PDE

INTRODUCTION

Let us recall that partial differential equation or PDE is an equation containing

the partial derivative with respect to several independent variable. solving PDEs
will be our main application of Fourier Transforms.
A PDE is said to be linear if the dependent variable and its derivative appear at
most to the first power and in no functions. we will only talk about linear PDEs.
Together with PDE, we usually specify some boundary condition, where the value
of the solution or its derivatives is given along the boundary region with some
initial condition. sometimes such conditions are mixed together and we will refer
to them simply as side conditions.
We will study three specific partial differential equations, each one representing a
general class of equations.
First, we will study the heat equation, which is an example of a parabolic PDE.
Next, we will study wave equation,which is an example of a hyperbolic PDE.
Finally, we will study the laplace equation, which is an example of an elliptic PDE.
Each of our examples will illustrate the behavior of its typical for the whole class.
Generalized Separation variable: The heat (diffusion) equation is an linear partial
differential equation with constant co.efficient. In thermal diffusivity (diffusion
co.efficient) depends on the heat transfer direction in in-homogeneous condition,
and it can depend co-ordinate condition and even on the temperature. In the last
case, the heat (diffusion) equation is non-linear various authors suggested a lot of
different relations to approximate the dependence of the transfer co.efficient on the
temperature or concentration, including linear, power-law and exponential.

1
Exact solutions of heat and mass transfer equations play an important role in
thermal and diffusion processes. Exact solutions of non linear equations make it
possible to look into the spatial localization of heat transfer, multiplicity and
absence of steady state under certain condition etc. Even those particular exact
solutions of PDEs which do not have a clear physical interpretation can be used as
test problems for checking the correctness and accuracy of various numerical,
asymptotic and approximate analytical methods. some of the problematic example
we can see here, with boundary condition of heat equation process in separation of
variable method and Fourier Transform based on it. And also see the algorithm
process to solve the partial differential equation with boundary value in Fourier
Transform.

2
Chapter 1

1.1 BASIC DEFINITION:

Definition 1.1.1. PDE:

A partial differential(PDE) is a mathematical equation that involves two or more
independent variables, an unknown function (dependent on those variable) and par-
tial derivatives of the unknown function with respect to the independent variables.

Example 1.1.2. Many physically important partial differential equations are second
order and linear.
uxx = ut (one dimensional heat equation)
uxx + uyy = 0 (Two dimensional laplace equation).

Definition 1.1.3. PDE used for?

Partial differential equations are used to mathematically formulate and thus the
solution of physical and other problems involving functions of several variable such
as the propagation of heat or sound, fluid flow, elasticity,etc., PDE is denoted by ∂.

Example 1.1.4. First and second PDE of ’f’ with respect to ’y’
∂f ∂2f
∂y
and ∂y 2

Definition 1.1.5. Another name for heat equations;

The heat equation is also known as diffusion equation and it describe a time

3
varying evolution of a function u(x, t) given its initial distribution u(x, 0) at t=0.
Heat equation is an example of elliptical partial differential equation.

Definition 1.1.6. What is meant by heat equation?

A partial differential equation the solution of which gives the distribution of tem-
perature in a region as a function of space and time when the temperature at the
boundaries, the initial distribution of temperature an the physical properties of the
medium are specified.

Definition 1.1.7. Heat equation on PDE.

The One dimensional heat equation is;
let u(x,t) denote the temperature at point x at time t.
∂u 2
∂t
= k ∂∂xu2 , where k > 0 is a constant (Thermal conductivity of the material)
(i.e.,)Change in heat at a specific point is proportional to the second derivative of
the heat along the wire.

Definition 1.1.8. THE SPACE V:

we need a further space.
V = {(y, v) ∈ H 2 (Λ) × H 2 (χ) : y|χ = v} −→ (V.a)
which is naturally embedded in H, and it is a hilbert space with respect to the scalar
product and norm inherited from H 2 (Λ) × H 2 (χ).

Definition 1.1.9. HEAT EQUATION:

we study the heat equation
wt − ∆w = 0
and the non homogeneous heat equation.
wt − ∆w = 0
be a initial and boundary conditions. Here t > 0 and x ∈ U ,where U ⊂ Rn is open.

4
 n
P
we denote ∆u = ∆x u = u xi xi .
i=n

FUNDAMENTAL SOLUTION OF HEAT EQUATION:


2
|x|
1 − 4t

e (x ∈ Rn , t > 0 )


(4πt)n/2
Φ(x, t) =
(x ∈ Rn , t > 0 )

0


is called the fundamental solution of the heat equation.

Definition 1.1.10. EXPONENTIAL SOLUTION OF HEAT EQUATION:

If w is
wt − ∆w = 0 = (iω + |y|2 )w = 0
given/ provided ω = i|y|2 . Hence
2t
u = eiyx−|y|

Definition 1.1.11. ? HEAT EQUATION(without heat source:)

∂ ∂ 2
∂t
u(x, t) = k. ∂x 2 u(x, t), x ∈ Ω, t > 0

? HEAT EQUATION(with heat source:)

∂ ∂ 2
∂t
u(x, t) = k. ∂x 2 u(x, t) + f (x, t), x ∈ Ω, t > 0

? HEAT EQUATION(with linear convection term:)

∂ ∂ 2
∂t
u(x, t) = k. ∂x 2 u(x, t) + λu(x, t), x ∈ Ω, t > 0

? HEAT EQUATION(with non linear convection term:)

∂ ∂ 2
∂t
u(x, t) = k. ∂x 2 u(x, t) + F (u(x, t), x, t), x ∈ Ω, t > 0

Example
∂ ∂ 2
∂t
u(x, t) = k. ∂x 2 u(x, t) + (µ − u(x, t))u(x, t), x ∈ Ω, t > 0

Definition 1.1.12. FICK’S LAW:

We assume that the rate of change of the density function due to these reasons
is f(t,x,p), which we usually call the reaction rate. Now we derive a differential
equation using the balance law. We choose any region O, then the total population
R
in O is p(t, x)dx and the rate of change of total population is.
0

