Professional Documents
Culture Documents
Ebook Numerical Solutions of Three Classes of Nonlinear Parabolic Integro Differential Equations PDF Full Chapter PDF
Ebook Numerical Solutions of Three Classes of Nonlinear Parabolic Integro Differential Equations PDF Full Chapter PDF
of Nonlinear Parabolic
Integro-Differential Equations - eBook
PDF
Visit to download the full and correct content document:
https://ebooksecure.com/download/numerical-solutions-of-three-classes-of-nonlinear-
parabolic-integro-differential-equations-ebook-pdf/
Numerical Solutions of Three Classes of Nonlinear
Parabolic Integro-Differential Equations
Numerical Solutions of Three Classes
of Nonlinear Parabolic Integro-Differential
Equations
Temur Jangveladze
Ilia Vekua Institute of Applied Mathematics of
Ivane Javakhishvili Tbilisi State University
Tbilisi, Georgia
&
Georgian Technical University
Tbilisi, Georgia
Zurab Kiguradze
Ilia Vekua Institute of Applied Mathematics of
Ivane Javakhishvili Tbilisi State University
Tbilisi, Georgia
Beny Neta
Naval Postgraduate School
Department of Applied Mathematics
Monterey, CA, U.S.A.
Academic Press is an imprint of Elsevier
32 Jamestown Road, London NW1 7BY, UK
225 Wyman Street, Waltham, MA 02451, USA
525 B Street, Suite 1800, San Diego, CA 92101-4495, USA
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK
This book and the individual contributions contained in it are protected under copyright
by the Publisher (other than as may be noted herein).
Notices
Knowledge and best practice in this field are constantly changing. As new research and
experience broaden our understanding, changes in research methods, professional
practices, or medical treatment may become necessary.
Practitioners and researchers must always rely on their own experience and knowledge
in evaluating and using any information, methods, compounds, or experiments
described herein. In using such information or methods they should be mindful of their
own safety and the safety of others, including parties for whom they have a professional
responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or
editors, assume any liability for any injury and/or damage to persons or property as a
matter of products liability, negligence or otherwise, or from any use or operation of
any methods, products, instructions, or ideas contained in the material herein.
ISBN: 978-0-12-804628-9
Temur Jangveladze,
Zurab Kiguradze,
Beny Neta
ACKNOWLEDGMENTS
Introduction
Abstract
The description of various kinds of integro-di erential equa-
tions and a brief history of their origin and applications are
given. The importance of investigations of integro-di erential
models is pointed out as well. Classi cation of integro-di eren-
tial equation is given. The main attention is paid on parabolic
type integro-di erential models. In particular, three types of
integro-di erential equations are considered. Two of them are
based on Maxwell’s equations describing electromagnetic eld
penetration into a substance. The third one is obtained by
simulation of heat ow. At the end of the chapter, as at the
end of each chapter, the comments and bibliographical notes is
given, which consists of description of references concerning to
the topic considered.
t
@ 2 U (x; t) @ 2 U (x; t) @ 2 U (x; )
Z
= + (t; )d ; (1.1)
@t2 @x2 0 @x2
" #
2
@ 2 U (x; t) @ 2 U (x; t)
Z
2 @U (x; t)
+ dx = 0; (1.2)
@t2 0 @x @x2
CHAPTER 1. INTRODUCTION 7
"Z #
2
@ 2 U (x; t) @U (x; t) @ 2 U (x; t)
a dx =0 (1.3)
@t2 0 @x @x2
and
@ 2 U (x; t) 1
a A 2 U (t) AU (t) = 0; (1.4)
@t2
where a(S) a0 = const > 0 is a known function of its argu-
ment and A is a self-adjoint positive operator, i.e., A = A > 0.
The norm used in (1.4) is the one de ned on the range of the
operator A.
In investigating (1.3) and (1.4) type models it is su cient
to mention the following publications: [9], [24], [36], [37], [55],
[61], [65], [130], [309], [327], [328], [349], [355], [356], [357], [373],
[380], [383], [390], [394], [395] though this list is not complete.
Let us note that equations (1.3) and (1.4) are also called Kirch-
ho equations. They, along with some similar equations, de-
scribe important physical processes, among which are linear and
nonlinear dynamics of di erent dimensional bodies (see, for ex-
ample, [36], [355], [373] and [471]).
In other questions connected with hereditary phenomena,
one nds the integro-di erential equations of a parabolic type,
which were investigated by Evans [160]. These equations look
like
Z t 2
@U (x; t) @ 2 U (x; t) @ U (x; )
2
+ A(t; )d = 0; (1.5)
@t @x t0 @x2
@W (x; t)
+r [a(S(x; t))r W (x; t)] = 0;
@t (1.6)
r W (x; t) = 0;
where Z t
S(x; t) = jr W (x; )j2 d : (1.7)
0
In system (1.6), r W and r W are the usual vector oper-
ators with respect to the variables x = (x1 ; x2 ; x3 ): Even one-
dimensional scalar version of this model is very complicated and
its investigation has been possible yet only for special cases. The
one-dimensional scalar case of the model (1.6), (1.7) has the fol-
lowing form
" Z ! #
t 2
@U (x; t) @ @U (x; ) @U (x; t)
= a d ; (1.8)
@t @x 0 @x @x
began in the works [138], [187], and [188]. Since then many
scienti c publications were devoted to the investigation of exis-
tence and uniqueness of their solutions under various types of
initial and boundary conditions. In this respect, especially sig-
ni cant are the works [49], [50], [141], [146], [147], [219], [220],
[238], [254], [256], [302], [303], [304], [305], [306], [307], [322],
[333], [334], and reference therein. Authors of this book have
also made contribution in this direction, for example, see [137],
[138], [139], [140], [141], [145], [146], [147], [186], [187], [188],
[219], [220], [223], [238], [247], [248], [253], [254], [256], [261],
and reference therein.
Making certain physical assumptions in mathematical de-
scription of the above-mentioned process of penetration of elec-
tromagnetic eld into a substance, Laptev [306] has constructed
a new integro-di erential model, which represents a general-
ization of the system introduced in [187] and [188]. Founded
on Maxwell’s system the following parabolic integro-di erential
model is obtained
Z tZ
@W (x; t)
=a jr W (x; )j2 dxd W (x; t): (1.9)
@t 0
Z t
@U (x; t)
=r a jrU (x; )j2 d rU (x; t) (1.10)
@t 0
and
Z tZ
@U (x; t)
=a jrU (x; )j2 dxd U (x; t); (1.11)
@t 0
respectively.
In equations (1.10) and (1.11) we have x = (x1 ; : : : ; xn ) and
the vector operator rU is given by
@U (x; t) @U (x; t)
rU = gradU = ;:::; = (D1 U; : : : ; Dn U ) :
@x1 @xn
Some generalizations of the models (1.6)-(1.11) are given in
the works [137], [139], [145], [219], [305], and [306]. One kind of
these models has the forms:
n
@U (x; t) X
Di a(S(x; t)) jrU (x; t)jq 2
Di U (x; t)
@t i=1 (1.12)
= f (x; t)
and
n
@U (x; t) X
a(S(t)) Di jrU (x; t)jq 2
Di U (x; t)
@t i=1 (1.13)
= f (x; t);
CHAPTER 1. INTRODUCTION 11
where Z t
S(x; t) = jrU (x; )jq d
0
and Z tZ
S(t) = jrU (x; )jq d :
0
The models of type (1.6)-(1.13) are complex and have been
intensively studied by many authors. The existence and unique-
ness of global solutions of initial-boundary value problems for
equations and systems of type (1.6) - (1.13) were studied in
[49], [50], [137], [138], [139], [140], [141], [145], [186], [187],[188],
[219], [223], [238], [247], [248], [253], [261], [302], [303], [304],
[305], [306], [307], [322], [333], [334], and in a number of other
works as well.
The existence theorems that are proved in [137], [138], [139],
[140], [141], [145], [187], [188], and [219] are based on a-priori
estimates, modi ed Galerkin’s method and compactness argu-
ments as in [327], [328], [461], and [462] for nonlinear elliptic
and parabolic equations.
For equation (1.8) with nonhomogeneous right-hand side and
a(S) = 1 + S, or for equation (1.13) in the one-dimensional
case with q = 2, such theorem for rst initial-boundary value
problem is proved in section 3.6.
The asymptotic behavior as t ! 1 of the solutions of such
models has been the object of intensive research in recent years,
see [28], [29], [31], [32], [33], [34], [35], [139], [141], [145], [146],
[147], [217], [218], [219], [220], [223], [227], [228], [229], [231],
[232], [233], [234], [235], [236], [237], [238], [239], [240], [241],
[242], [243], [245], [246], [247], [248], [251], [252], [253], [254],
[256], [257], [261], [276], [277], [278], [280], [282], [283], and ref-
erence therein.
Another model considered in this book and studied by one
of the authors of this monograph [376] is
12 CHAPTER 1. INTRODUCTION
Z t
@U (x; t) @ @U (x; )
= a(t ) d + f (x; t): (1.14)
@t 0 @x @x
[452], [453], [454], [470], [475], [476], [477], [485], [486], [487],
[489], [491], and references therein.
As we have already pointed out, by using Maxwell’s system
[300] for mathematical simulation of the process of electromag-
netic eld penetrating into a substance, new class of integro-
di erential models (1.6), (1.7), (1.8), and (1.10) arises (see [187]
and [188]).
Integro-di erential equations arise in many other practical
processes as well, see, for example, [11], [21], [22], [36], [37],
[55], [59], [74], [84], [119], [127], [128], [164], [185], [187], [188],
[295], [298], [306], [310], [311], [312], [355], [356], [357], [363],
[373], [380], [384], [405], [425], [459], [463], [465], [466], [467],
[468], [469], [471], [484], and in a number of other works as well.
The motivation for studying integro-di erential problems
comes from the many physical models in such elds as elec-
tromagnetic wave propagation, heat transfer, nuclear reactor
dynamics, and thermoelasticity. Besides the integro-di erential
equations arise in many spheres of human activity as well. For
example, the second order fully nonlinear integro-di erential
equations are derived from the pricing problem of nancial deriva-
tives and optimal portfolio selection problem in a market [59].
In [84] nonlinear integro-di erential equations that arise from
stochastic control problems with purely jump Levy processes
are considered.
Many problems of modern science and engineering can be
described by partial integro-di erential equations. Since quite
a lot of these problems are time-dependent, most of them are
evolution equations and especially nonlinear evolution parabolic
equations, see [93], [177], [179], [180], [191], [216], [310], [311],
[312], [322], [332], [345], [346], [358], [451], and references therein).
Many scienti c works are devoted to investigation and nu-
merical solution of parabolic integro-di erential models, see [6],
[18], [45], [53], [62], [63], [69], [74], [83], [86], [87], [93], [98], [103],
16 CHAPTER 1. INTRODUCTION
[108], [115], [118], [128], [137], [139], [145], [155], [156], [163],
[175], [177], [181], [182], [191], [198], [247], [253], [263], [265],
[266], [267], [276], [279], [281], [306], [308], [310], [311], [312],
[313], [319], [320], [322], [323], [327], [353], [358], [370], [376],
[378], [389], [393], [427], [430], [432], [433], [447], [448], [449],
[450], [451], [456], [479], [483], [488], [492], [497], and references
therein.
Study of the models of type (1.6), (1.7), and (1.8) has begun
in the works [138], [187] and [188]. In these works, in particu-
lar, the theorems of existence of solution of the initial-boundary
value problem (with rst (Dirichlet) boundary conditions) for
scalar equation with one-dimensional space variable are proved.
Investigations of higher space dimensions for model (1.10) car-
ried out initially in [137] and [140].
In [306] some generalization of the system of type (1.6), (1.7)
was proposed. In particular, assuming that the temperature
of the considered body depends on the time variable, but in-
dependent of the space coordinates, then the same process of
penetration of a magnetic eld into the material is simulated
by the averaged (as the author of [306] has named it) integro-
di erential models (1.9) and (1.11).
Study of the models of type (1.9) and (1.11) has started in
the works [217] and [219].
One must note that some works were devoted to the study
of modeling of physical process of electromagnetic eld pene-
tration in the case of cylindrical conductors. In this case, the
above-mentioned integro-di erential model (1.6), (1.7), written
in cylindrical coordinates, was given in [148]. The work [333]
is devoted to the investigation of periodic problem for one-
dimensional (1.8) type model in cylindrical coordinates.
Interest in the above-mentioned integro-di erential (1.6) -
(1.11) models is increasing. Some generalizations of (1.10) and
(1.11) models, which have the forms (1.12) and (1.13) corre-
1.1. COMMNETS AND BIBLIOGRAPHICAL NOTES 17
Mathematical Modeling
Abstract
The chapter consists of six sections. The chapter concerns
mathematical modeling of the investigated in the monograph
equations. Some mathematical features of these models are
studies as well. In the rst section general statement of di usion
process is given. Sections two and three are dedicated to a re-
duction of Maxwell’s equations to the integro-di erential mod-
els. Consequently, two types of integro-di erential equations
are obtained which are called Model I and Model II accordingly.
Both models in di erent physical assumptions describe process
of penetration of an electromagnetic eld into a substance. In
the third section Model III is considered, which represents a
special model for one-dimensional heat ow in materials with
memory. This model arises in the theory of one-dimensional
viscoelasticity as well. Next two sections are devoted to some
mathematical features of all three models above. The existence
and uniqueness properties of the solutions as well as asymp-
totic behavior of solutions of the appropriate initial-boundary
value problems are presented. This chapter closes with some
concluding remarks and bibliographical overview of the three
investigated models.
1 @( H)
= r E; (2.1)
c @t
r ( H) = 0; (2.2)
4
E = r H; (2.3)
c
where E = (E1 ; E2 ; E3 ) and H = (H1 ; H2 ; H3 ) are vectors of
electrical and magnetic elds, respectively, is the magnetic
permeability, is the conductivity of substance, and c is the
speed of light in vacuum.
In (2.3), following an assumption of quasi-stationarity, the
term proportional to @E=@t is omitted and Ohm’s law is used,
relating the vector E to a vector of density of a current J by
the following relation
J = E: (2.4)
2.1. ELECTROMAGNETIC DIFFUSION PROCESS 21
@
cv = EJ = E 2 ; (2.5)
@t
where cv is the speci c heat capacity of the environment. Thus
the factor of heat capacity can also depend on temperature (as
a rule, in a power-like form).
Equations (2.1), (2.3), (2.5) form the closed system for def-
inition of an electromagnetic eld and temperature under an
appropriate initial and boundary conditions.
The questions of existence and uniqueness of the solutions of
linear di erential problems ( = (x)), in general enough state-
ment, are considered in [136], [297], and in a number of other
works. Thus, in [297] the movement of environment is supposed
within the framework of magnetic-hydrodynamical approach. In
these works the transition from classical statement to the gener-
alized one is made. The requirement that the functions satisfy
equation (2.2) and boundary conditions is replaced with the re-
22 CHAPTER 2. MATHEMATICAL MODELING
@
0 cv = J 2: (2.7)
@t
Here is density of the substance, 0 is the density at time t = 0,
and cv (as was already mentioned in 2.1.1) is its speci c heat
capacity. Generally they also depend on temperature. Equation
(2.7) does not take into account transfer of heat due to heat
conductivity and radiation. Number of other physical e ects
are not considered. However, from the mathematical point of
view system (2.6), (2.7) is complex enough.
Now we shall begin the reduction of system (2.6), (2.7) to a
system of nonlinear integro-di erential equations.
Let us rewrite equation (2.7) in the following form
cv @
0( ) = J 2:
( ) @t
we have
@S
= J 2:
@t
Let us assume that the process begins at the moment of time
t = 0 in which the temperature 0 of the substance is constant.
Integrating this equation on a segment [0; t]; we shall nd
Z t
S( (x; t)) S( 0 ) = J 2d :
0
24 CHAPTER 2. MATHEMATICAL MODELING
t
c2
Z
@H c 2
+ r r H d r H = 0;
@t 4 0 4
r H = 0:
Let us introduce the notations
c2 c
a(S) = ( (S)); W = H
4 4
and rewrite the system in the following form:
Z t
@W
+r a jr W j2 d r W = 0;
@t 0 (2.8)
r W = 0:
2.1. ELECTROMAGNETIC DIFFUSION PROCESS 25
@U @U
r W = ; ;0
@x2 @x1
and (2.8) takes the form
Z t
@U
=r a jrU j2 d rU : (2.9)
@t 0
Let us note again that equations such as (2.8), (2.9) for the
rst time have arisen in [187] and [188]. In these works, as well
as in [137], [138], [139], [140], [141], [145], [186], [219], [223],
[238], [247], [253], and [261] together with other questions, the
uniqueness of the solutions of the initial-boundary value prob-
lems for equations (2.8) and (2.9) is given, under a rather general
assumptions on the function a = a(S).
Assume that 0 and cv are constants. Let us give examples
of function ( ), inducing a(S), see, e.g., [145], [304], and [306].
If ( ) = , > 1, then
a(S) = C1 (C0 S) 1;
where C0 and C1 are some positive constants. Thus, the power
growth of the resistance ( ) gives a coe cient determined only
on a nite interval. Let us notice that the physical substances
do not have such property.
