real applications, parallel and distributed computers have sparse underlying networks. Both local and global algorithms described ....

....Richardson scheme using Chebyshev acceleration over the first order scheme. 3.3 The Second Order Richardson Method There are several forms of the Second Order Richardson method. Details on the different versions of Second Order Richardson can be found in standard numerical analysis texts e.g. [Ame77, You71, HY81]. For load balancing, we consider the Semi Iterative form of Second Order Richardson for two reasons. ffl It is computationally the most stable form of the Second Order Richardson Methods. ffl We will show that when used for load balancing on a given graph, it suffices for the algorithm to know ....

[Article contains additional citation context not shown here]

W. F. Ames. Numerical Methods for Partial Differential Equations. Academic Press, New York, 1977.

....of constraint groups. Each group interacts with every other group. Written in terms of a matrix problem (there are no source terms) A Delta u = 0; 2) where the m Theta m matrix A (which itself is composed of matrices) is dense. Relaxation techniques are generally O(n 2 ) when not accelerated[1], with the actual convergence rate being determined by the eigenvalues of the iteration matrix. This suggests that the order of the iteration process is O(m 2 ) though the complexity of the process of linear constraint solving which takes place at each step makes this far from certain. 4 ....

William F. Ames. Numerical Methods for Partial Differential Equations. Barnes & Nobel, New York, 1st edition, 1969.

....the behavior of other kinds of semiconductor lasers [2] The numerical integration of the two dimensional governing equations involves a great amount of computational work of the order of several hours, or even days, of CPU time. In this paper, a finite difference method based on the Q technique [3] which includes the explicit, implicit and Crank Nicolson schemes is used to solve the MSSCLA model equations because of the relatively simple geometries considered. The Q method results in a system of nonlinear algebraic equations which are linearized by means of the Newton Raphson method, and ....

Ames, W.F., Numerical Methods for Partial Differential Equations, Academic Press, 1992.

....1997 DRAFT 42 If min (A) approaches a nonzero value as h 0, then the norm of the global error kek 2 0 at least as fast as the truncation error kfk 2 . For more information on the finite difference method and corresponding error analysis and implementational details see [183] 187] 184] [188], 189] 190] 191] C. The Finite Element Method As we have seen above, in the classical numerical treatment for partial differential equations the finite difference method the solution domain is approximated by a grid of uniformly spaced nodes. At each node, the governing differential ....

Ames, W., Numerical Methods for Partial Differential Equations, Academic Press, New York, 1977.

....The curl of v, 5 Theta 5 Phi = 0. This formulation will be suitable for study of irrotational flow fields. Substitution of components of (10) into equation (9) results in the following hyperbolic PDE: Psi x E y Gamma Psi y E x E t = 0 (12) which is a first order equation whose characteristics [2] are the level curves of the projection pictures, E. We can perform the following integration along a curve C in order to invert equation (12) Psi (x; y) Psi 0 Z (x;y) x0 ;y0 ) 5 Psi Delta tds (13) where Psi 0 is the value of Psi at (x 0 ; y 0 ) and t is the tangent to C. If C is a ....

....is a non negative function of x and y. Discretizing the above integral on the pixel grid, we obtain a sum with central difference approximations for partial derivatives of Psi , as well as partial derivatives of E(x; y) Solution of the minimization problem at each pixel is obtained by SOR [2]. As Psi can only be determined up to an additive constant, we set Psi = 0 on the lower boundary and Psi = Gamma on the upper boundary. The latter quantity is the total mass flux in a given vessel with no branchings and may be determined using (14) or with a second variational principle ....

William F. Ames. Numerical Methods for Partial Differential Equations. Academic Press, New York, 1992.

....of the Ctadel system has been developed using the public domain SWI Prolog package [23] 3 Finite Difference Methods In this section a brief introduction to finite difference methods and staggered grids will be given. For a more detailed discussion the reader is referred to a textbook, e.g. [2]. 3.1 Partial Differential Equations A scientific model can be written generally as a set of n partial differential equations in the form t L i (u i ) F i (u1 ; un ) i = 1; n; 1) u v T u v T u v T u v T u v T u v T u v T u v T u v T x y x y T T T T T T T T T u v u v u ....

W.F. Ames, Numerical Methods for Partial Differential Equations, second edition, Academic Press, New York, 1977.