5
 d
R
dt
p(t, x)dx
0
Net growth of the population inside the region O is
R
f (t, x, p(t, x))dx
0
and total out flux is
R
J(t, x).n(x)dx.
∂O

Definition 1.1.13. For the heat equation, we must also have some boundary con-
dition.
We assume that the ends of the wire are either exposed and touching some body
of constant heat, or the ends are insulated.
? If the ends of the wire are kept at temperature 0, then the condition are u(0, t)=
0 and u(L, t)= 0.
? If on the other hand the ends are also insulated, the conditions are ux (0, t) = 0
and ux (L, t) = 0.
? If ux is positive at some point x0 , then at a particular time, u is smaller to the
left of x0 and higher to the right of x0 . Heat is flowing from high heat to low heat,
(i.e.,) to the left.
On the other hand, if ux is negative then heat is again flowing from high heat to
low heat,(i.e.,) To the right.So,when ux is Zero, that is a point through which heat
is not flowing.
? In other words, ux (0, t) = 0 means no heat is flowing or out of the wire at the
point x= 0.
We have two conditions along the x-axis as there are two derivatives in the x-
direction. These side conditions are said to be homogeneous. [i.e., u or a derivatives
of u is set to be zero.]
? We also need an initial condition,
The temperature distribution at time t= 0.
(i.e.,)u(x, 0)= f(x) for some known function f(x)
this initial condition is not a homogeneous side conditions.

6
Definition 1.1.14. Separation of variables.
The heat equation is linear as u and its derivative do not appear to any pow-
ers or in any function.Thus the principle of superposition still applies for the heat
equations(without side conditions).
If u1 and u2 are solutions and c1 ,c2 are constants then u = c1 u1 + c2 u2 is also a
solution.

Definition 1.1.15. THEOREM STATEMENTS:

? MORREY’S INEQUALITY:
Assume n < p 6 ∞, then there exists a constant C, depending only on P and N
such that
kukC 0,γ (Rn ) 6 CkukW 1,p(Rn )
for all u ∈ C 1 (Rn ), where γ := 1 − n/p, n < p < ∞.
? POINCARE’S INEQUALITY:
Let U be a bounded, connected, open subset of Rn , with a C 1 boundary ∂u.
Assume 1 6 p 6 ∞. Then there exists a constant C, depending only on n, p and U,
such that
ku − (u)U kLp (U ) 6 CkDukLp (u)
for each function u ∈ W 1,p (U ).
? POINCARE’S INEQUALITY FOR A BALL:
Assume 1 6 p 6 ∞. Then there exists a constant C, depending only on n and p
such that
ku − (u)x,r kLp (B(x,r)) 6 CrkDukLp (B(x,r))
for each ball B(x, r) ⊂ Rn and each function u ∈ W 1,p (B 0 (x, r)).
? TRACE THEOREM:
Assume a domain U is bounded and the ∂u is C 1 . Then there exists a bounded
linear operator
T : W 1,p (U ) −→ Lp (∂u)

7
such that
i). T u = u|∂u if u ∈ W 1,p (U ) ∩ C(Ū )
ii). kT ukLp (∂u) 6 CkukW 1,p (U ) for each u ∈ W 1,p (U ), with C depending only on P
and U.
? CLOSED GRAPH:
If X and y are Banach space and T : X → Y is a linear operator, then T is
continuous if and only if its graph is closed in X × Y .(with the product topology).

Definition 1.1.16. Fourier transform:

The fourier transform is a generalization of the complex fourier series in the
limit as L −→ ∞. Replace the discrete An with the continuous F(k)dk while letting
n
L
−→ k.
then, change the sum to an integral and the equations become,

R∞
f(x) = F (k)e2πikx dk
−∞
R∞
F(k) = f (x)e−2πikx dx
−∞
here, F(k) = Fx [f (x)](k)
R∞
= f (x)e−2πikx dx
−∞
is called the forward (-i) fourier transform and
f(x) = Fk−1 [f (k)](x)
R∞
= f (k)e2πikx dk
−∞
is called the inverse(+i) Fourier transform.

The notation fˆ(k) and fˇ(x) are also used to denote the fourier transform and inverse
fourier transform respectively.

Definition 1.1.17. ? DEFINIYION OF CONVOLUTION SHORTLY:

R∞
f ∗g = f (T )g(X − T )
−∞

8
 =⇒ F[f ∗ g] = F[f ]F[g].
? PROPERTIES OF FOURIER TRANSFORM:
There are four fourier transform properties:
(1) Linearity property
(2) Transform of a transform
(3) Transform of a convolution
(4) Transform of product

Definition 1.1.18. ? FOURIER TRANSFORM AND INTEGRAL:

R∞ −iλx
f (x) = √12π e f (λ)dλ 0<x<∞
−∞
and
R∞ R∞
f (x) = √1
2π
e−iλx f (t)dtdλ.
−∞ −∞
? FOURIER SINE TRANSFORM AND INTEGRAL:
q R∞
f (x) = π2 f (x)sinλxdx 0<x<∞
0
and
q R∞ R∞
2
f (x) = π
f (t)sinλt sinλx dtdλ.
0 0
? FOURIER COSINE TRANSFORM AND INTEGRAL:
q R∞
f (x) = π2 f (x)cosλxdx 0<x<∞
0
and
q R∞ R∞
2
f (x) = π
f (t)cosλt cosλx dtdλ.
0 0

Definition 1.1.19. ALGORITHM:

To solve partial differential equation with boundary value:
1. Apply the suitable transform to give partial differential equation. for this check
the range of x.
i). If −∞ < x < ∞, then apply Fourier transform.
ii). If 0 < x < ∞, then check initial value conditions.
a). If value of u(0, t) is given, then apply Fourier sine transform.

9
  
∂u
b). If value of ∂x
is given, then apply Fourier cosine transform.
x=0
An ordinary differential equation will be formed after applying the transform.
2. Solve the differential equation using usual methods.
3. Apply Boundary value conditions to evaluate arbitrary constants.
4. Apply inverse transform to get the required expression for u(x, t).

Definition 1.1.20. What do you mean by fourier transform?

The fourier transform is a mathematical technique that transforms a function
of time, X(t) to a function of frequency, X(Λ).It is closely related to the fourir series.

Definition 1.1.21. uses(Application)

The fourier transform is an important image processing tool which is used to
decompose an image into its sine and cosine components.[The output of the trans-
formation represent the image in the fourier or frequency domain, while the input
image is the spatial domain equivalent]. Comparing with the signal process which is
often using 1-dimensional fourier transform, in imaging anlysis, 2 or higher dimen-
sional fourier transform are being used.

10
 Part I

The One Dimensional Heat

Equation.

11
1.2 Physical derivative:

In a metal rod with non uniform temperature, heat (thermal energy) is transformed
from regions of higher temperature temperature to the region of lower temperature.
It follows that three physical conditions/principles are used here,
1.Heat (or thermal) energy of a body with uniform properties:
Heat energy = Cmu,
where, m is the body mass,
u is the temperature,
C is the specific heat,
units[C] = L2 T −2 U −1
(basic units are M-mass, L length, T time, U temperature).
C is the energy required to raise a unit mass of the substance one unit in tempera-
ture .
2.Fourier’s law of heat transfer: rate of heat transfer proportional to negative tem-
perature gradient,
Rateof heattransf er
area
= −k0 ∂u
∂x
−→ (1)
Where k0 is the thermal conductivity, units [k0 ] = M LT −3 U −1 . In other words,
heat is transferred from areas of high temperature to low temperature.
3.Conversation of energy:
Consider a uniform rod of length l with non uniform temperature lying on the x
axis from x = 0 to x = l. By uniform rod, we mean the density ρ, specific heat C,
thermal conductivity k0 , cross sectional area A are ALL constant. Assume the sides
of the rod are insulate and only the ends may be exposed. Also assume there is no
heat source within the rod. Consider an arbitrary thin slice of the rod of width ∆x
between x and x + ∆x. The slice is so thin that the temperature throughout the
slice is u(x, t). Thus,
Heat energy of segment = C × ρA∆x × u
=CρA∆xu(x, t).
By conservation of energy,

12
change of heat energy of segment in time ∆t} = {heat in from left boundary -
{ heat out from right boundary.
from fourier law(1)
C ρA∆xu(x, t + ∆t) − CρA∆xu(x, t) = ∆tA(−k0 ∂u ) - ∆tA(−k0 ∂u
∂x x
)
∂x x+∆x

Rearranging yields, [recall ρ, c, A, k0 are constant]

 ∂u ∂u
u(x, t + ∆t) − u(x, t) −k0 ( ∂x )x+∆x − ( ∂x )x

= Cρ
∆t ∆x
Taking the limit ∆t , ∆x −→ 0 give the heat equation,
∂u 2
∂t
= k ∂∂xu2 −→ (2)
k0
where, k= Cρ
−→ (3) [ also denoted by ut = kuxx ]. is called the
thermal diffusivity units [k] = L2 /T . since the slice was chosen arbitrarily, heat
equation (2) applies throughout the rod.
Initial Condition and boundary condition:
we need clear information.
1.Initial Condition(IC):
The initial temperature distribution in the rod u(x, 0).
2.Boundary Condition(BC):
In this case, the temperature of the rod is affected by what happens at the ends,
x = 0,1. Here BC were consider three simple cases for the boundary at x= 0.
I. Temperature prescribed at a boundary. For t > 0, u(0, t) = u1 (t)
II. Insulated boundary, The heat flow can be prescribed at the boundaries −k0 ∂u
∂x
(0, t) =
φ1 (t)
III.Mixed Condition: An equations involving u(0, t), ∂u
∂x
(0, t) etc.,
NOTE:
The heat equation and corresponding IC and BCs are,
PDE: ut = k.uxx 0<x<l
IC: u(x, 0) = f (x) 0<x<l
BC: u(0, t) = u(L, t) = 0 t > 0.

13
1.3 Two Derivation of Heat Equations:

Heat equation: ut − kuxx = 0, k > 0 −→ (A)

this equation is also known as the diffusion equation.
DIFFUSION:
Consider a liquid in which a dye is begin diffused through the liquid. The dye
will move from higher concentration to lower concentration. Let u(x, t) be the con-
centration (mass per unit length) of the dye at position x in the pipe at time t. The
total mass of dye in the pipe from x0 to x1 at time t is given by,

Rx1
M (t) = u(, t)dx
x0
dM
Rx1
Therefore, dt
= ut (x, t)dx
x0
dM
By fick’s law, dt
= flow in - flow out = kux (x1 , t) − kux (x0 , t)

where k > 0 is a proportionality constant. i.e., The flow rate is proportional to the
Rx1
concentration gradient. Therefore , ut (x, t)dx = kux (x1 , t) − kux (x0 , t).
x0
Now differentiating with respect to x1 , we have
ut (x, t) = kuxx (t) or
ut = kuxx . This is known as diffusion equation.
HEAT FLOW:
We now given an alternate of equation (A) from the study of heat flow. Let D
be a region in Rn . Let u(x, t) be the temperature at point x, time t, and let H(t)
be the total amount of heat (in calories) contained in D. Let C be the specific heat
of the material and ρ its density (mass per unit volume). Then
R
H(t) = Cρu(x, t)dx
D
Therefore, the change in heat is given by
dH
R
dt
= Cρut (x, t)dx.
D
”Fourier’s law says that heat flows from hot to cold region at a rate k > 0 propor-
tional to the temperature gradient”. The only way heat will leave D is through the
boundary.

14
 dH
R
(i.e.,) dt
= k.∇unds
∂D
where ∂D is the boundary of D, n is the outward unit normal vector to ∂D and ds
in the surface measure over ∂D. we have
R R
Cρut (x, t)dx = k.∇unds
D ∂D
Recall that for a vector field F, the ”divergence theorem” says,
R R
F unds = ∇F dx.
∂D D
Therefore, we have
R R
Cρut (x, t)dx = ∇(k∇u)dx
D D
This leads us to the partial differential equation.
Cρut = ∇(k∇u)
If C, ρ and k are constants, then we have a heat equation,
n
k
P
ut = k∆u, where k = Cρ > 0 and ∆u = u xi xi
i=1

1.4 Heat Equations on an interval in R:

SEPARATION OF VARIABLE:
Consider the initial/ boundary value problem on a interval I in R,


ut = kuxx x ∈ I, t > 0






 u(x, 0) = φ(x) x∈I



satisfies certain BCs −→ (X1 )

u

In practice, the most common boundary condition are the following:

1. Dirichlet (I = (0, l)) : u(0, t) = 0 = u(l, t).
2. Neumann (I = (0, l)) : ux (0, t) = 0 = ux (l, t).
3. Robin (I = (0, l)) : ux (0, t) − a0 u(0, t) = 0 and ux (l, t) + a1 u(l, t).
4. periodic (I = (−l, l)) : u(−l, t) = u(l, t) and ux (−l, t) = ux (l, t).
We will give specific examples below where we consider some of these boundary
condition. However, we represent the technique of the separation of variable. This

15
technique involves looking for a solution of aparticular form. In general, we look for
a solution of the form,
u(x, t) = X(x)T (t)
for function X, T to be determined. Suppose we can find a solution of (X1 ) of this
form.

u(x,t) = X(x)T(t)

for function X, T to be determined. Suppose we can find a solution of (X1 ) of the

form.
Plugging a function u = X T into the heat equation.

0 00
XT − KX T = 0

Dividing this equation by K X T, we have

0 00
T X
KT
= X
= −λ

for some constant λ. Therefore, if there exists a solution u(x,t) = X(x)T(t) of the
heat equation, then T and X must satisfy the equations

0
T
KT
= −λ
00
X
X
= −λ for some constant λ.

In addition, in order for u to satisfy our boundary conditions,we need our function
X to satisfy our boundary condition.
i.e., we need to find functions X and scalars λ such that


 00
−X (x) = λX(x), x∈I



X
 satisfies our BCs −→ (X2 )

16
 This problem is known as an eigen value problem. In particular, a constant
λ which satisfies (X2 ) for some function X, not identically zero, is called an eigen
value of −∂x2 for the given boundary conditions. The function X is called an eigen
function with associated eigenvalue λ.

17
Chapter 2

2.1 Heat Equation With Dynamical Boundary Con-

dition of Reactive diffusive Type:

INTRODUCTION:
We study some evolution problem consisting in the standard heat equation posed
in a bounded domain, with supply some boundary condition. The problem is,


yt − ∆y = 0 in (0, ∞) × Λ






 yt = myϑ + n∆χ y = 0 on(0, ∞) × χ



on Λ −→ (H1 )

y(0, a) = y (a)
0

Here y = y(t, a), t > 0, a ∈ Λ, where Λ is a C ∞ regular bounded domain of

Rn (N > 2) and χ = ∂Λ. The first equation represents the law of standard dif-
fusion or heat conduction in Λ, and ∆ = ∆a denotes the laplacian operator with
respect to space variable. In the boundary conditions (H12 ).
y = Assume the trace of the function w defined for a ∈ Λ.
∆χ = Laplace Beltrami operator on χ.
ϑ = Outward normal to Λ and m ∈ R, n > 0 are given constants.
myϑ = interaction domain boundary.

18
n∆χ y = stands for boundary diffusion.
A first step in this study has been performed by the author’s paper, where we con-
sider the laplace equation instead of the heat equation as domain equation. These
modified problem admits a simple functional framework; because it helpus to under-
stand the dynamical boundary condition (H12 ). A new estimate is needed to study
the problem (H1 ), for the frame work of L2 (Λ) × L2 (χ).
we propose the problem (H1 ) work in the space


0 0
H = (y, ϑ) ∈ H (Λ) × H (χ) : y/χ = ϑ −→ (H2 )
0 0
where y/χ denotes trace of y on χ, with natural topology on H (Λ) × H (χ). we
denote for any r ∈ R, H r (Λ) and H r (χ) be the sobolev spaces of complex valued dis-
tributions respectively on Λ and χ. For our identification, H is an isomorphic coun-
0 0
terpart {y ∈ H (Λ) : y/χ ∈ H (χ)} through the simplification (y, y/χ) −→ y, w ∈ H
for functions defined on Λ.

2.2 THEOREM 1

For any y0 ∈ H problem (H1 ) has a unique solution y = y(y0 ) such that.
w ∈ C([0, ∞); H 1 (Λ)) ∩ C 1 ((0, ∞); H 1 (Λ)) ∩ C((0, ∞); H 3 (Λ))
w/χ ∈ C([0, ∞); H 1 (χ)) ∩ C 1 ((0, ∞); H 1 (χ)) ∩ C((0, ∞); H 3 (χ)) −→ (H3 ).
Moreover/ therefore,
k ∇y(t)k2L2 (Λ) +kdχ y/χ(t)k2L2 (χ) +ky/χ(t)k2L2 (χ) 6 e2λ0 t k ∇y0 (t)k2L2 (Λ) +kdχ y/χ(t)k2L2 (χ) +
ky/χ(t)k2L2 (χ) −→ (H4 )
for all t > 0, where λ0 > 0 is a constant depending on Λ. Finally, the family of
maps {y0 7−→ y(y0 )(t), t > 0} extends to an analytic quasi-contractive semi group
in H and consequently

y ∈ C ∞ ((0, ∞) × Λ̄) −→ (H5 )

19
The solutions are in principle complex valued but it is clear that for real valued
data the solution is likewise real valued. For more regular initial data satifying
usual compatibility condition. This is the content of the following regularity result.

2.3 THEOREM:2

If y0 ∈ H 2n+1 (Λ) and y0 |χ ∈ H 2n+1 (χ) for some n ∈ N and

(∆i y0 )|χ = m(∆i−1 y0 )ϑ + n∆χ ((∆i−1 y0 )|χ), for all i = 1,2,...,n. −→ (H6 )
then,
0
w ∈ C([0, ∞); H 2n+1 (Λ)) ∩ C 1 ([0, ∞); H 2n−1 (Λ)) ∩ ... ∩ C n ([0, ∞); H (Λ))
0
w/χ ∈ C([0, ∞); H 2n+1 (χ))∩C 1 ([0, ∞); H 2n−1 (χ))∩...∩C n ([0, ∞); H (χ)) −→ (H7 )
Finally, if y0 ∈ C ∞ (Λ̄) and (H6 ) hold for all i ∈ N , then
w ∈ C ∞ ([0, ∞) × Λ̄). −→ (H8 )
The proof of theorems 1 and 2 will be after some lemma conditions of the resolvent
problem with eigen value dependent boundary condition, that is



∆w + λw = H inΛ



−mwϑ − n∆χ w + λw = H
 onχ −→ (H9 )

where λ ∈ C and H ∈ H.
NOTATION:
We clearly denote by x, y ∈ C N , that is
N
P
xy = xi yi , when x = (x1 , ..., xN ), y = (y1 , ..., yN ) −→ (H10 ).
i=1
Therefore k.kp , 1 6 r 6 ∞, denotes the norm in Lp (Λ) and also the norm in
Lp (Λ; RN ) since no confusion is expected.

20
 Part II

Preliminaries and Functional

setting

21
2.4 Laplace Beltrami Operator:

For our convenience, we clearly determined χ i a Riemannian manifold endowed

with the natural metric inherited from RN , given in local coordinates by (Hij ) =
1, ..., N − 1. we denote by dσ the natural volume element on χ, given in local co-
√
ordinates by hdx1 , ..., dxN −1 , where h = det(hij ). we denote by ∇χ the Rie-
mannian gradient and by dχ be the total differential on χ. Then, it is clear that
(dχ w|dχ v) = (∇χ w, ∇χ v) for w, v ∈ C 1 (χ). So the use of vectors or forms in the
sequel is optional.
The Laplace Beltrami operator ∆χ can be at first defined on C ∞ (χ) by the formula.

R R
− (∆/χw)v̄dv = (dχ w|dχ v)dσ −→ (LB.1)
χ χ

for any w, v ∈ C ∞ (χ), and it is given in local coordinates by

−1
NP
−∆χ w = h−1/2 ∂
∂yi
∂u
(H 1/2 ∂y j
). −→ (LB.2)
i,j=1

where (H ij ) = (H ij )−1 where ∇χ can be considered as a bounded operator form

H r+2 (χ) to H r (χ) for any r ∈ R.

2.5 NOTE:

Note 2.5.1. DIRICHLET TO NEUMANN OPERATOR:

For any w ∈ H s (χ), r ∈ R, the non homogeneous Dirichlet problem



∆v = 0 inΛ



v = w
 onχ −→ (N.1)
For all w, v ∈ C ∞ (χ), integrating by part twice we have
R R R
Awv̄ = ∇(Dw)∇(Dv̄) = uAv̄ −→ (N.2)
χ Λ χ
which, by density,holds fo all w, v ∈ H 1 (χ)

22
Remark 2.5.2. since ∆χ 1 = 0 the operator is not injective,by (LB. 1) we have
R
− (−∆χ w + w)W̄ =k dχ w k2L2 (χ) + k w k2L2 (χ) −→ (R.1)
χ

Note 2.5.3. FUNCTIONAL SETTING

In the sequel we equip H 1 (χ) with the equivalent norm with the equivalent norm in
(R.1), so we denote
R R
(w, v)1H (χ) = wv̄ + (dχ w/dχ v). kw k2H 1 (χ) = (w, w)H 1 (χ) −→ (N.3)
χ χ
Moreover, we denote as usual
kW k2H 1 (Λ) = ku k22 +k∇u k22 −→ (N.4).

Lemma 2.5.4. Integral of fundamental solution:

R
For each time t > 0 Φ(x, t)dx = 1.
Rn

Proof. we calculate
|x|2
Φ(x, t)dx = (4πt)1 n/2 e− 4t dx
R R
Rn RRn | 2
1
= πn/2 e z| dz
Rn
1
R∞ 2
= πn/2 Πni=1 ezi dzi
−∞
= 1.

Note 2.5.5. THE SPACE H:

The space H is given in (2), which by the trace theorem is a closed subset of
H 1 (Λ) × H 1 (χ), hence hilbert space with respect to scalar product H 1 (Λ) × H 1 (χ).
R
By our clarity, notate w/χ, we shall write kwk2,χ , χ w and so on.

Note 2.5.6. Elliptic theory:

By a solution of equation(H9 ), we mean a function y ∈ ϑ such that (H9 ) holds true
in L2 (Λ), while H92 holds true in L2 (χ).
Space ϑ was just introduced in (z. a). Before stating the main result of this section
we introduce, for any r > 1,
Let H s = {(y, ϑ) ∈ H r (Λ) × H r (χ) : y|χ = ϑ} −→ (E.a)

23
where H r is a Hilbert space equipped with the norm inherited norm, H r (Λ) × H r (χ)
is a closed produce space. Moreover H and H 1 = H, H 2 = z are naturally embedded
naturally (like k.kH 2 and k.kz ).

Lemma 2.5.7. we set, for any w, z ∈ H,

R R R
(y, z)H = ∇y∇z̄ + (dχ y|dχ z) + yz̄, k y k2H = (w, w)H −→ (L1 )
Λ χ χ
Then is equality in H to be the standard norm inherited by H 1 (Λ) × H 1 (χ) .

Proof. we show that,

If in the standard norm of H 1 (Λ) × H 1 (χ).
” By poincaré-type inequality ” (in real valued case extension of complex valued
one).
R
=⇒ k y − y k6 C1 k∇k2 ∀w ∈ H 1 (Λ)
χ
where C1 = C1 (M, Λ) > 0, 2∗ is the sobolev critical exponent.
(i.e.,) 2∗ 6 2M/(M − 2) when M > 3, 1 6 2 ≤ ∞ when M = 2. Consequently, since
Λ is boundded and χ is compact.,
we get,
R R
kyk2 6 ky − yk2 + k yk2
χ χ R
6 C1 k∇yk2 + λM (Λ) |w|
χ
6 C2 (k∇yk2 + ky2.χ ).
where λM denotes the usual lebesgue measure in RN and C2 = C2 (M, χ) > 0. This
estimate completes the proof.

Lemma 2.5.8. If we set, for any y, ϑ ∈ z.

R R R
(y, ϑ)z = ∆y∆z̄ + ∆χ y∆χ z̄ + yz̄, kyϑ2 = (y, y)ϑ , −→ (L. 2)
Λ χ χ
then k.kϑ is equivalent in z to the standard norm inherited by H 2 (Λ) × H 2 (χ).

Proof. It follows by elliptic regularity.

For any y ∈ H 2 (Λ),
kyk2H (Λ) 6 C2 (k∇yk2 + ky/χkH 3/2 (χ) ).

24
and Consequently, since H 2 (χ) is continuously embedded in H 3/2 (χ),
for any y ∈ z, we get
kyk2H (Λ) 6 k∆yk2 + ky/χkH 2 (χ) ) −→ (L.2.1)
” By using functional setting definition”
=⇒ which completes the proof.

2.6 THEOREM 3:

Theorem 2.6.1. There is a positive constant (or) non negative constant µ0 , de-
pending on p, g, Λ, M such that µ ∈ C, Reµ > µ0 and t ∈ H equation (9) has a
unique solution y ∈ z, which also belongs to H 3 . Moreover, if t ∈ H 3 for some
r > 1, then y ∈ H r+2
Finally, there is C4 = C4 (p, g, Λ, s, µ) > 0 such that
kwkH r+2 6 C4 ktkH r ∀t ∈ H r −→ (4)
solve by elliptic order by variational method:
so, we introduce a sesquilinear form, which leads to weak solutions.
From equation (9) =⇒ multiply (atleast formally) the equation ∆y + µy = t by a
test function Ψ̄ ∈ C ∞ (Λ) and integrate over Λ to get.
R R R
− ∆y ψ̄ + µ y ψ̄ = tψ̄
Λ Λ Λ
Integrating by parts, when y ∈ H 2 (Λ)
R R R R
∇y∇ψ̄ − yν ψ̄ + µ y ψ̄ = tχ̄
Λ χ Λ Λ
Then, using the boundary equation (9), we get
∇y∇ψ̄ + g1 tψ̄ − µg y ψ̄ + gp ∆χ y ψ̄ + µ y ψ̄ = tψ̄
R R R R R R
∵ whenk 6= 0
Λ χ χ χ Λ Λ
Finally by equation (11) we arrive to
∇y∇ψ̄ − kp (dχ y/dχ ψ) − µg y ψ̄ + µ y ψ̄ = − k1 tψ̄ + tψ̄
R R R R R R
−→ (5)
Λ χ χ Λ χ Λ
Now, check L.H.S of sesquilinear form equation(15) is indefinite in the case of k > 0.
so we check the another side of positive definite sesquilinear form atleast Reµ large
enough.

25
so it exactly defines the content of the following two lemmas.

Lemma 2.6.2. 3(1)

Let t ∈ H. Then y ∈ z solves equation(9) if and only if
bµ (y, z) = (t, z)H forall z ∈ z −→ (6)
where the sesquilinear form aµ onz is defined by the formula.
R R R R
bµ (y, z) = ∆y∆z̄ + p ∆χ y∆χ z̄ + µ ∇y∇z̄ + (µ + p) (dχ y \ dχ z) −
ΛR χR RΛ R Rχ
p ∆χ yzν + g yν ∆χ z̄ − g yν zν − g yν z̄ + µ yz̄ −→ (7)
χ χ χ χ χ
By following this case, y ∈ H 3 (Λ) and y \ χ ∈ H 3 (χ).

Proof. It is divided into several steps.

CLAIM I:
If y ∈ z is a solution of (H6 ), then y ∈ H 3 (Λ) and y|χ ∈ H 3 (χ).
To recognize that our claim is true, by using elliptic regularity both Λ and χ as
follows. since y ∈ H 2 (Λ) we have yν ∈ H 1/2 (χ) by the trace theorem. so begin
t|χ ∈ H 1 (χ) and y|χ ∈ H 1/2 (χ) is follows that −∆χ y + y|χ ∈ H 1/2 (χ) from H92 , so
that we use isomorphism property of −∆χ + 1. so we conclude that y|χ ∈ H 5/2 (χ).
Consequently, using elliptic regularity for non homogeneous Dirichlet problems. we
obtain by H91 that y ∈ H 3 (Λ).
By using trace theorem , we get again yν ∈ H 3/2 (Λ) using H92 . so as before
y|χ ∈ H 3 (χ) completing the proof of our first claim.
CLAIM II:
If y ∈ z is a solution of (H9 ), then formula (H6 ) holds. By the first claim we have
∆y ∈ H 1 (Λ). Follow this H91 we get
(∆y)|χ = µy|χ − t|χ ∈ H 1 (χ) −→ (8)
Consequently, we get that ∆y ∈ H, so from H91 we have.
(−∆y, z)H + µ(y, z)H = (t, z) for all z ∈ H.
By using the definition of (., .)H given in (L1 ) in L.H.S terms of formula we write it,
R R R R R
∇(−∆y)∇z̄+ (dχ (−∆y)|dχ z)− ∆yz̄+µ (dχ y\dχ z)+µ yz̄ = (t, z)H −→ (9)
Λ χ χ Λ χ

26
Now, we using (8),
(∆y)|χ = gyν + p∆χ y|χ. in the form of (9)2 −→ (10)
Plugging (10) into (9), we get
R R R R R
∇(−∆y)∇z̄ − g (dχ yz \ dχ z) − p (dχ ∆χ y \ dχ z) − g yν z̄ − p ∆χ yz̄ +
R Λ R χR χ χ χ
µ ∇y∇z̄ +µ (dχ y \dχ z)+µ yz̄ = (t, z)H −→ (11) for all z ∈ H.
Λ χ χ
Now, we restrict to test functions z ∈ z, Next we use integrate by parts the first
integral in (11) and we use (11) in the first one we get.
R R R R R R
∆y∆z̄ − ∆y z¯µ − g (dχ yz \ dχ z) + p ∆χ y∆χ z̄ − g yν z̄ − p ∆χ yz̄ +
R Λ R χ Rχ χ χ χ
µ ∇y∇z̄ + µ (dχ y \ dχ z) + µ yz̄ = (t, z)H −→ (12)
Λ χ χ
Plugging (10) once again in the second integral in the L.H.S of (12) and (11) and
we finally we get (LB.1).
CLAIM III:
To complete the proof ,
Now, we now suppose that (16) holds for somey ∈ z. we have to prove that y solves
(9).
Integrating by parts in the third integral in (17) and in the first one in (L.1), then
we write as (16).
R R R R R R
∆y∆z̄ + p ∆χ y∆χ v̄ − µ y∆z̄ + µ yzν + (µ + p) (dχ y \ dχ z) − p ∆χ yzν +
ΛR χR R Λ R χ χ χ
g yν ∆χ z̄ − g yν zν − g yν z̄ + µ yz̄
χ χR Rχ R χ R
= - t∆z̄ + tzν + (dχ t \ dχ z) + tz̄ −→ (13)
Λ χ χ χ
for all v ∈ V
By plugging using (LB.1), we can write (13)as
R R R R R R R
∆y∆z̄ + p ∆χ y∆χ v̄ − µ y∆z̄ + µ yzν − µ y∆χ z̄ − p ∆χ yz̄ − p ∆χ y z¯ν +
ΛR χR Λ
R χ R χ χ χ
g yν ∆χ v̄ − g yν ∆χ z¯ν − g yν ∆χ z̄ + µ yz̄
χ χR R χ R Rχ
= - t∆z̄ + tzν + tzν + tz̄ −→ (14)
Λ χ χ χ
(i.e.,)we grouping the terms with respect to the test function.
R R
(∆y−µy+t)∆z̄+ (−p∆χ y−guν +µy−t)(−∆χ z + zν + z) = 0 −→ (15)
Λ χ
we, now to proceed this form of (15), By restrict to our test functions , atleast to
get (H92 ). we get that in distributional sense, which is not (H92 ).

27
so, it is more useful to prove (H92 ).
By restrict (26) to test function Dz, where z ∈ H 2 (χ) to get
R
(−p∆χ y − guν + µy − t)(−∆χ z + Cz + z) = 0 −→ (16)
χ
where C denotes the Dirichlet to neumann operator. now, we claim that (16).
R
(−p∆χ y − guν + µy − t)ψ̄ = 0 f orall ψ ∈ L2 (χ) −→ (17)
χ
from which we hold (H92 ) in L2 (χ).
To prove our claim, Given an arbitrary ψ ∈ L2 (χ),
-∆χ z + Cz + z = ψ −→ (18)
has a solution u ∈ H 2 (χ), which turns out to be unique.
hence our claim is nothing but a refinement, By our previous result of [lemma 1]
which says that given p > 0 and g̃ ∈ R there is ω̃ > 0 such that for ω > ω̃ the
problem.
−p̃∆χ z − g̃Cz + ωz = ψ −→ (19).
has a unique solution for any z ∈ H 2 (χ) for any ψ ∈ L2 (χ). In particular, our claim
is proved,
if we prove this more explicit form.
R R R R
− g̃Cz ψ̃ + p̃ (dχ z \ dχ ψ) + ω zψ = ψ φ̄ −→ (20) for all
χ χ χ χ
φ ∈ H 1 (χ)
then, we apply Lax-Milgram theorem to the sesquilinear form.
R R R
b(z, φ) = − g̃Cz φ̃ + p̃ (dχ z \ dχ φ) + ω z ψ̄, z, φ ∈ H 1 (χ)
χ χ χ
which is trivially Hermitian and continuous. It is also co-ercive for ω > 1 . since
g̃ < 0, we have
b(z, z) = −g̃k∇(Dz)k22 + p̃kdχ zk22,χ + ωkzk22,χ 6 min{p̃, ω}kzk2H 1 (χ) ,
whenever ω > 0 . Then from Lax-Milgram we get the existence of a solution z ∈
H 1 (χ) of (19). By the isomorphism property of −∆χ + 1 recalled and it follows that
z ∈ H 2 (χ) . Completing the proof of our claim.
R
(∆y − µy + t)∆z̄ = 0. forallz ∈ V −→ (21)
Λ
which implies (H9 ), since for any ψ ∈ L2 (Λ) there are z ∈ V such that ∆z = φ.
For example, by taking the unique solution z ∈ H 2 (Λ) ∩ H01 (Λ) with homogeneous

28
dirichlet boundary conditions.
The proof is then complete.

Lemma 2.6.3. 3(2):

There are positive constants µ0 and A5 , depending on p, g, Λ, N such that for all
µ ∈ C, Reµ > µ0 , we have
Rebµ (y, y) > A5 kyk2V for all y ∈ V .

Proof. Proof of theorem: 3

By using lemma 3, equation (H9 ) can be equivalently written as (16) equation.
The sesquilinear form bµ in V is trivially continuous and by lemma 3(2), it is also
coercive, when Reµ > mu0 .
we then apply Lax-Milgram theorem to get the existence of a unique solution y of
(H9 ) in V .
suppose that t ∈ H r , r > 1 . To recognize that y ∈ H r+2 , we apply the same
bootstrap procedure applied in lemma 3(1). so, we have to prove that, for any
m ∈ M0 , we have
y ∈ H min{r+2,m+7/2} (Λ) and y|χ ∈ H min{r+2,m+4} (χ) −→ (22)
for large n is enough our claim.
From equation (22) =⇒ prove by induction method
when n = 0, It holds that
By (H91 ) =⇒ ∆y = µ − t ∈ H min{r,3} (Λ)
and y|χ ∈ H 3 (χ),
then by elliptic regularity we have
y ∈ H min{r+2,5,7/2} (Λ) = H min{r+2,7/2} (Λ) which is the required regularity
condition on Λ, when n = 0
According to trace theorem,
we have yν ∈ H min{r+1/2,2} (χ). since gyν ∈ H min{r+1/2,2} (χ), µy ∈ H 3 (χ) and
t ∈ H min{r,r+1/2,2,3} (χ) by (H92 ) we have
and consequently/ equivalently

29
−∆χ y|χ + y|χ ∈ H min{r,2} (χ) By isomorphism property: −∆χ + 1 can be recalled,
then we get y|χ ∈ H min{r+2,4} (χ) which completes the proof when n = 0. To complete
the induction process, we now suppose that (22) holds. by continuing the case n =
0 by (H91 ),
we get, ∆y = µy − t ∈ H min{r,m+9/2} (Λ) and
so again elliptic regularity, y|χ ∈ H min{r+2,n+9/2} (χ) and also by trace theorem
yν ∈ H min{r+1/2,n+3} , with the help of (H92 ) −∆χ y + y|χ ∈ H min{r,n+3} (χ)
As before as, y |χ ∈ H min{r+2,n+5 (χ) completing the induction process.
At last , we prove an equation (4),
so we set an operator Fµ : B(Fµ ) → H r , where
B(Fµ ) = {(y, z) ∈ H r+2 : (∆y)|χ = gyν + p∆χ z}
   
y −∆y + µy
and Fµ   =   so, we notice that B(Fµ ) is closed in
z −gyν − p∆χ z + µz
H r+2 , so it is a Hilbert space with respect to the scalar product and Fµ is bounded
and also y ∈ H r+2 solve (H9 ) if and only if y ∈ B(Fµ ) and Bµ y = t . Therefore, by
our analysis, Fµ is bi-jective, hence (4) follows that Closed graph theorem.

Theorem 2.6.4. Operator D generates an analytic semigroup {K(s), s > 0} in H,

and
kK(s)kL(H) 6 eµ0 t for all s > 0 −→ (23)
where µ0 is the positive number given in theorem (3), so {K(s), s > 0} is quasi
contractive.

Proof. Now introduce the unbounded operator O in H by X(O) = X(D) and O =

D − µ0 I
Then, given y ∈ X(O) , we have that y solves (H9 ), when µ = µ0 and h = −Oy.
Hence by equation (6),
(Oy, y)H = −(h, y)H = −bµ0 (y, y) for all y ∈ y(O) −→ (24)
Then, by using lemma 3(2). we get Re(Oy, y)H > 0 , for all y ∈ Y (0)

30
(i.e.,) O is a dissipative operator in H. ” By using theorem (3)” =⇒ R (I - 0) = H
Then, we get X(0) is dense in H. Meanwhile, given any y ∈ Y (0) by lemma 3.2 and
(22) , we have
Re(−Oy, y)H = Rebµ0 (y, y) > A5 kyk2V −→ (25)
while, equation (35) is continuity of bµ0
|Im(−Oy, y)H | 6 |bµ0 (y, y)| 6 A9 kyk2V −→ (26)
for some A9 = A9 (g, p, M, Λ) > 0 . combining (26) and (25), we get that -0 is a
densely defined M-sectional operator in H, generates an analytic semigroup theory.
which shows that D generates an analytic contraction semigroup {K(s), s > 0} in
H and consequently D generates an analytic semigroup.
so (24), Hence it follows that

Now, we can prove the proofs of our main results.

Proof. theorem 1
By theorem (3(2)), the operator D generates the analytic and hence differentiable,
quasi contractive semigroup {K(s), s >} in H. Then, by semigroup theory given any
y0 ∈ H there is a unique solution,
y ∈ C([0, ∞); H) ∩ C 1 ((0, ∞); H) −→ (27)
of the abstract cauchy problem,


y 0 (s) = Dy(s),

s>0



y(0) = y0


−→ (28)
we clearly compare that the equation (27) is nothing but (20) and (28) is the ab-
stract form of problem (H1 ) parallely (H4 ) is nothing but (24) due to lemma 3.1.
Next, by using the differentiability property of semigroup {K(s), s > 0}, we get
0
that and equivalently Oy = Dy − µ0 y = y − µ0 y ∈ C ∞ ((0, ∞); H) . By (4), then
we get . As by standard procedure, then gives that,y ∈ C ∞ ((0, ∞); H 3 ) , since
y ∈ C ∞ ((0, ∞); H 2n+1 ) for all n ∈ N. By Morrey’s theorem, Then we get (H5 )

31
holds.

Proof. theorem 2:
we introduce, by recurrence on n ∈ N, the space.
X(Om ) = {y ∈ Y (Om−1 ) : Oy ∈ X(Om−1 )} −→ (29)
endowed with the graph norm
m
kyk2X(Om ) = kX i yk2H
P
i=0
By using theorem 3, we get
X(O)m = {y ∈ H 2n+1 : (∆i y)|χ = g(∆i−1 y)ν + p∆χ (∆i−1 y)|χ, i = 1, ..., n} −→ (30)
and that the graph norm is equivalent to the norm of H 2n+1 . since O is a dissipative
operator in H and R( I - 0) = H, we are able to apply the procedure outlined in the
proof
in the real case, which works as well in the complex one. consequently, since y0 ∈
O(X m ) , we get
y ∈ C([0, ∞); X(Om )) ∩ C 1 ([0, ∞); X(Om−1 )) ∩ ... ∩ C n ([0, ∞); H)
which satisfies the previous note/remark. Finally, if y0 ∈ C ∞ (Λ) and (H6 ) holds for
all i ∈ N, we apply previous analysis for any n ∈ N, together with Morrey’s theorem
to get (H8 ).

32