• Keine Ergebnisse gefunden

Construction of exact control for a one-dimensional heat equation with delay

N/A
N/A
Protected

Academic year: 2022

Aktie "Construction of exact control for a one-dimensional heat equation with delay"

Copied!
12
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

Universität Konstanz

Construction of Exact Control for a One-Dimensional Heat Equation with Delay

Denys Ya. Khusainov Michael Pokojovy Elvin I. Azizbayov

Konstanzer Schriften in Mathematik und Informatik Nr. 317, Juli 2013

ISSN 1430-3558

© Fachbereich Mathematik und Statistik Universität Konstanz

Fach D 197, 78457 Konstanz, Germany

Konstanzer Online-Publikations-System (KOPS)

URL: http://nbn-resolving.de/urn:nbn:de:bsz:352-241387

(2)
(3)

Equation with Delay

Denys Ya. Khusainov

, Mihael Pokojovy

, Elvin I. Azizbayov

July21, 2013

Abstrat

Weproveanexatontrollabilityresultforaone-dimensionalheatequationwithdelayinboth

lowerand highestordertermsandnonhomogeneousDirihletboundaryonditions. Moreover,we

giveanexpliit representationoftheontrol funtion steeringthesysteminto agivennal state.

UnderertaindeaypropertiesfororrespondingFourieroeientswhihanbeinterpretedasa

suientlyhigh Sobolev regularityof thedata,bothontrolfuntion andthe solutionare proved

to beregularinthelassialsensebothwithrespettotimeandspaevariables.

1 Introdution

Studying and developing mathematial models to desribe various phenomena in physis, eonomis,

eologyandpopulationdynamis,et.,areoneofentralproblemsofthemodernappliedmathematis

(f. [6℄, [14℄). Integral and dierential equations with lumped and distributed parameters proved to

be a useful and eient tool for suh studies. Whereas evolution equations with lumped parameters

have alreadybeenrather well investigated (see,e.g.,[7 ℄), therestill remaina lotofopen questionsfor

the ase of dynamialsystems withdistributed parameters (p. monographs [12 ℄, [13℄ and referenes

therein).

The sope of the present paper is a linear one-dimensional heat equation in a bounded domain with

disrete delay interms of both lower and highestorders. Reently, an abstrat semigroup treatment

wasproposedfor distributedsystemswithdelays (viz. [3 ℄,[4 ℄). Thoughthis rathergeneral framework

provides good analytialand ontrol-theoretial tools for various delaysenarios, tehnial diulties

may arrive when applying to problems with delayin thehighestorder terms whih have nevertheless

been solved in[5℄for ertainparaboli-type equations.

Another important problem onsistsinobtaining expliitrepresentation formulas for thesolutions to

distributed evolution equations with delay. We refer to [2 ℄, [9℄, [10 ℄, [11 ℄ for details. Suh represen-

tation formulas an then be naturally used to arefully study the solutions, obtain semi-analytial

approximations, addressontrollabilityand optimal ontrol problems,et.

2 Representation of solutions to the heat equation with delay

In [11℄,anonhomogeneous one-dimensional heatequationwithdelay

v t (x, t) = a 2 1 v xx (x, t)+a 2 2 v xx (x, t −τ )+b 1 v x (x, t)+b 2 v x (x, t −τ )+d 1 v(x, t)+d 2 v(x, t −τ )+g(x, t),

(2.1)

DepartmentofCybernetis,KievNationalTarasShevhenkoUniversity,Ukraine

DepartmentofMathematisandStatistis,UniversityofKonstanz,Germany

DepartmentofMehanisandMathematis,BakuStateUniversity,Azerbaijan

(4)

dened for

0 ≤ x ≤ l

and

t ≥ 0

(

l > 0

), wasstudied. The oeients for the phasederivatives were assumed to be proportional, i.e., theremust exista onstant

µ ∈ R

suhthat

µ = − 2a b 1 2

1

= − 2a b 2 2 2

holds

true. A Dirihlet initial boundary valueproblemwithnonhomogeneous initial

v(x, t) = ψ(x, t)

for

0 ≤ x ≤ l, −τ ≤ t ≤ 0

(2.2)

and boundaryonditions

v(0, t) = θ 1 (t), v(l, t) = θ 2 (t)

for

t ≥ −τ

(2.3)

wasonsideredunderan additional ompatibilityondition onthedata:

ψ(0, t) = θ 1 (t), ψ(l, t) = θ 2 (t)

for

t ≥ −τ

Performing thesubstitution

v(x, t) := e µx u(x, t)

with

µ = − b 1

2a 2 1 = − b 2 2a 2 2 ,

Equation (2.1 )wastransformed to

u t (x, t) = a 2 1 u xx (x, t) + a 2 2 u xx (x, t − τ ) + c 1 u(x, t) + c 2 u(x, t) + f (x, t)

(2.4)

with

c 1 := d 1 − b 2 1

4a 2 1 , c 2 := d 2 − b 2 2

4a 2 2 , f (x, t) := e −µx g(x, t)

wherebythe initial andboundaryonditions readas

u(x, t) = ϕ(x, t)

for

0 ≤ x ≤ l, −τ ≤ t ≤ 0, ϕ(x, t) := e −µx ψ(x, t)

(2.5)

and

u(x, 0) = µ 1 (t), u(l, t) := µ 2 (t)

for

− τ < t < 0, µ 1 (t) := θ 1 (t), µ 2 (t) := e −µl θ 2 (t),

(2.6)

respetively.

Following [10 ℄, the delayed exponential funtion

exp τ (b, ·)

wasintrodued.

Denition 1. For

τ > 0

,

b ∈ R

(or

b ∈ C

),dene for eah

t ∈ R

:

exp τ (b, t) :=

 

 

 

 

 

 

 

 

 

 

0, −∞ < t < −τ,

1, −τ ≤ t < 0,

1 + b 1! t , 0 ≤ t < τ, 1 + b 1! t + b 2 ( b−τ) 2! 2 , τ ≤ t < 2τ,

. . . . . .

1 + b 1! t + · · · + b k (t−(k−1)τ)

k

k! , (k − 1)τ ≤ t < kτ,

. . . . . .

(2.7)

See Figure 1for a plotof the delayed exponential funtion.

Using thespeialfuntion giveninEquation (2.7 ),thelassial solutionto theinitial boundary value

problem (2.4 )(2.6 )withdelay anbe representedas

u(x, t) = S 1 (ϕ, µ 1 , µ 2 )(x, t) + S 2 (f, µ 1 , µ 2 )(x, t) + µ 1 (t) + 2 (t)−µ l 1 (t) x

(2.8)

(5)

−1 −0.5 0 0.5 1 1.5 2 0

1 2 3 4 5 6 7

t ex p τ ( − 1 , t )

τ = 0.01 τ = 0.1 τ = 0.5

−1 −0.5 0 0.5 1 1.5 2

0 1 2 3 4 5 6 7

t ex p τ (0 , t )

τ = 0.01 τ = 0.1 τ = 0.5

−1 −0.5 0 0.5 1 1.5 2

0 1 2 3 4 5 6 7

t ex p τ (1 , t )

τ = 0.01 τ = 0.1 τ = 0.5

Figure1: Delayed exponential funtion

exp τ (b, ·)

withlinear operators

S 1 (ϕ, µ 1 , µ 2 )(x, t) =

X

n=1

e L n (t+τ) exp τ (D n , t)Φ n (−τ )+

Z 0

−τ

e L n (t−s) exp τ (D n , t − τ − s)( ˙ Φ n (s) − L n Φ n (s))ds

sin( πn l x), S 2 (f, µ 1 , µ 2 )(x, t) =

X

n=1

Z t 0

e L n (t−s) exp τ (D n , t − τ − s)F n (s)ds

sin( πn l x ,

(2.9)

where

F n (t) = 2 l

Z t 0

f (ξ, t) sin( πn l x)ξdξ + M n (µ 1 , µ 2 )(t), Φ n (t) = 2

l Z l

0

ϕ(ξ, t) sin( πn l ξ)dξ + m n (µ 1 , µ 2 )(t)

(2.10)

with

M n1 , µ 2 )(t) = 2 l

Z l 0

− d dt

µ 1 (t) + µ 2 (t)−µ l 1 (t) ξ

+ b l 21 (t − τ ) − µ 2 (t − τ ))

sin( πn l ξ)ξdξ+

c 1 m n (µ 1 , µ 2 )(t) + c 21 , µ)(t − τ ), m n1 , µ 2 )(t) = 2

l Z l

0

µ 1 (t) + ξ

l (µ 2 (t) − µ 1 (t))

sin( πn l ξ)dξ

and

L n = c 1πn l a 1 2

, D n =

c 2πn l a 2 2 e

c 1 − ( πn l a 1 ) 2 τ .

(2.11)

(6)

inEquation (2.9 ). Regarding itsonvergene, thefollowing resultwasshown in[11℄(f. also[10℄).

Theorem 2. For

T > 0

,

m := ⌈ T τ

and

α > 0

, let

F ∈ C 0 ([0, l] × [0, T ], R )

,

Φ, ∂ t Φ, ∂ tt Φ ∈ C 0 ([0, l] × [0, T ], R )

be suh thattheir Fourier oeients

F n

and

Φ n

satisfy

n→∞ lim n 2m+1+δ max

s∈[−τ,0]

′′ n (s)| + n 2 n (s)| + n 4 |Φ n (s)|

= 0,

n→∞ lim max

1≤k≤m n 2(m−k)+1+δ max

(k−1)τ≤s≤kτ

|F n (s)| + n 2 |F n (s)|

= 0.

Undertheseonditions,problem(2.4 )(2.6 )possessesauniquelassialsolution

u ∈ C 0 ([0, l]×[0, T ], R )

with

t u, ∂ xx u ∈ C 0 ([0, l] × [0, T ], R )

. Moreover, the funtions

u

,

t u

, and

xx u

are represented by uniformlyandabsolutelyonvergentFourierseriesgivenin(2.8)orobtainedbyaterm-wiseappliation

of

∂ t

or

∂ xx

to (2.8 ), respetively.

Remark 3. Using standard arguments from the ellipti theory, the onditions of Theorem 2 an be

interpreted as arequirement for the data

ϕ

,

µ 1

,

µ 2

,

f

to belong to ertain Sobolev spaes (f. [1 ℄)of

funtions withsuiently many weak derivatives (s. [11℄). The larger

T

and

α

are,thesmoother the

dataare supposedto be.

Representing ofthe solution to the initial boundary value problem(2.4 )(2.6 )intheform (2.8 )is not

alwaysonvenient when the impatoftheinitial andboundary valuesor theinhomogeneityhasto be

treated separately. For our purposes,itis neessarytosplit orrespondingterms into dierent sums.

Expanding therst sumin(2.8) and performing integration byparts, we obtain

S 1 (ϕ, µ 1 , µ 2 )(x, t)

=

X

n=1

e L n (t+τ) exp τ (D n , t)Φ n (−τ )

sin( πn l x)+

X

n=1

Z 0

−τ

e L n (t−s) exp τ (D n , t − τ − s)( ˙Φ n (s) − L n Φ n )ds

sin( πn l x)

=

X

n=1

e L n (t+τ) exp τ (D n , t)Φ n (−τ )

sin( πn l x)+

X

n=1

e L n (t−s) exp τ (D n , t − τ − s)Φ n (s)

s=0 s=−τ

sin( πn l x)−

X

n=1

Z 0

−τ

−L n e L n (t−s) exp τ (D n , t − τ − s)

Φ n (s)ds

sin( πn l x)−

X

n=1

Z 0

−τ

−e L n (t−s) D n exp τ (D n , t − 2τ − s)

Φ n (s)ds

sin( πn l x)

=

X

n=1

e L n t exp τ (D n , t − τ )Φ n (0) + D n Z 0

−τ

e L n (t−s) exp τ (D n , t − 2τ − s)Φ n (s)ds

sin( πn l x).

Plugging

Φ n

from Equation(2.10 ),weget

S 1 (ϕ, µ 1 , µ 2 )(x, t)

=

X

n=1

e L n t exp τ (D n , t) 2

l Z l

0

ϕ(ξ, 0) sin( πn l ξ)dξ + m n1 , µ 2 )(0)

sin( πn l x)+

X

n=1

D n

Z 0

−τ

e L n (t−s) exp τ (D n , t−2τ −s) 2

l Z l

0

ϕ(ξ, s) sin( πn l x)dξ+m n (µ 1 , µ 2 )(s)

ds

sin( πn l x).

(7)

Expanding thesum inEquation(2.9) and pluggin

F n

fromEquation (2.10 )yields

S 2 (ϕ, µ 1 , µ 2 )(x, t)

=

X

n=1

Z t 0

e L n (t−s) exp τ (D n , t − τ − s)F n (s)

sin( πn l x)

=

X

n=1

Z t 0

e L n (t−s) exp τ (D n , t − τ − s) 2

l Z l

0

f (ξ, s) sin( πn l ξ)dξ+M n (µ 1 , µ 2 )

ds

sin( πn l x).

Thus,thesolution

u

to the initial boundary value problem(2.4)(2.6)given inEquation (2.8 ) an be

written asfollows:

u(x, t) =

X

n=1

e L n t exp τ (D n , t − τ ) 2

l Z t

0

ϕ(ξ, 0) sin( πn l ξ)

sin( πn l x)+

X

n=1

e L n t exp τ (D n , t)m n1 , µ 2 )(0)

sin( πn l x)+

X

n=1

D n

Z 0

−τ

e L n (t−s) exp τ (D n , t − 2τ − s)

2 l

Z l 0

ϕ(ξ, s) sin( πn l ξ)dξ

ds

sin( πn l x)+

X

n=1

D n

Z 0

−τ

e L n (t−s) exp τ (D n , t − 2τ − s)m n (µ 1 , µ 2 )(s)ds

sin( πn l x)+

X

n=1

Z t 0

e L n (t−s) exp τ (D n , t − τ − s)

2 l

Z l 0

f (ξ, s) sin( πn l ξ)dξ

ds

e α 2 x sin( πn l x)+

X

n=1

Z t 0

e L n (t−s) exp τ (D n , t − τ − s)M n1 , µ 2 )ds

sin( πn l x) + µ 1 (t) + µ 2 (t)−µ l 1 (t) x.

Now, we ollet appropriate terms inthe following three operators therst one depending on the

initial data:

S ˜ 1 (ϕ)(x, t) =

X

n=1

e L n t exp τ (D n , t − τ ) 2

l Z t

0

ϕ(ξ, 0) sin( πn l ξ)

sin( πn l x)+

X

n=1

D n

Z 0

−τ

e L n (t−s) exp τ (D n , t − 2τ − s)

2 l

Z l 0

ϕ(ξ, s) sin( πn l ξ)dξ

ds

sin( πn l x),

theseond one depending on theboundary data:

S ˜ 21 , µ 2 )(x, t) =

X

n=1

e L n t exp τ (D n , t)m n (µ 1 , µ 2 )(0)

sin( πn l x)+

X

n=1

D n

Z 0

−τ

e L n (t−s) exp τ (D n , t − 2τ − s)m n1 , µ 2 )(s)ds

sin( πn l x)+

X

n=1

Z t 0

e L n (t−s) exp τ (D n , t − τ − s)M n (µ 1 , µ 2 )ds

sin( πn l x) + µ 1 (t) + µ 2 (t)−µ l 1 (t) x,

and thethird one depending onthe inhomogeneity:

S ˜ 2 (f )(x, t) =

X

n=1

Z t 0

e L n (t−s) exp τ (D n , t − τ − s)

2 l

Z l 0

f(ξ, s) sin( πn l ξ)dξ

ds

sin( πn l x).

Thus,we arrive at

u(x, t) = ˜ S 1 (ϕ)(x, t) + ˜ S 21 , µ 2 )(x, t) + ˜ S 2 (f )(x, t) + µ 2 (t)−µ l 1 (t) x.

(2.12)

(8)

In this setion, we onsider the following exat ontrollability problem. Given an initial state

ϕ

and

boundary data

γ 1 , γ 2

, replae

f

with a ontrol funtion

U

suh that the solution

u

to (2.4 )(2.6 ) is steered into agiven nalstale

Ψ

at a presribed time

T > 0

,i.e.,

u(x, T ) = Ψ(x)

for

0 ≤ x ≤ l.

(3.1)

Sineweareinterestedinlassialsolutions,aompatibilityonditionontheboundaryonditionsand

theend state hasto beimposed:

Ψ(0) = µ 1 (T ), Ψ(l) = µ 2 (T).

As it follows from the representation formula given in Equation (2.12 ), Equation (3.1 ) is satised if

and only if

S ˜ 1 (ϕ)(x, T ) + ˜ S 2 (µ 1 , µ 2 )(x, T ) + ˜ S 2 (U )(x, T ) + µ 2 (T )−µ l 1 (T ) x = Ψ(x)

for

0 ≤ x ≤ l.

(3.2)

Weexpandthefuntions

Ψ

and

x 7→ µ 2 (T )−µ l 1 (T ) x

ontheinterval

(0, l)

intoFourier serieswithrespet

to the eigenfuntions ofthe orresponding elliptioperator. Equation(2.10 )yields then

Ψ(x) =

X

n=1

Ψ n sin( πn l x)

with

Ψ n (x) = 2 l

Z l 0

Ψ(ξ) sin( πn l ξ)dξ

for

n ∈ N , µ 1 (t) + µ 2 (t)−µ l 1 (t) =

X

n=1

m n1 , µ 2 ) sin( πn l x).

Weassume now

U

to alsohave anexpansion inFourier series ofthe form:

U (x, t) =

X

n=1

U n (t) sin( πn l x).

(3.3)

The operator

S ˜ 3 (U )

readsthenas

S ˜ 3 (U)(x, t) =

X

n=1

Z t 0

e L n (t−s) exp τ (D n , t − τ − s)U n (s)ds

sin( πn l x).

Thus,theontrollability onditionrewrites as

S ˜ 1 (ϕ)(x, T ) + ˜ S 21 , µ 2 )(T, x) +

X

n=1

Z T 0

e L n (t−s) exp τ (D n , t − τ − s)U n (s)ds

sin( πn l x)+

X

n=1

m n (µ 1 , µ 2 )(T ) sin( πn l x) =

X

n=1

Ψ n sin( πn l x).

Denote

s 1n (t) = e L n (t) exp τ (D n , t) 2

l Z l

0

ϕ(ξ, 0) sin( πn l ξ)dξ

+ D n

Z l 0

e L n (t−s) exp τ (D n , t − 2τ − s) 2

l Z l

0

ϕ(ξ, s) sin( πn l ξ)dξ

ds, s 2n (t) =

Z t 0

e L n (t−s) exp τ (D n , t − τ − s)M n (µ 1 , µ 2 )(s)ds.

(9)

There follows thenfrom (3.2) that the ontrollabilityproblem for an arbitrarytime

T > 0

redues to

nding funtions

u n

satisfying the following ondition

X

n=1

(s 1n (T ) + S 2n (T )) sin( πn l x) +

X

n=1

Z t 0

e L n (t−s) exp τ (D n , T − τ − s)U n (s)ds

sin( πn l x)+

X

n=1

m n1 , µ 2 )(T ) sin( πn l x) =

X

n=1

Ψ n sin( πn l x),

whih isinits turnequivalent toa systemof ountably manyFredholmintegral equations oftherst

type:

s 1n (T ) + s 2n (T ) + Z T

0

e L n (T −s) exp τ (D n , T − τ − s)U n (s)ds + m n1 , µ 2 )(T ) = Ψ n

for

n ∈ N .

(3.4)

Lemma 4. For

τ > 0

,

D 6= 0

,thereholds for arbitrary

T > 0 Z T −τ

−τ

exp τ (D, s)ds = D 1 (exp τ (D, T ) − 1).

Proof. There existsa unique

k ∈ N

suh that

(k − 2)τ ≤ T − τ < (k − 1)τ

. Therefore,

Z T −τ

−τ

exp τ (D, s)ds = Z 0

−τ

ds + Z τ

0

1 + D 1! s ds +

Z 2τ τ

1 + D 1! s + D 2 ( s−τ) 2! 2 ds+

Z 3τ 2τ

1 + D 1! s + D 2 ( s−τ) 2! 2 + D 3 ( s−2τ) 3! 3

ds + · · · + Z (k−1)τ

(k−2)τ

1 + D 1! s + D 2 ( s−τ) 2! 2 + D 3 ( s−2τ) 3! 3 + · · · + D k (s−(k−2)τ) (k−1)! k−1 ds.