....fixed. When we use finite difference techniques to convert equation (9) into a matrix problem, Ax = b, the resulting matrix problem is also linear. For an excellent discussion of finite difference techniques applied to partial differential equations (PDE) see Press et al. 1988] chapter 19, or Ames [1977]. When the maximum density perturbation in the simulation is small (i.e. less than 5 ) a spectral method can solve equation (9) very rapidly. To apply this method we write it in the form r 2 OE m 1 = rn Delta rOE m rn Theta rOE m T (x; y) n: 10) Oppenheim, Otani and Ronchi: ....

Ames, W. F., Numerical Methods for Partial Differential Equations , Academic Press, 1977.

....studying iterative numerical techniques in general (and relaxation techniques more specifically) I can accomplish my first goal, and that the second will naturally follow from the first. As a starting point, I plan on studying the following references, beginning with the book by William F. Ames: [1, 2, 4, 6, 7]. 3 Bibliography ....

William F. Ames. Numerical Methods for Partial Differential Equations. Barnes & Nobel, New York, 1st edition, 1969.

....function is performed over an occupancy grid representation of the local map. The computation of the harmonic function is executed as a separate process from the path executioner. In order to reduce the number of iterations to be linear in the the number of grid points, Methods of Relaxation (Ames 1992) techniques such as Gauss Seidel iteration with Successive Over Relaxation, combined with the Alter nating Direction (ADI) method were used. A local window is used for computing the potential field to further limit the computation time, and because of the rapid decay of the harmonic function, ....

....distance path. This problem was overcome by making a slight modification to the quad tree representation by using a framed quad tree representation (Chen, Szczerba, Uhran 1997) Computing the harmonic function with a quad tree representation bears some resemblance to the multi grid methods (Ames 1992) for computing solutions to differential equations with the slight difference that in this approach there is no need to compute all of the grid locations at their highest resolution. To take into account the irregular grid spacing in the quad configuration, the iteration kernel is redeveloped from ....

Ames, W. F. 1992. Numerical Methods for Partial Differential Equations. Academic Press Inc.

....grid points as functions of the values of nearby grid points and of higher order terms. Then by neglecting higher order terms, we obtain an equation that expresses the value of a grid point at time t 1 as a function of nearby grid points at time t, and which is accurate to first order. See Ames[2] or Ghez[7] for more detailed discussion of this. This technique leads to the following set of equations, which are accurate to first order and which represent an iterative method of solving the Fokker Planck equation: p t 1=4 x;y; p t x;y; Gamma cos 8 : p t x;y; Gamma p t ....

.... Deltat) 2( Delta ) 2 . These equations require a few comments. First we have split the computation of p t 1 x;y; into four separate steps. This standard fractional method allows us to compute p t 1 x;y; with a series of separate convolutions in each of the three spatial directions. See Ames[2] for comments on the accuracy of this method, and descriptions of other related methods. Second, notice that our computations of the first two terms depend on the sign of cos and sin . To see why this is necessary, consider the problem of computing the first term for a particle traveling in the ....

[Article contains additional citation context not shown here]

W. Ames, Numerical Methods for Partial Differential Equations, Academic Press Inc., 1992.

....of the map. The role of the global planning module is to provide global information to the local path planner. This is necessary because the spatial extent of the local path planner is limited due to its computational requirements increasing linearly, proportional to the number of grid elements [44]. D.1 Local Dynamic Path Planning The problem addressed by the local planning module is discovery and on line planning [45] i.e. dynamic path planning) In this situation, the workspace is initially unknown or partially known. As the robot moves about, it acquires new partial information via its ....

....15 . The computation of the harmonic function is performed over an occupancy grid representation of the local map and is executed as a separate process from the path executioner. In order to make the number of iterations linear in the the number of grid points, Methods of Relaxation [44] techniques such as Gauss Seidel iteration with Successive Over Relaxation, combined with the Alternating Direction (ADI) method were employed. A local window is used for computing the potential field for two reasons: 1) the computation time will linearly increase with the number of grid ....

W. F. Ames, Numerical Methods for Partial Differential Equations. Academic Press Inc., 1992.

.... the number of iterations as much as possible, Methods of Relaxation To appear: IROS 98: IEEE RSJ International Conference on Intelligent Robots and Systems 4 techniques such as Gauss Seidel iteration with Successive Over Relaxation, combined with the Alternating Direction (ADI) method (Ames 1992) were used to achieve O(n) rate of convergence, where n is the number of grid points. A good initial guess to the computation was obtained by using the the heuristic potential field technique (Khatib 1986) This was also at a cost because computing the radiating heuristic potential functions took ....

