[14] G. Golub and C. V. Loan, Matrix Computations, Johns Hopkins. [15] E. Haber and U. Ascher, Fast nite volume modeling of 3d electromagnetic
|
|
- Deborah Goodwin
- 5 years ago
- Views:
Transcription
1 [4] G. Golub and C. V. Loan, Matrix Computations, Johns Hopkins University Press, 983. [5] E. Haber and U. Ascher, Fast nite volume modeling of 3d electromagnetic problems with highly discontinuous coecients, submited to SISC, (999). [6] R. Hiptmair, Discrete hodge operators, tech. report, Report No. 6, SFB 38, Universitt Tbingen, October 999. [7] J. Jin, The Finite Element Method in Electromagnetics, John Wiley and Sons, 993. [8] P. R. McGillivray, Forward Modelling and Inversion of DC Resistivity and MMR Data, PhD thesis, University of British Columbia, 99. [9] J. D. Moulton, Numerical implementation of nodal methods, PhD thesis, Institute of Applied Mathematics, University of British Columbia, 996. [0] T. Nataka and K. Fujiwara, Summary on the results of benchmark problem 3 (3d nonlinear magnetostatic model), COMPREL, (99), pp. 345{369. [] I. Perugia, V. Simoncini, and M. Arioli, Linear algebra methods in mixed approximation of magnetostatic problems, SISC, (999), pp. 085{0. [] P. Raviart and J. Thomas, A mixed nite element method for second order elliptic problems, vol. 606, Springer-Verlag, Berlin adn New york, 977. [3] Y. Saad, Iterartive Methods for Sparse Linear Systems, PWS Publishing Company, 996. [4] J. Sanz-Serna and M. Calvo, Numerical Hamiltonian Problems, Chapman and Hall, 994. [5] S. Turek, Ecient Solvers For Incompressable Flow Problems, New York: Macmillan,
2 rithms for constrained multibody dynamics, J. Nonlinear Dynamics, (to appear). [] O. Biro and K. Preis, On the use of the magnetic vector potential in the nite element analysis of three-dimensional eddy currents, IEEE Transactions on Magnetics, 5 (989), pp. 345{359. [3] A. Bossavit, Whitney forms: a class of nite elements for threedimensional computations in electromagnetism, IEEE Proc. A, 35 (988), pp. 493{500. [4], Computational Electromagnetism, Academic Press, 998. [5] R. J. Bowers and B. A. Bidwell, Geophysics and uxo detection, The leading edge, 8 (999). n.. [6] F. Brezzi and M. Fortin, Mixed and Hybrid Finite Element Methods, Springer-Verlag, New York, 99. [7] G. Carey and T. Oden, Finite Elements, a second course, Prentice- Hall, 983. [8] N. Demerdash and R. Wang, Theoretical and numerical diculties in 3d vector potential methods in nite element magnetostatic computations, IEEE Trans on Magnetics, MAG-6 (990), pp. 656{658. [9] H. Elman, Iterative methods for problems in computational uid dynamics, Published in Iterative Methods in Scientic Computing, R. Chan, T. Chan and G. Golub, Eds. Springer-Verlag, (997), pp. 7{ 37. [0], Fast stokes solvers, (998). submitted. [] M. Fortin and R. Glowinski, Augmented lagrangian methods: Applications in the numerical solution of boundary-value problems, Studies in Mathematics and Its Applications, 5 (983). [] P. E. Gill, W. Murray, and M. H. Wright, Practical Optimization, Academic Press, New York, 98. [3] V. Girault and P. Raviart, Finite Element Methods for Navier- Stokes Equations, Springer, Berlin-Heidelberg,
3 k grid size CG iterations Flops 8 5 4: : : : : : : : : : : : : : : :0 0 0 Table : Summary of the numerical experiments for the edge elements. roughly comparable and for this small extra eort, we obtain H with second order accuracy vs the rst order accuracy that the edge elements give. In terms of comparison to other codes. It was reported in [8] that nodal methods break when = 0 = 000 and thus our method is obviously superior to the nodal methods. Our method is also better than [] because it requires roughly a third of the storage and it is simpler to program while giving similar results in terms of accuracy. 7 Acknowledgment The author would like to thank Dr. Chen Greif for his thorough reading and useful remarks. References [] E. Bayo and A. Avello, Singularity-free augmented lagrangian algo- 5
4 k grid size CG iterations Flops 8 5 4: : : : : : : : : : : : : : : :3 0 0 Table : Summary of the numerical experiments. In order to check the accuracy of our method and compare its eciency, we use edge elements as suggested in []. The edge elements have been generated on the same grid. In this case the stiness matrix has size of the number of edges, that is and 800. We see that the size of the stiness matrices which evolve from edge elements and the size of the stiness matrix in our method is roughly equivalent (this is because the number of unknowns we have is equivalent tothenumber of faces while in [] the number of unknowns is equivalent to the number of edges). In terms of ll-in, the storage of all our matrices M B ^L requires roughly.5 more disk space. In order to solve the edge element system we have used CG with SSOR preconditioner which makes for a good comparison with our new method. In terms of accuracy, the results of the two methods agree within the discretization error. The computational eort of the edge element discretization is presented in Table. We can see that the edge elements take more iterations to converge but require less oating point operations. This is due to the inversion of the mass matrix in our method. However, the amount ofwork is 4
5 to overcome this problem. It is possible to approximate M by a diagonal mass matrix. This procedure is often called lumping [6]. One such lumping strategy is to use the diagonal nite volume mass matrix [5] as an approximation to M H. Let us dene this matrix as ^MH. In this case ^M ; is a diagonal matrix and we can calculate the matrix ^L = B T ^M; B with moderate expense. The matrix ^L is a good approximation to the matrix L. Therefore, we can use it in order to precondition the system [4]. In this work we have used the SSOR [4, 3] of ^L. The SSOR requires the choice of a parameter 0 <!<whichwehave set to be 0:5. Thus below we apply the CG method to (37) and precondition with the SSOR of ^L. 6 Computational examples In order to test and compare the results we use a model problem similar to []. We assume a uniform cube made of permeable material with permeability = k 0 in the ground where = 0. The grid is uniform spanning the space [;0:8m 0:8m] 3. The cube is of size 0:m 3 and it is located in the middle of the grid from ;0: to0:. The boundary conditions are Dirichlet with (n = n [0 0 ] T. Our goal is to test the accuracy of the method as well as its numerical properties. Since our method results in large but manageable systems we are able to rene our grid and solve the system with and 64 3 cells. This corresponds to problems of size 78, 3056, 0376 and unknowns respectively. Wealsochange the discontinuity ratio k from to 0 3. This allows us to check the method and its robustness. The results are summarized in Table. In the experiments we stop the iterations when the relative residual is less than 0 ;6.From the above we see that the iteration count aswell as the oating point operations grows which iswhatwe expect from a well conditioned system when using preconditioned CG. We also see that the problem is harder to solve as the discontinuity is increased. This is no surprise and our method shares this property with the div-grad system [9]. However although the number of iterations grows with the discontinuity and with the grid, the solution is still manageable even for the 64 3 grid with large discontinuity wherethenumber of unknowns isvery large. 3
6 5 Solution of the resulting linear system In this section we discuss the solution of the discrete system (35). Although, it is possible to design preconditioners directly for the system (35) (see for example [9, 5, ]) we use a Schur complement approach toreducethe system into a smaller dimensional system and precondition the Schur complement. Similar approach is taken in mixed methods for the electrostatic case where the (,) block is easily invertible [7]. We note that such approach is not possible in the formulation [] because the (,) block there is not invertible. The Schur complement system of (36) is obtained by block elimination. First, we can write that h = ;M ; Ba and therefore La= B T M ; Ba= f (37) This system is symmetric positive denite and therefore standard techniques such as CG are eective for its solution. The size of L is equivalent tothe number of faces we have in our discretization and it is therefore a large matrix. Note that in order to apply CG one need not calculate the dense matrix M ; explicitly but rather calculate its action on a vector M ; v. This is easily done by solving the system Mu = v (38) with other iterative methods (here we use CG again). The solution of this system is fast because M is a mass matrix and as discussed in the previous section, its condition number is grid independent. In our numerical experiments we have found that roughly 5-6 preconditioned iterations are needed in order to solve the mass matrix system to accuracy of 0 ;6. The precondition is done by simple scaling [4] and require very little eort. However, while it is very easy to calculate the product of the system (37) with a vector, there remain the problem of preconditioning this system. As stated before, although M is easily invertible M ; is not sparse and therefore L = B T M ; B is not sparse as well. However, there is a simple way
7 and M = e h x h y h z (33) Using the element matrices we can rewrite the variational form as L d = (ht M H h + T M ) ; a T Ch + T Da + f T a (34) where we nowhave the stiness matrices M H M C and D. The matrices M H and M are weighted mass matrices. The matrices M H has six diagonals and the matrix M is diagonal. The matrix C is a representation of the curl operator and as such, it contains a null space which isspannedby the divergence matrix represented by D. Dierentiating the discrete Lagrangian with respect to h a yields the following discrete system. M H 0 ;C T 0 M D ;C D T 0 0 h a 0 A 0 0 A (35) ;f The system (35) is a KKT system of the form M B MH 0 where M = B T 0 0 M C T and B = D (36) The matrix B represents the discretization of the operator [r r ] T which upon consistent discretization (as we have) is of full-rank [3]. The block M represent a mass matrix. For constant, such matrices have condition number which is O() [0] independent on the mesh, and therefore, are easily invertible. In the case that is not constant, the matrix has a condition number which is proportional to max()=min(). However, the condition number of the matrix can be reduced to O() by simple scaling []. This plays an important role when inverting the matrix as we see in the following section. With the above characteristics of M and B, the KKT matrix (36) is invertible [3] however, it is indenite. We now discuss the solution of the discrete system.
8 where h x h y and h z are the cells dimensions. We now rewrite the variational form (5) in a discrete form: h T NN T dv h + dv ; (9) L = X e h T a T e e e (r N W T )dv (W f) dv a ; a T e e (r W )dv + Calculating the integrals in (9) we obtain the element matrices and (9) is rewritten X as h i L = h T M h H + M ; a T C h ; D T a + a T f e Where the matrices M H M C and D are the element matrices. The curl matrix over an element isgiven by where C = 0 h xh z 6 C C 0 ; hxhy C 6 h xh y C 6 0 ; hyhz ; hxhz C h yh z C ; ; ; ; and C = 6 C A (30) ; ; ; ; The div matrix over an element isgiven by D = ; ;h z h y h z h y ;h z h x h z h x ;h x h y h x h y Finally, the mass matrices are given by 0 M 36 eh x h y h z M e 0 0 H 0 36 eh x h y h z M e eh x h y h z M e where M e = C A (3) A (3)
9 Thus in each element, the unknowns A H can be written as where and W i N i A = H = 6X X i= i= = i W i a i = W T a (6a) N i h i = N T h (6b) = W xi ^x W i+ = W yi ^y W i+4 = W zi ^z i = : = N xi ^x N i+4 = N yi ^y N i+8 = N zi ^z i = 3 4: (6c) and = are a constant function over the element. The functions W are given by W x W y W z = h x ( h x = h y ( h y = h z ( h z and the functions N and are given by N x = ( h y h y h z ; y)(h z N x3 = ( h y h y h z ; y)(h z N y = ( h x h x h z ; x)(h z N = y3 ( h x h x h z ; x)(h z N z = ( h x h x h y ; x)(h y N z3 = ( h x h x h y ; x)(h y ; x) W x ; y) W y ; z) W z = h x ( h x + x) = h y ( h y + y) = h z ( h z + z) ; z) N x = h y h z ( h y + y)(h z ; z) + z) N x4 = h y h z ( h y + y)(h z + z) ; z) N y = h x h z ( h x + x)(h z ; z) + z) N y4 = h x h z ( h x + x)(h z + z) ; y) N z = h x h y ( h x + x)(h y ; y) + y) N z4 = h x h y ( h x + x)(h y + y) 9
10 = r A at the nodes. However, the elds A at the nodes in cases of large discontinuities are not well behaved and may present singularities (see [3]) which may result in poor accuracy of r A at the nodes (where we calculate in the case of low order interpolation). On the other hand, (3) implies that we have A H(div) and = r A needs to be only in H 0 which is similar to dene it at cell centers where A and r A are well dened. We therefore use the weak form (5) in order to discretize the system. 4 Discretization of the weak formulation In order to discretize (5) we divide our domain into rectangular elements and pick the variables to be of a low level interpolation functions over the elements. The vector potential A is chosen to be approximated by art element []. The magnetic eld is chosen to be a lowest order Whitney or edge element [3, 7] and nally, the scalar is chosen to be piecewise constant. The numbering of the unknowns on an element is in Figure. Az() Hx(3) Hz(3) Hy() Ay() Hz(4) Hy(4) Hx(4) Ax() Hy() Hz() Hx() Hy(3) Hz() Ax() Az() Hx() Ay() Figure : Discretization of the unknowns over an element 8
11 3 Weak form of the system In order to discretize the system () we rst present it in its weak form. It is easy to verify that the system () corresponds to the following constraint optimization problem minimize (H + ) dv (a) subject to r H ;r ; f =0 (b) Note that A does not appear in the constrained problem. This is similar to the weak form of the div-grad system (3a) and A is introduced as a Lagrange multiplier in the equivalent Lagrangian L = (H + ) ; A (r H ;r ; f) dv (3) So far in this paper we avoided the question of boundary conditions. However, the above Lagrangian is equivalent to the PDE()only for the following Natural boundary conditions (n = 0 = 0 (4b) Note that as discussed, the BC (4b) implies that r A =0. Here we assume for simplicity only natural BC but other BC can be added [7]. In this weak form we have A H 0, H H(curl) and H and the smoothness of A is relaxed while the smoothness of H and is enhanced. This is important as it allows us to nd better approximations for the magnetic eld H than for the magnetic vector potential A. While it is important tohave the smoothness of H enhanced, is only a ctitious variable and we can relax its smoothness byintegrating by parts, obtaining L = (H + ) ; r H A ; (r A) + A f dv (5) Although it is possible to discretize either (3) or (5) we have found that (5) behaves better for very large discontinuities. We suspect that the main reason is that in (3) is at least in H which implies continuity of 7
12 which is closely connected to augmented Lagrangian and penalty methods [,, ]. It is possible to write equations (6a) and (7) in their weak form as min [ ; (r A) ; A f]dv (8) and min [ ; ((r A) +(r A) ) ; A f]dv (9) The condition (6b) can be applied to (8) by choosing A H(curl) (as in []) or it can be approximately applied in (9) (as in [7, 8]). However, as stated in the introduction, these methods lead to low accuracy in H = ; r A because it is obtain by numerical calculation of the curl, however, H is exactly the quantity desired from such calculations. Therefore, we take a dierent approach and similarly to the electrostatic problem, we do not eliminate H from the system. Furthermore, we introduce a scalar function such that and the system is rewritten as r A = (0) r A ; H = 0 (a) r A ; = 0 (b) r H ;r = f (c) Note that in our formulation, equation (c) corresponds to the positive denite system (7). Furthermore, since r f = 0 equation () implies that r = 0 and therefore if we demand Dirichlet boundary conditions on of the form =0,we obtain r A = 0 identically. The advantage of adding the unknowns H is similar to mixed formulations in the electrostatic case and we can design a method that allows higher order interpolation functions for H. Also, introducing allows us to relax the divergence-free condition which is an arbitrary gauge condition. As we show in the next section this formulation implies a dierent weak forms that allows dierent smoothness for the dierent unknowns. 6
13 equipped with the usual norms (see, e.g., [3]). We now note that while in () H the weak formulation (3a) requires to be only in H 0. Also note that in (), the ux J can be obtained only by taking the gradient ofand therefore J H(curl) rather in H(div) as it should be. However, after the decoupling we obtain J H(div) and this allows the use the \right" interpolation for the ux, whichkeeps its continuity across an interface and not across an edge (see [7, 9, 6]). In most practical applications J is approximated by Raviart-Thomas elements whereas is chosen to be approximated by piecewise constant elements. This has been shown to be important especially in the case of large discontinuities in [9]. Our approach to the magnetostatic problem which is driven from the above mixed approach, is stable for very large discontinuities in and as we shall see in Section 5 it is easy to reduce the discrete system into a positive denite system which allows the use of standard and rapidly converging iterative techniques and preconditioners. The paper is divided as follows. In Section we reformulate the problem as a rst order system similar to the electrostatic case. In Section 3 we derive the weak formulation and show that its multidimensionality allows us to use two dierent mixed formulations. In Section 4 we use mixed nite elements in order to discretize the system. In Section 5 we discuss the solution of the system by employing sparse matrix techniques. We conclude this paper in Section 6 where we perform numerical experiments on a model problem. Reformulation of the magnetostatic problem Similarly to the electrostatic case we begin with the equations for the magnetic potential r ; r A = f (6a) r A = 0 (6b) For now, let us ignore the question of BC assuming that they are proper for this problem (see [7, 4]). We shall return to this after we derive our weak form. The system (6) is in principle an over-determined yet consistent. This system is made positive denite by adding a stabilizer [8, 5] r ; r A ;r ; r A = f (7) 5
14 In this paper we take a new approach which can be seen as a combination of the two above mentioned approaches namely, we use the magnetic potential but view the constitutive relation (9c) in a weak sense. Our motivation is drawn from the electrostatic problem where mixed type methods are commonly used to solve for large discontinuities (see for example [7, 6,,, 9] and others). In the electrostatic problem the div-grad system r r = f (0) can be written in a weak form as a minimization problem min [ (r) + f] dv () A common discretization of this weak formulation is done by using nodal elements [7, 7, 9]. In mixed methods for electrostatics, the div-grad equation is decoupled into two rst order systems r J = f (a) ; J ;r = 0 (b) In its weak form, the problem is a constraint optimization problem of the form min J dv (3a) subject to r J ; f =0 (3b) Note that does not appear in the system and it can be viewed as a Lagrange multiplier of the equivalent Lagrangian L = [ ; J + (r J ; f)] dv (4) In order to see the advantage of the weak form (4) we dene the regular Sobolov spaces H 0 = fv L ()g (5) H = fv L () rv L ()g H(div) =fv L () r v L ()g H(curl) =fv L () r v L ()g 4
15 Note that (5) diers from () only in the source term. The solution of the system (5) has been addressed in [7, 3,,, 8] and others. In [, 7, 8] (and many others) the magnetic potential A was introduced such that H = r A (6) This condition satises (5b) and we obtain a system for the vector potential A (and appropriate BC for A [7]). r ; r A = f (7) This equation does not have a unique solution due to the null space of the curl operator and therefore a gauge condition for A is introduced r A =0 (8) In [8, 7] a solution of the problem using nodal elements is derived, employing the gauge condition as penalty. This method has been shown to be unstable when contains large discontinuities. Better results where obtained using edge elements for A []. However, this method suer from low accuracy because H is obtained from A using numerical dierentiation. Furthermore, the resulting discretization of the system (7) is semidenite and therefore convergence of iterative methods may beslow. A second approach to the problem is in [3, ] where the magnetic ux B is introduced and the constitutive relation is added to the system. This yields the system r H = f (9a) r B = 0 (9b) B ; H = 0 (9c) and the constitutive relation (9c) is viewed only in a weak sense. This method, although very stable for large discontinuities requires the introduction of three extra Lagrange multipliers and therefore suers from a large number of unknowns. Furthermore the solution of the resulting system requires some non-trivial linear algebra (see the work of []). The main dif- culty of this method is that it leads to very large indenite systems of ve vector elds and therefore it requires large memory and sophisticated solution techniques. 3
16 of permeability in the ground. The problem is also a corner-stone to the solution of Maxwell's equations in the quasi-static case and therefore ecient techniques for its solution in the case of large discontinuities are important. The problem evolves from Maxwell's equations in the static limit (@=@t(:)= 0) which can be written as r E = 0 (a) r H ; E = J s (b) r (H) = 0 (c) with appropriate boundary conditions on H (see [7]). In this paper we assume only Dirichlet type BC i.e. (n =0: There are two dierent magnetostatic problems. First, if the source is divergence free (such as the earth's magnetic eld) i.e. r J s = 0 then, it is easy to see that E = 0 and the system reduces to r H = J s (a) r (H) = 0 (b) This is an overdetermined yet consistent system for H, that is, although there are four equations and three unknowns, it is possible to nd H which solves the system (). Secondly, ifr J s 6= 0 (for example in the MMR problem [8]) then there exist an electric potential, such thate = r and we obtain r H ; r = J s (3a) r (H) = 0: (3b) However, this system can be reduced to a similar form as () by taking the div of (3a), obtaining the well known div-grad electrostatic system r (r) =;r J s (4) Once this system is solved, is recovered and one could obtain H by solving the system r H = J s ; r = f (5a) r (H) = 0 (5b) Note that the solution of the electrostatic equation requires an additional BC on which is not given in the system (3a). This BC is driven from other physical considerations
17 A mixed nite element method for the solution of the magnetostatic problem with highly discontinuous coecients in 3D E. Haber January 4, 000 Abstract This paper present a new a mixed nite element method for the simulation of magnetostatic problems with highly discontinuous permeability. The method is derived from the well studied mixed formulation for the div-grad system that is known to be accurate for very large discontinuities. test model problem. The method robustness is demonstrated on a Keywords: vector potential, Maxwell's equations, solution discontinuities, mixed nite elements, Krylov subspace methods, preconditioning. Introduction Magnetostatic simulation is important and has many applications [, 8, 0, 7]. In geophysics, this simulation is important in order to simulate the response of unexploded ordnances (UXO's) [5] and for the MMR experiment [8]. In the UXO problem one needs to simulate the response of metal with permeability of roughly =5 0 0 in a medium of = 0 and therefore it is important to be able to model very large discontinuities. In the MMR problem one needs to simulate the response of any arbitrary distribution Department of Computer Science and Institute of Applied Mathematics, University of British Columbia, Vancouver, BC, V6T 4, Canada. (haber@cs.ubc.ca).
A MULTIGRID ALGORITHM FOR. Richard E. Ewing and Jian Shen. Institute for Scientic Computation. Texas A&M University. College Station, Texas SUMMARY
A MULTIGRID ALGORITHM FOR THE CELL-CENTERED FINITE DIFFERENCE SCHEME Richard E. Ewing and Jian Shen Institute for Scientic Computation Texas A&M University College Station, Texas SUMMARY In this article,
More informationFast Iterative Solution of Saddle Point Problems
Michele Benzi Department of Mathematics and Computer Science Emory University Atlanta, GA Acknowledgments NSF (Computational Mathematics) Maxim Olshanskii (Mech-Math, Moscow State U.) Zhen Wang (PhD student,
More informationFAST FINITE VOLUME SIMULATION OF 3D ELECTROMAGNETIC PROBLEMS WITH HIGHLY DISCONTINUOUS COEFFICIENTS
SIAM J. SCI. COMPUT. Vol., No. 6, pp. 1943 1961 c 001 Society for Industrial and Applied Mathematics FAST FINITE VOLUME SIMULATION OF 3D ELECTROMAGNETIC PROBLEMS WITH HIGHLY DISCONTINUOUS COEFFICIENTS
More informationParallelizing large scale time domain electromagnetic inverse problem
Parallelizing large scale time domain electromagnetic inverse problems Eldad Haber with: D. Oldenburg & R. Shekhtman + Emory University, Atlanta, GA + The University of British Columbia, Vancouver, BC,
More informationLecture Note III: Least-Squares Method
Lecture Note III: Least-Squares Method Zhiqiang Cai October 4, 004 In this chapter, we shall present least-squares methods for second-order scalar partial differential equations, elastic equations of solids,
More informationSolving Large Nonlinear Sparse Systems
Solving Large Nonlinear Sparse Systems Fred W. Wubs and Jonas Thies Computational Mechanics & Numerical Mathematics University of Groningen, the Netherlands f.w.wubs@rug.nl Centre for Interdisciplinary
More informationDELFT UNIVERSITY OF TECHNOLOGY
DELFT UNIVERSITY OF TECHNOLOGY REPORT 18-05 Efficient and robust Schur complement approximations in the augmented Lagrangian preconditioner for high Reynolds number laminar flows X. He and C. Vuik ISSN
More information7.4 The Saddle Point Stokes Problem
346 CHAPTER 7. APPLIED FOURIER ANALYSIS 7.4 The Saddle Point Stokes Problem So far the matrix C has been diagonal no trouble to invert. This section jumps to a fluid flow problem that is still linear (simpler
More informationEfficient Augmented Lagrangian-type Preconditioning for the Oseen Problem using Grad-Div Stabilization
Efficient Augmented Lagrangian-type Preconditioning for the Oseen Problem using Grad-Div Stabilization Timo Heister, Texas A&M University 2013-02-28 SIAM CSE 2 Setting Stationary, incompressible flow problems
More informationAMS526: Numerical Analysis I (Numerical Linear Algebra) Lecture 23: GMRES and Other Krylov Subspace Methods; Preconditioning
AMS526: Numerical Analysis I (Numerical Linear Algebra) Lecture 23: GMRES and Other Krylov Subspace Methods; Preconditioning Xiangmin Jiao SUNY Stony Brook Xiangmin Jiao Numerical Analysis I 1 / 18 Outline
More informationA Finite Element Method for an Ill-Posed Problem. Martin-Luther-Universitat, Fachbereich Mathematik/Informatik,Postfach 8, D Halle, Abstract
A Finite Element Method for an Ill-Posed Problem W. Lucht Martin-Luther-Universitat, Fachbereich Mathematik/Informatik,Postfach 8, D-699 Halle, Germany Abstract For an ill-posed problem which has its origin
More informationwith Applications to Elasticity and Compressible Flow Daoqi Yang March 20, 1997 Abstract
Stabilized Schemes for Mixed Finite Element Methods with Applications to Elasticity and Compressible Flow Problems Daoqi Yang March 20, 1997 Abstract Stabilized iterative schemes for mixed nite element
More informationHigh Order Differential Form-Based Elements for the Computation of Electromagnetic Field
1472 IEEE TRANSACTIONS ON MAGNETICS, VOL 36, NO 4, JULY 2000 High Order Differential Form-Based Elements for the Computation of Electromagnetic Field Z Ren, Senior Member, IEEE, and N Ida, Senior Member,
More informationKey words. Parallel iterative solvers, saddle-point linear systems, preconditioners, timeharmonic
PARALLEL NUMERICAL SOLUTION OF THE TIME-HARMONIC MAXWELL EQUATIONS IN MIXED FORM DAN LI, CHEN GREIF, AND DOMINIK SCHÖTZAU Numer. Linear Algebra Appl., Vol. 19, pp. 525 539, 2012 Abstract. We develop a
More informationFinite element approximation on quadrilateral meshes
COMMUNICATIONS IN NUMERICAL METHODS IN ENGINEERING Commun. Numer. Meth. Engng 2001; 17:805 812 (DOI: 10.1002/cnm.450) Finite element approximation on quadrilateral meshes Douglas N. Arnold 1;, Daniele
More informationA 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 informationRening the submesh strategy in the two-level nite element method: application to the advection diusion equation
INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN FLUIDS Int. J. Numer. Meth. Fluids 2002; 39:161 187 (DOI: 10.1002/d.219) Rening the submesh strategy in the two-level nite element method: application to
More informationDomain Decomposition Preconditioners for Spectral Nédélec Elements in Two and Three Dimensions
Domain Decomposition Preconditioners for Spectral Nédélec Elements in Two and Three Dimensions Bernhard Hientzsch Courant Institute of Mathematical Sciences, New York University, 51 Mercer Street, New
More informationOUTLINE 1. Introduction 1.1 Notation 1.2 Special matrices 2. Gaussian Elimination 2.1 Vector and matrix norms 2.2 Finite precision arithmetic 2.3 Fact
Computational Linear Algebra Course: (MATH: 6800, CSCI: 6800) Semester: Fall 1998 Instructors: { Joseph E. Flaherty, aherje@cs.rpi.edu { Franklin T. Luk, luk@cs.rpi.edu { Wesley Turner, turnerw@cs.rpi.edu
More informationFEniCS Course. Lecture 0: Introduction to FEM. Contributors Anders Logg, Kent-Andre Mardal
FEniCS Course Lecture 0: Introduction to FEM Contributors Anders Logg, Kent-Andre Mardal 1 / 46 What is FEM? The finite element method is a framework and a recipe for discretization of mathematical problems
More informationA Robust Preconditioned Iterative Method for the Navier-Stokes Equations with High Reynolds Numbers
Applied and Computational Mathematics 2017; 6(4): 202-207 http://www.sciencepublishinggroup.com/j/acm doi: 10.11648/j.acm.20170604.18 ISSN: 2328-5605 (Print); ISSN: 2328-5613 (Online) A Robust Preconditioned
More informationA Review of Preconditioning Techniques for Steady Incompressible Flow
Zeist 2009 p. 1/43 A Review of Preconditioning Techniques for Steady Incompressible Flow David Silvester School of Mathematics University of Manchester Zeist 2009 p. 2/43 PDEs Review : 1984 2005 Update
More informationfield using second order edge elements in 3D
The current issue and full text archive of this journal is available at http://www.emerald-library.com using second order edge elements in 3D Z. Ren Laboratoire de GeÂnie Electrique de Paris, UniversiteÂs
More informationfor Finite Element Simulation of Incompressible Flow Arnd Meyer Department of Mathematics, Technical University of Chemnitz,
Preconditioning the Pseudo{Laplacian for Finite Element Simulation of Incompressible Flow Arnd Meyer Department of Mathematics, Technical University of Chemnitz, 09107 Chemnitz, Germany Preprint{Reihe
More informationEfficient Solvers for the Navier Stokes Equations in Rotation Form
Efficient Solvers for the Navier Stokes Equations in Rotation Form Computer Research Institute Seminar Purdue University March 4, 2005 Michele Benzi Emory University Atlanta, GA Thanks to: NSF (MPS/Computational
More informationMULTIGRID PRECONDITIONING FOR THE BIHARMONIC DIRICHLET PROBLEM M. R. HANISCH
MULTIGRID PRECONDITIONING FOR THE BIHARMONIC DIRICHLET PROBLEM M. R. HANISCH Abstract. A multigrid preconditioning scheme for solving the Ciarlet-Raviart mixed method equations for the biharmonic Dirichlet
More informationMultigrid Approaches to Non-linear Diffusion Problems on Unstructured Meshes
NASA/CR-2001-210660 ICASE Report No. 2001-3 Multigrid Approaches to Non-linear Diffusion Problems on Unstructured Meshes Dimitri J. Mavriplis ICASE, Hampton, Virginia ICASE NASA Langley Research Center
More informationPARTITION OF UNITY FOR THE STOKES PROBLEM ON NONMATCHING GRIDS
PARTITION OF UNITY FOR THE STOES PROBLEM ON NONMATCHING GRIDS CONSTANTIN BACUTA AND JINCHAO XU Abstract. We consider the Stokes Problem on a plane polygonal domain Ω R 2. We propose a finite element method
More informationJasmin Smajic1, Christian Hafner2, Jürg Leuthold2, March 23, 2015
Jasmin Smajic, Christian Hafner 2, Jürg Leuthold 2, March 23, 205 Time Domain Finite Element Method (TD FEM): Continuous and Discontinuous Galerkin (DG-FEM) HSR - University of Applied Sciences of Eastern
More information0 Finite Element Method of Environmental Problems. Chapter where c is a constant dependent only on and where k k 0; and k k ; are the L () and H () So
Chapter SUBSTRUCTURE PRECONDITIONING FOR POROUS FLOW PROBLEMS R.E. Ewing, Yu. Kuznetsov, R.D. Lazarov, and S. Maliassov. Introduction Let be a convex polyhedral domain in IR, f(x) L () and A(x) be a suciently
More informationThe Deflation Accelerated Schwarz Method for CFD
The Deflation Accelerated Schwarz Method for CFD J. Verkaik 1, C. Vuik 2,, B.D. Paarhuis 1, and A. Twerda 1 1 TNO Science and Industry, Stieltjesweg 1, P.O. Box 155, 2600 AD Delft, The Netherlands 2 Delft
More informationt x 0.25
Journal of ELECTRICAL ENGINEERING, VOL. 52, NO. /s, 2, 48{52 COMPARISON OF BROYDEN AND NEWTON METHODS FOR SOLVING NONLINEAR PARABOLIC EQUATIONS Ivan Cimrak If the time discretization of a nonlinear parabolic
More informationDepartment of Computer Science, University of Illinois at Urbana-Champaign
Department of Computer Science, University of Illinois at Urbana-Champaign Probing for Schur Complements and Preconditioning Generalized Saddle-Point Problems Eric de Sturler, sturler@cs.uiuc.edu, http://www-faculty.cs.uiuc.edu/~sturler
More informationANDREA TOSELLI. Abstract. Two-level overlapping Schwarz methods are considered for nite element problems
OVERLAPPING SCHWARZ METHODS FOR MAXWELL'S EQUATIONS IN THREE DIMENSIONS ANDREA TOSELLI Abstract. Two-level overlapping Schwarz methods are considered for nite element problems of 3D Maxwell's equations.
More informationLinear Solvers. Andrew Hazel
Linear Solvers Andrew Hazel Introduction Thus far we have talked about the formulation and discretisation of physical problems...... and stopped when we got to a discrete linear system of equations. Introduction
More informationMathematics and Computer Science
Technical Report TR-2007-002 Block preconditioning for saddle point systems with indefinite (1,1) block by Michele Benzi, Jia Liu Mathematics and Computer Science EMORY UNIVERSITY International Journal
More informationPreconditioners for the incompressible Navier Stokes equations
Preconditioners for the incompressible Navier Stokes equations C. Vuik M. ur Rehman A. Segal Delft Institute of Applied Mathematics, TU Delft, The Netherlands SIAM Conference on Computational Science and
More informationNovel determination of dierential-equation solutions: universal approximation method
Journal of Computational and Applied Mathematics 146 (2002) 443 457 www.elsevier.com/locate/cam Novel determination of dierential-equation solutions: universal approximation method Thananchai Leephakpreeda
More informationSolution of the Two-Dimensional Steady State Heat Conduction using the Finite Volume Method
Ninth International Conference on Computational Fluid Dynamics (ICCFD9), Istanbul, Turkey, July 11-15, 2016 ICCFD9-0113 Solution of the Two-Dimensional Steady State Heat Conduction using the Finite Volume
More information9.1 Preconditioned Krylov Subspace Methods
Chapter 9 PRECONDITIONING 9.1 Preconditioned Krylov Subspace Methods 9.2 Preconditioned Conjugate Gradient 9.3 Preconditioned Generalized Minimal Residual 9.4 Relaxation Method Preconditioners 9.5 Incomplete
More informationStructure preserving preconditioner for the incompressible Navier-Stokes equations
Structure preserving preconditioner for the incompressible Navier-Stokes equations Fred W. Wubs and Jonas Thies Computational Mechanics & Numerical Mathematics University of Groningen, the Netherlands
More informationDiscontinuous Galerkin Methods
Discontinuous Galerkin Methods Joachim Schöberl May 20, 206 Discontinuous Galerkin (DG) methods approximate the solution with piecewise functions (polynomials), which are discontinuous across element interfaces.
More informationThe Algebraic Multigrid Projection for Eigenvalue Problems; Backrotations and Multigrid Fixed Points. Sorin Costiner and Shlomo Ta'asan
The Algebraic Multigrid Projection for Eigenvalue Problems; Backrotations and Multigrid Fixed Points Sorin Costiner and Shlomo Ta'asan Department of Applied Mathematics and Computer Science The Weizmann
More informationSubstructuring Preconditioning for Nonconforming Finite Elements engineering problems. For example, for petroleum reservoir problems in geometrically
SUBSTRUCTURING PRECONDITIONING FOR FINITE ELEMENT APPROXIMATIONS OF SECOND ORDER ELLIPTIC PROBLEMS. I. NONCONFORMING LINEAR ELEMENTS FOR THE POISSON EQUATION IN A PARALLELEPIPED R.E. Ewing, Yu. Kuznetsov,
More informationParallel numerical solution of the time-harmonic Maxwell equations in mixed form
NUMERICAL LINEAR ALGEBRA WITH APPLICATIONS Numer. Linear Algebra Appl. (2011) Published online in Wiley Online Library (wileyonlinelibrary.com)..782 Parallel numerical solution of the time-harmonic Maxwell
More informationRobust multigrid methods for nonsmooth coecient elliptic linear systems
Journal of Computational and Applied Mathematics 123 (2000) 323 352 www.elsevier.nl/locate/cam Robust multigrid methods for nonsmooth coecient elliptic linear systems Tony F. Chan a; ; 1, W.L. Wan b; 2
More informationFast solvers for steady incompressible flow
ICFD 25 p.1/21 Fast solvers for steady incompressible flow Andy Wathen Oxford University wathen@comlab.ox.ac.uk http://web.comlab.ox.ac.uk/~wathen/ Joint work with: Howard Elman (University of Maryland,
More informationMathematics Research Report No. MRR 003{96, HIGH RESOLUTION POTENTIAL FLOW METHODS IN OIL EXPLORATION Stephen Roberts 1 and Stephan Matthai 2 3rd Febr
HIGH RESOLUTION POTENTIAL FLOW METHODS IN OIL EXPLORATION Stephen Roberts and Stephan Matthai Mathematics Research Report No. MRR 003{96, Mathematics Research Report No. MRR 003{96, HIGH RESOLUTION POTENTIAL
More informationMultigrid and Iterative Strategies for Optimal Control Problems
Multigrid and Iterative Strategies for Optimal Control Problems John Pearson 1, Stefan Takacs 1 1 Mathematical Institute, 24 29 St. Giles, Oxford, OX1 3LB e-mail: john.pearson@worc.ox.ac.uk, takacs@maths.ox.ac.uk
More informationMULTIGRID PRECONDITIONING IN H(div) ON NON-CONVEX POLYGONS* Dedicated to Professor Jim Douglas, Jr. on the occasion of his seventieth birthday.
MULTIGRID PRECONDITIONING IN H(div) ON NON-CONVEX POLYGONS* DOUGLAS N ARNOLD, RICHARD S FALK, and RAGNAR WINTHER Dedicated to Professor Jim Douglas, Jr on the occasion of his seventieth birthday Abstract
More informationReduced Synchronization Overhead on. December 3, Abstract. The standard formulation of the conjugate gradient algorithm involves
Lapack Working Note 56 Conjugate Gradient Algorithms with Reduced Synchronization Overhead on Distributed Memory Multiprocessors E. F. D'Azevedo y, V.L. Eijkhout z, C. H. Romine y December 3, 1999 Abstract
More informationMIXED FINITE ELEMENT METHODS FOR PROBLEMS WITH ROBIN BOUNDARY CONDITIONS
MIXED FINITE ELEMENT METHODS FOR PROBLEMS WITH ROBIN BOUNDARY CONDITIONS JUHO KÖNNÖ, DOMINIK SCHÖTZAU, AND ROLF STENBERG Abstract. We derive new a-priori and a-posteriori error estimates for mixed nite
More informationPreconditioned GMRES Revisited
Preconditioned GMRES Revisited Roland Herzog Kirk Soodhalter UBC (visiting) RICAM Linz Preconditioning Conference 2017 Vancouver August 01, 2017 Preconditioned GMRES Revisited Vancouver 1 / 32 Table of
More informationPreconditioning a mixed discontinuous nite element method for radiation diusion
NUMERICAL LINEAR ALGEBRA WITH APPLICATIONS Numer. Linear Algebra Appl. 2004; 11:795 811 (DOI: 10.1002/nla.347) Preconditioning a mixed discontinuous nite element method for radiation diusion James S. Warsa
More informationMultilevel Preconditioning of Graph-Laplacians: Polynomial Approximation of the Pivot Blocks Inverses
Multilevel Preconditioning of Graph-Laplacians: Polynomial Approximation of the Pivot Blocks Inverses P. Boyanova 1, I. Georgiev 34, S. Margenov, L. Zikatanov 5 1 Uppsala University, Box 337, 751 05 Uppsala,
More informationContents. Preface... xi. Introduction...