If ( ) = , then
a(S) = C1 eC0 S ;
with positive constants C0 and C1 , i.e., the linear growth of the
function ( ) gives exponential function a(S). The linear growth
26 CHAPTER 2. MATHEMATICAL MODELING
a(S) = C1 (C0 + S) 1 ;
with positive constants C0 and C1 . Thus, under-linear growth
of the function ( ) gives sedate growth of the coe cient a(S).
If ( ) = , < 0, then
da c2 d d c2 d =d c2
= = = ( ) 0 ( ):
dS 4 d dS 4 dS=d 4 cv
Z tZ
@W 1
+r a jr W j2 dxd r W = 0;
@t j j 0
28 CHAPTER 2. MATHEMATICAL MODELING
r W = 0:
Here j j is the volume of the substance. Due to averaging the
coe cient of this equation depends only on the variable t, and
consequently the equation can be rewritten in the form
Z tZ
@W
=a jr W j2 dxd W: (2.15)
@t 0
r r W = W + r(r W ):
For a at eld W = (0; 0; U ), where U = U (x1 ; x2 ; t) is a
function of two spatial variables, system (2.15) becomes
Z tZ
@U
=a jrU j2 dxd U: (2.16)
@t 0
@U @ 2U
=k 2;
@t @x
where U (x; t) is the absolute temperature and k > 0 is a con-
stant called thermal di usivity. This parabolic equation allows
a thermal disturbance at any point in the medium to be in-
stantly felt at other points, see [196]. Gurtin and Pipkin [196]
2.3. MATHEMATICAL MODEL III 29
which is (1.14).
" Z ! #
t 2
@U @ @U @U
a d = f (x; t) (2.22)
@t @x 0 @x @x
and
2.4. SOME FEATURES OF MODELS I AND II 31
!
Z tZ 1 2
@U @U @ 2U
a dxd = f (x; t): (2.23)
@t 0 0 @x @x2
Z Z
@U
+ A(U ) V dxdt = f V dxdt; (2.26)
QT @t QT
@ 2V @ @V
2
= a ;
@t @x @x
which generally does not possess global solutions.
The following existence and uniqueness statement takes place
for problem (2.22), (2.24), (2.25) (see [138]).
2.4. SOME FEATURES OF MODELS I AND II 33
Theorem 2.1 If
@ 2U
2 L2 (QT );
@x2
p @ 2U
T t 2 L2 (QT ):
@t@x
If
2a0
;
C
where C here and below is a positive constant independent of t,
the following estimate takes place
t=2
jjU ( ; t)jj e jjU0 jj :
2.4. SOME FEATURES OF MODELS I AND II 35
Now, using well-known fact (see, for example, [25], [270], and
[296]) it is clear that
jjU ( ; t)jjW 1 ( ) ! 0; t ! 1:
2
@U @ @U
= a(S) ; (x; t) 2 Q; (2.27)
@t @x @x
@U @U a0
+ C exp t :
@x @t 2
(2.28), we obtain
2 Z 1 2
@U 1 @ @U
+ a(S) dx = 0: (2.31)
@t 2 0 @t @x
Let us di erentiate relation (2.27) with respect to t:
@ 2U @ @a(S) @U @ 2U
+ a(S) = 0: (2.32)
@t2 @x @t @x @t @x
We multiply (2.32) by U and integrate the resulting relation
over the interval [0; 1]:
Z 1 2 Z 1 2
@ U @a(S) @U
2
U dx + dx
0 @t 0 @t @x
(2.33)
Z 1 2
1 @ @U
+ a(S) dx = 0:
2 0 @t @x
It follows from (2.31) and (2.33) that
Z 1 2 2 Z 1 2
@ U @U @ @U
2
U dx + + a(S) dx
0 @t @t 0 @t @x
Z 1 2
@a(S) @U
+ dx = 0;
0 @t @x
which, in view of the relation
Z 1 2 2
@ U @U 1 d2
2
U dx + = kU k2 ;
0 @t @t 2 dt2
acquires the form
1 2
1 d2
Z
d @U
2
kU k2 + a(S) dx = 0: (2.34)
2 dt dt 0 @x
2.4. SOME FEATURES OF MODELS I AND II 39
The brass liner for the lower seat was passed through the upper
seat by being made thinner than the upper liner. Those for the valve
were made ¹⁄₈ inch too long, and guttered in the lower edge. They
were then driven down by a set and sledge on an anvil. By going
around them three times the lower edges were spread out to fill the
chamfer, and the flanges brought down to their seats. Those for the
lower valve were put in in halves.
William R. Jones
CHAPTER XXIII