....on the harmonic function is not necessarily the shortest distance but helps in dealing with uncertainty by minimizing the hitting probability in cooperation with minimizing the distance to the goal. The computation with a quad tree representation bears some resemblance to the multi grid methods (Ames 1992) for computing solutions to differential equations with the slight difference that in this approach there is no need to compute all of the grid locations at their highest resolution. Figure 4 illustrates a quad tree representation for a particular configuration of obstacles, robot and goal ....

Ames, W. F. 1992. Numerical Methods for Partial Differential Equations. Academic Press Inc.

.... of Gm on the stability of explicit methods partially explains the widespread popularity of the semi implicit method referred to in the electrophysiological literature as the Crank Nicolson method (e.g. 45, 52, 115] This is not a direct extension of the wellknown Crank Nicolson method[1] from the linear diffusion problem to the monodomain problem. Rather, it generally refers to a semi implicit treatment of the charge conservation equation, so that the linear term Gm v m is treated implicitly (as in the classic Crank Nicolson method) and the nonlinear term Jm (q; v m ) ....

....a system of N linear algebraic equations at each integration step. For one dimensional domains, the coefficient matrix is tridiagonal and thus may be solved directly in O(N ) operations. For two dimensional domains, iterative methods[91] or implicit treatment in one dimension by the ADI method[1, 16, 52] have generally been preferred. Implicit integration methods In contrast to the common usage of semi implicit methods, fully implicit methods (i.e. those that are implicit in both q and v m ) have rarely been used. Interestingly, Cooley and Dodge[20] used the Trapezoidal Rule method[55] q ....

W. F. Ames. Numerical methods for partial differential equations. Academic Press, Second edition, 1977.

....and conclusions and deal with some issues related to further work. 2 Finite Difference Methods and Staggered Grids In this section, a brief introduction to finite difference methods and staggered grids will be given. For a more detailed discussion the reader is referred to a textbook, e.g. [2]. Finite difference methods combined with staggered grids are at the moment the basic methods employed by the prototype implementation of the Ctadel system to efficiently solve a model numerically. By means of an example the numerical solution of a physical model using finite difference methods ....

....of partial derivatives by finite difference schemes. These schemes are based on difference quotients, that are derived from e.g. Taylor series. In addition, values that are required in the interval between grid points are obtained using interpolation methods. See for more details e.g. [2]. An important consideration in the choice of finite difference schemes is the arrangement of variables on the grid. The use of staggered grids is a common technique employed to u v T u v T u v T u v T u v T u v T u v T u v T u v T x y x y T T T T T T T T T u u u u u u v v v v v v x y T T T T T ....

W.F. Ames, Numerical Methods for Partial Differential Equations, second edition, Academic Press, New York, 1977.

....stable finite difference algorithms. This is due to their necessarily wide stencil which degrades diagonal dominance for implicit schemes. 6] For an interesting discussion on how cross derivatives affect weather prediction, for example, and the methods used to circumvent problems, confer Ames. [7] In the author s estimation, the reduction in terms, and in complexity, from Eq. 32 to that of Eqs. 35 and 36 is surprising. Just as surprising is the fact that the resultant equations have an exact solution form. 8] The solution is non trivial and is being studied by the author. Finite difference ....

W. F. Ames, Numerical Methods for Partial Differential Equations, 2 nd edition, Academic Press, New York, 1977.

No context found.

W. F. Ames, Numerical Methods for Partial Differential Equations. Academic Press, New York (1977).

No context found.

W. F. Ames, Numerical Methods for Partial Differential Equations, 2nd ed. (Academic Press, New York, 1977).

No context found.

Ames W.F. Numerical Methods for Partial Differential Equations, second edi- tion, Academic Press, New York, 1977.

No context found.

W. F. Ames. Numerical Methods for Partial Differential Equations. Academic Press, 1977.

No context found.

W.F. Ames. Numerical Methods for Partial Differential Equations. Barnes and Noble, New York, 1971.

No context found.

W. F. Ames, Numerical Methods for Partial Differential Equations, Academic Press, New York, 1977.

No context found.

W.F. Ames, Numerical Methods for Partial Differential Equations, Academic Press, New York, NY, 1977, pp. 148-157.

No context found.

Ames, W. F., Numerical Methods for Partial Differential Equations, Academic Press, London, 1992.

No context found.

Ames92 W. Ames, Numerical Methods for Partial Differential Equations, Academic Press Inc., 1992.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC