%PDF-1.4
%
1 0 obj
<>stream
Iterative methodsPreconditioningIncomplete LU factorizationsGraph partitioningMatrix reorderings[EN] Computational electromagnetics based on the solution of the integral form of Maxwell s
equations with boundary element methods require the solution of large and dense linear
systems. For large-scale problems the solution is obtained by using iterative Krylov-type
methods provided that a fast method for performing matrix vector products is available.
In addition, for ill-conditioned problems some kind of preconditioning technique must
be applied to the linear system in order to accelerate the convergence of the iterative
method and improve its performance. For many applications it has been reported that
incomplete factorizations often suffer from numerical instability due to the indefiniteness
of the coefficient matrix. In this context, approximate inverse preconditioners based on
Frobenius-norm minimization have emerged as a robust and highly parallel alternative.
In this work we propose a two-level ILU preconditioner for the preconditioned GMRES
method. The computation and application of the preconditioner is based on graph
partitioning techniques. Numerical experiments are presented for different problems and
show that with this technique it is possible to obtain robust ILU preconditioners that
perform competitively compared with Frobenius-norm minimization preconditioners.Reconocimiento - Sin obra derivada - No comercial (by-nd-nc)Embargadohttp://hdl.handle.net/10251/105506Elsevierapplication/pdf José Marín Mateos-Aparicio José Mas MaríJournal of Computational and Applied MathematicsInglés
endstream
endobj
2 0 obj
<>/ProcSet[/PDF/Text/ImageB/ImageC/ImageI]/XObject<>/ColorSpace<>/Font<>>>/MediaBox[0.0 0.0 595.32 841.92]/StructParents 0/Rotate 0>>
endobj
20 0 obj
<>
endobj
23 0 obj
<>
endobj
29 0 obj
<>
endobj
42 0 obj
<>
endobj
54 0 obj
<>
endobj
58 0 obj
<>
endobj
63 0 obj
<>
endobj
69 0 obj
<>
endobj
76 0 obj
<>
endobj
81 0 obj
<>
endobj
89 0 obj
<>
endobj
95 0 obj
<>
endobj
105 0 obj
<>
endobj
114 0 obj
<>
endobj
121 0 obj
<>
endobj
126 0 obj
<>
endobj
129 0 obj
<>
endobj
138 0 obj
<>
endobj
141 0 obj
<>
endobj
150 0 obj
<>
endobj
158 0 obj
<>
endobj
161 0 obj
<>
endobj
164 0 obj
<>
endobj
167 0 obj
<>
endobj
170 0 obj
<>
endobj
173 0 obj
<>
endobj
176 0 obj
<>
endobj
179 0 obj
<>
endobj
4 0 obj
<>
endobj
3 0 obj
<>
endobj
16 0 obj
[/ICCBased 183 0 R]
endobj
183 0 obj
<>stream
HyTSwoɞc
[5laQIBHADED2mtFOE.c}088GNg9w߽ '0 ֠Jb
2y.-;!KZ ^i"L0-
@8(r;q7Ly&Qq4j|9
V)gB0iW8#8wթ8_٥ʨQQj@&A)/g>'K t;\
ӥ$պFZUn(4T%)뫔0C&Zi8bxEB;Pӓ̹Aom?W=
x- [ 0}y)7ta>jT7@tܛ`q2ʀ&6ZLĄ?_yxg)˔zçLU*uSkSeO4?c. R
߁-25 S>ӣVd`rn~Y&+`;A4 A9 =-tl`;~p Gp| [`L`< "AYA+Cb(R, *T2B-
ꇆnQt}MA0alSx k&^>0|>_',G!"F$H:R!zFQd?r9\A&GrQhE]a4zBgE#H *B=0HIpp0MxJ$D1D, VĭKĻYdE"EI2EBGt4MzNr!YK ?%_(0J:EAiQ(()ӔWT6U@P+!~mDeԴ!hӦh/']B/ҏӿ?a0nhF!X8܌kc&5S6lIa2cKMA!E#ƒdV(kel
}}Cq9
N')].uJr
wG xR^[oƜchg`>b$*~ :Eb~,m,-ݖ,Y¬*6X[ݱF=3뭷Y~dó tizf6~`{v.Ng#{}}jc1X6fm;'_9 r:8q:˜O:ϸ8uJqnv=MmR 4
n3ܣkGݯz=[==<=GTB(/S,]6*-W:#7*e^YDY}UjAyT`#D="b{ų+ʯ:!kJ4Gmt}uC%K7YVfFY.=b?SƕƩȺy
چk5%4m7lqlioZlG+Zzmzy]?uuw|"űNwW&e֥ﺱ*|j5kyݭǯg^ykEklD_p߶7Dmo꿻1ml{Mś
nLl<9O [$h՛BdҞ@iءG&vVǥ8nRĩ7u\ЭD- u`ֲK³8%yhYѹJº;.!
zpg_XQKFAǿ=ȼ:ɹ8ʷ6˶5̵5͵6ζ7ϸ9к<Ѿ?DINU\dlvۀ܊ݖޢ)߯6DScs
2F[p(@Xr4Pm8Ww)Km
endstream
endobj
5 0 obj
<>stream
x+ |
endstream
endobj
6 0 obj
<>stream
HUo0>ƓᩪT:R+,Җm$ºs d%ݽwyjYϞݤ߯$܀)JHpaT%\gOE-@l;2+]=J
Z@JDLe$elMxyYVD%0\
[Lymi]յSǶZ gfxRQ<",5)"ViMe7x00YaѰAl 9)@Dt<"Ǥ)W`A~o*۸&O B7 QÑ CՄp=gYX5S!L7&ӰInʤp}fQ.j^=o-2&5ϨIv@nD
4R\p.{]PDCjY3$B@n-<k"H:R3Pp+&`|;