Contents Preface... xi Introduction... xv Chapter 1. Computer Architectures... 1 1.1. Different types of parallelism... 1 1.1.1. Overlap, concurrency and parallelism... 1 1.1.2. Temporal and spatial parallelism
More informationVector and scalar penalty-projection methods
Numerical Flow Models for Controlled Fusion - April 2007 Vector and scalar penalty-projection methods for incompressible and variable density flows Philippe Angot Université de Provence, LATP - Marseille
More informationSolving the curl-div system using divergence-free or curl-free finite elements
Solving the curl-div system using divergence-free or curl-free finite elements Alberto Valli Dipartimento di Matematica, Università di Trento, Italy or: Why I say to my students that divergence-free finite
More informationThe solution of the discretized incompressible Navier-Stokes equations with iterative methods
The solution of the discretized incompressible Navier-Stokes equations with iterative methods Report 93-54 C. Vuik Technische Universiteit Delft Delft University of Technology Faculteit der Technische
More information1e N
Spectral schemes on triangular elements by Wilhelm Heinrichs and Birgit I. Loch Abstract The Poisson problem with homogeneous Dirichlet boundary conditions is considered on a triangle. The mapping between
More informationThe antitriangular factorisation of saddle point matrices
The antitriangular factorisation of saddle point matrices J. Pestana and A. J. Wathen August 29, 2013 Abstract Mastronardi and Van Dooren [this journal, 34 (2013) pp. 173 196] recently introduced the block
More informationSimultaneous estimation of wavefields & medium parameters
Simultaneous estimation of wavefields & medium parameters reduced-space versus full-space waveform inversion Bas Peters, Felix J. Herrmann Workshop W- 2, The limit of FWI in subsurface parameter recovery.
More informationAMS subject classifications. Primary, 65N15, 65N30, 76D07; Secondary, 35B45, 35J50
A SIMPLE FINITE ELEMENT METHOD FOR THE STOKES EQUATIONS LIN MU AND XIU YE Abstract. The goal of this paper is to introduce a simple finite element method to solve the Stokes equations. This method is in
More informationREGLERTEKNIK AUTOMATIC CONTROL LINKÖPING
Generating state space equations from a bond graph with dependent storage elements using singular perturbation theory. Krister Edstrom Department of Electrical Engineering Linkoping University, S-58 83
More informationA parallel method for large scale time domain electromagnetic inverse problems
A parallel method for large scale time domain electromagnetic inverse problems Eldad Haber July 15, 2005 Abstract In this work we consider the solution of 3D time domain electromagnetic inverse problems
More informationFrom the Boundary Element DDM to local Trefftz Finite Element Methods on Polyhedral Meshes
www.oeaw.ac.at From the Boundary Element DDM to local Trefftz Finite Element Methods on Polyhedral Meshes D. Copeland, U. Langer, D. Pusch RICAM-Report 2008-10 www.ricam.oeaw.ac.at From the Boundary Element
More informationLocal discontinuous Galerkin methods for elliptic problems
COMMUNICATIONS IN NUMERICAL METHODS IN ENGINEERING Commun. Numer. Meth. Engng 2002; 18:69 75 [Version: 2000/03/22 v1.0] Local discontinuous Galerkin methods for elliptic problems P. Castillo 1 B. Cockburn
More information2.29 Numerical Fluid Mechanics Spring 2015 Lecture 9
Spring 2015 Lecture 9 REVIEW Lecture 8: Direct Methods for solving (linear) algebraic equations Gauss Elimination LU decomposition/factorization Error Analysis for Linear Systems and Condition Numbers
More informationAn Efficient Low Memory Implicit DG Algorithm for Time Dependent Problems
An Efficient Low Memory Implicit DG Algorithm for Time Dependent Problems P.-O. Persson and J. Peraire Massachusetts Institute of Technology 2006 AIAA Aerospace Sciences Meeting, Reno, Nevada January 9,
More informationSIMULTANEOUS DETERMINATION OF STEADY TEMPERATURES AND HEAT FLUXES ON SURFACES OF THREE DIMENSIONAL OBJECTS USING FEM
Proceedings of ASME IMECE New ork, N, November -6, IMECE/HTD-43 SIMULTANEOUS DETERMINATION OF STEAD TEMPERATURES AND HEAT FLUES ON SURFACES OF THREE DIMENSIONAL OBJECTS USING FEM Brian H. Dennis and George
More informationLinear Algebra (part 1) : Matrices and Systems of Linear Equations (by Evan Dummit, 2016, v. 2.02)
Linear Algebra (part ) : Matrices and Systems of Linear Equations (by Evan Dummit, 206, v 202) Contents 2 Matrices and Systems of Linear Equations 2 Systems of Linear Equations 2 Elimination, Matrix Formulation
More informationUNIVERSITY OF MARYLAND PRECONDITIONERS FOR THE DISCRETE STEADY-STATE NAVIER-STOKES EQUATIONS CS-TR #4164 / UMIACS TR # JULY 2000
UNIVERSITY OF MARYLAND INSTITUTE FOR ADVANCED COMPUTER STUDIES DEPARTMENT OF COMPUTER SCIENCE PERFORMANCE AND ANALYSIS OF SADDLE POINT PRECONDITIONERS FOR THE DISCRETE STEADY-STATE NAVIER-STOKES EQUATIONS
More informationMultigrid Methods for Saddle Point Problems
Multigrid Methods for Saddle Point Problems Susanne C. Brenner Department of Mathematics and Center for Computation & Technology Louisiana State University Advances in Mathematics of Finite Elements (In
More informationCentro de Processamento de Dados, Universidade Federal do Rio Grande do Sul,
A COMPARISON OF ACCELERATION TECHNIQUES APPLIED TO THE METHOD RUDNEI DIAS DA CUNHA Computing Laboratory, University of Kent at Canterbury, U.K. Centro de Processamento de Dados, Universidade Federal do
More informationA note on accurate and efficient higher order Galerkin time stepping schemes for the nonstationary Stokes equations
A note on accurate and efficient higher order Galerkin time stepping schemes for the nonstationary Stokes equations S. Hussain, F. Schieweck, S. Turek Abstract In this note, we extend our recent work for
More information1. Introduction. The Stokes problem seeks unknown functions u and p satisfying
A DISCRETE DIVERGENCE FREE WEAK GALERKIN FINITE ELEMENT METHOD FOR THE STOKES EQUATIONS LIN MU, JUNPING WANG, AND XIU YE Abstract. A discrete divergence free weak Galerkin finite element method is developed
More informationNumerical behavior of inexact linear solvers
Numerical behavior of inexact linear solvers Miro Rozložník joint results with Zhong-zhi Bai and Pavel Jiránek Institute of Computer Science, Czech Academy of Sciences, Prague, Czech Republic The fourth
More informationON A GENERAL CLASS OF PRECONDITIONERS FOR NONSYMMETRIC GENERALIZED SADDLE POINT PROBLEMS
U..B. Sci. Bull., Series A, Vol. 78, Iss. 4, 06 ISSN 3-707 ON A GENERAL CLASS OF RECONDIIONERS FOR NONSYMMERIC GENERALIZED SADDLE OIN ROBLE Fatemeh anjeh Ali BEIK his paper deals with applying a class
More informationAchieving depth resolution with gradient array survey data through transient electromagnetic inversion
Achieving depth resolution with gradient array survey data through transient electromagnetic inversion Downloaded /1/17 to 128.189.118.. Redistribution subject to SEG license or copyright; see Terms of
More informationLuca F. Pavarino. Dipartimento di Matematica Pavia, Italy. Abstract
Domain Decomposition Algorithms for First-Order System Least Squares Methods Luca F. Pavarino Dipartimento di Matematica Universita dipavia Via Abbiategrasso 209 27100 Pavia, Italy pavarino@dragon.ian.pv.cnr.it.
More informationVolume Integral Equations in Nonlinear 3D. Tampere University of Technology South Cass Avenue. Argonne, Illinois 60439, U.S.A.
Volume Integral Equations in Nonlinear 3D Magnetostatics Lauri Kettunen and Kimmo Forsman Tampere University of Technology Laboratory of Electricity and Magnetism, P.O. Box 692 FIN-33101 Tampere, Finland
More informationPreconditioning Techniques Analysis for CG Method
Preconditioning Techniques Analysis for CG Method Huaguang Song Department of Computer Science University of California, Davis hso@ucdavis.edu Abstract Matrix computation issue for solve linear system
More informationNumerical Methods I Non-Square and Sparse Linear Systems
Numerical Methods I Non-Square and Sparse Linear Systems Aleksandar Donev Courant Institute, NYU 1 donev@courant.nyu.edu 1 MATH-GA 2011.003 / CSCI-GA 2945.003, Fall 2014 September 25th, 2014 A. Donev (Courant
More informationNumerical problems in 3D magnetostatic FEM analysis
Numerical problems in 3D magnetostatic FEM analysis CAZACU DUMITRU Department of electronics, computers and electrical engineering University of Pitesti Str.Tirgu din Vale nr.1 Pitesti ROMANIA cazacu_dumitru@yahoo.com
More informationJ.TAUSCH AND J.WHITE 1. Capacitance Extraction of 3-D Conductor Systems in. Dielectric Media with high Permittivity Ratios
JTAUSCH AND JWHITE Capacitance Extraction of -D Conductor Systems in Dielectric Media with high Permittivity Ratios Johannes Tausch and Jacob White Abstract The recent development of fast algorithms for
More informationA simple iterative linear solver for the 3D incompressible Navier-Stokes equations discretized by the finite element method
A simple iterative linear solver for the 3D incompressible Navier-Stokes equations discretized by the finite element method Report 95-64 Guus Segal Kees Vuik Technische Universiteit Delft Delft University
More informationCrouzeix-Velte Decompositions and the Stokes Problem
Crouzeix-Velte Decompositions and te Stokes Problem PD Tesis Strauber Györgyi Eötvös Loránd University of Sciences, Insitute of Matematics, Matematical Doctoral Scool Director of te Doctoral Scool: Dr.
More informationAn Optimal Finite Element Mesh for Elastostatic Structural. Analysis Problems. P.K. Jimack. University of Leeds. Leeds LS2 9JT, UK.
An Optimal Finite Element Mesh for Elastostatic Structural Analysis Problems P.K. Jimack School of Computer Studies University of Leeds Leeds LS2 9JT, UK Abstract This paper investigates the adaptive solution
More informationA High-Order Nodal Finite Element Formulation for Microwave Engineering
A High-Order Nodal Finite Element Formulation for Microwave Engineering Rubén Otín CIMNE - International Center For Numerical Methods in Engineering Parque Tecnológico del Mediterráneo, Edificio C4 - Despacho
More informationPreconditioned all-at-once methods for large, sparse parameter estimation problems
INSTITUTE OF PHYSICS PUBLISHING Inverse Problems 17 (2001) 1847 1864 INVERSE PROBLEMS PII: S0266-5611(01)20474-2 Preconditioned all-at-once methods for large, sparse parameter estimation problems E Haber
More informationA DELTA-REGULARIZATION FINITE ELEMENT METHOD FOR A DOUBLE CURL PROBLEM WITH DIVERGENCE-FREE CONSTRAINT
A DELTA-REGULARIZATION FINITE ELEMENT METHOD FOR A DOUBLE CURL PROBLEM WITH DIVERGENCE-FREE CONSTRAINT HUOYUAN DUAN, SHA LI, ROGER C. E. TAN, AND WEIYING ZHENG Abstract. To deal with the divergence-free
More informationOn solving linear systems arising from Shishkin mesh discretizations
On solving linear systems arising from Shishkin mesh discretizations Petr Tichý Faculty of Mathematics and Physics, Charles University joint work with Carlos Echeverría, Jörg Liesen, and Daniel Szyld October
More informationON AUGMENTED LAGRANGIAN METHODS FOR SADDLE-POINT LINEAR SYSTEMS WITH SINGULAR OR SEMIDEFINITE (1,1) BLOCKS * 1. Introduction
Journal of Computational Mathematics Vol.xx, No.x, 200x, 1 9. http://www.global-sci.org/jcm doi:10.4208/jcm.1401-cr7 ON AUGMENED LAGRANGIAN MEHODS FOR SADDLE-POIN LINEAR SYSEMS WIH SINGULAR OR SEMIDEFINIE
More informationwhere K e 2 h RN h e N h e is a symmetric positive denite (SPD) sparse system matrix, u h 2 R N h e the solution vector, and fh 2 R N h e the load vec
Algebraic Multigrid for Edge Elements Stefan Reitzinger y Joachim Schoberl z 15th June 2000 Abstract This paper presents an algebraic multigrid method for the ecient solution of the linear system arising
More informationLeast-Squares Finite Element Methods
Pavel В. Bochev Max D. Gunzburger Least-Squares Finite Element Methods Spri ringer Contents Part I Survey of Variational Principles and Associated Finite Element Methods 1 Classical Variational Methods
More informationSolving Updated Systems of Linear Equations in Parallel
Solving Updated Systems of Linear Equations in Parallel P. Blaznik a and J. Tasic b a Jozef Stefan Institute, Computer Systems Department Jamova 9, 1111 Ljubljana, Slovenia Email: polona.blaznik@ijs.si
More informationDivergence-conforming multigrid methods for incompressible flow problems
Divergence-conforming multigrid methods for incompressible flow problems Guido Kanschat IWR, Universität Heidelberg Prague-Heidelberg-Workshop April 28th, 2015 G. Kanschat (IWR, Uni HD) Hdiv-DG Práha,
More informationChapter 1 Mathematical Foundations
Computational Electromagnetics; Chapter 1 1 Chapter 1 Mathematical Foundations 1.1 Maxwell s Equations Electromagnetic phenomena can be described by the electric field E, the electric induction D, the
More informationGauge finite element method for incompressible flows
INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN FLUIDS Int. J. Numer. Meth. Fluids 2000; 34: 701 710 Gauge finite element method for incompressible flows Weinan E a, *,1 and Jian-Guo Liu b,2 a Courant Institute
More information2 IVAN YOTOV decomposition solvers and preconditioners are described in Section 3. Computational results are given in Section Multibloc formulat
INTERFACE SOLVERS AND PRECONDITIONERS OF DOMAIN DECOMPOSITION TYPE FOR MULTIPHASE FLOW IN MULTIBLOCK POROUS MEDIA IVAN YOTOV Abstract. A multibloc mortar approach to modeling multiphase ow in porous media
More information