Performing theintegration, we obtain

Z T −τ

−τ

exp τ (D, s)ds = 1! s

s=0 s=−τ +

s

1! + D s 2! 2

s=t s=0 +

s

1! + D s 2! 2 + D 2 ( s−τ) 3! 3

s=2τ s=τ + s

1! + D s 2! 2 + D 2 (s−τ) 3! 3 + D 3 (s−2τ) 4! 4

s=3τ

s=2τ + · · · + s

1! + D s 2! 2 + D 2 ( s−τ) 3! 3 + D 3 ( s−2τ) 4! 4 + · · · + D k (s−(k−2)τ)

k

k!

s=T −τ s=(k−2)τ

= 1! τ +

τ

1! + D τ 2! 2

+

1! + D (2τ) 2! 2 + D 2 τ 3! 3

τ

1! + D τ 2! 2 + 3τ

1! + D (3τ) 2! 2 + D 2 (2 3! τ) 3 + D 3 τ 4! 4

1! + D (2τ) 2! 2 + D 2 τ 3! 3

+ · · · + T −τ

1! + D (T −τ) 2! 2 + D 2 ( T −2τ) 3! 3 + D 3 ( T −3τ) 4! 4 + · · · + D k (T −τ−(k−1)τ)

k +1

(k+1)!

− (k−1)τ

1! + D ((k−1)τ) 2! 2 + D 2 (( k−2)τ) 3! 3 + D 3 (( k−3)τ) 4! 4 + · · · + D k−1 τ k! k

=

(T −τ)

1! + D (T −τ) 2! 2 + D 2 ( D−2τ) 3! 3 + D 3 ( T −3τ) 4! 4 + · · · + D k (T −τ−(k−1)τ) (k+1)! k+1 + 1! τ .

Thus,we an write

Z T −τ

−τ

exp τ (D, s)ds = 1 D

1 + D T 1! + D 2 ( T −τ) 2! 2 + D 3 ( T −3τ) 3! 3 + D 4 ( T −3τ) 4! 4 + · · · + D k+1 ( T (k+1)! −kτ) k+1 − 1 .

Thisompletes the proof.

(10)

Z T 0

e L n (T −s) exp τ (D n , T − τ − s)U n (s)ds = R n (T )

(3.5)

where

R n (T) := Ψ n − s 1n (T ) − s 2n (T ) − m n1 , µ 2 )(T ).

Substituting

t := T − τ − s

into (3.5 ),we further obtain

Z T −τ

−τ

e L n (τ+t) exp τ (D n , t)U n (T − τ − t)dt = R n (T ).

(3.6)

Welooknow fora solutionof Equation(3.6) intheform

U n (T − τ − t) = e −L n (τ+t) A n (T ),

where

A(T )

areonstants dependingon

T

. Plugging this into (3.6 ) yields

A n (T )

Z T −τ

−τ

exp τ (D n , t)dt = R n (T ).

Exploiting Equation(3.5 ) fromLemma 4,wean write

A n (T )

D n (exp τ (D n , T ) − 1) = R n (T ).

Thus,we obtain the following Fourier oeientsfor theontrol funtion

U n (t) = e −L n (T −t) exp R n (T )D n

τ (D n ,T )−1 .

Summarizing the alulations above,we have proved thefollowing statement.

Theorem 5. Let

ϕ

,

µ 1

,

µ 2

and

Ψ

be suh that theonditions of Theorem 2 are fullled. Then the

ontrol funtion

U (x, t) =

X

n=1

U n (t) sin( πn l x)

with

U n (t) = e −L n (T −t) exp R n (T )D n

τ (D n ,T )−1 ,

R n (T ) = Ψ n − s 1n (T ) − s 2n (T ) − m n (µ 1 , µ 2 )(T ), s 1n (t) = e L n t exp τ (D n , t)

2 l

Z l 0

ϕ(ξ, 0) sin( πn l ξ)dξ

+ D n

Z 0

−τ

e L n (t−s) exp τ (D n , t − 2τ − s) 2

l Z l

0

ϕ(ξ, s) sin( πn l ξ)dξ)

ds, s 2n (t) =

Z t 0

e L n (t−s) exp τ (D n , t − τ − s)M n1 , µ 2 )(s)ds

solvestheexat ontrollability problem(2.4 )(2.6 ),(3.1 ).

(11)

Weproved anexat ontrollability result inthe lassial settingsfor a one-dimensional heatequation

with delay. For pratial appliations, it wouldthough be desirable to extend these results to a weak

framework as, for example, the one desribed in [5℄ or even beyond it. For

p, q ∈ (1, ∞)

, using the

maximal

L p

-regularity property (f. [15℄, [16 ℄) for the ellipti operator in (2.4), the existene of a unique solution to (2.4 )(2.6 )

u ∈ W 1,p (0, T ), L q (0, l)

∩ L p (0, T ), W 2,q (0, l)

for the data

f ∈ L p (0, T ), L q (0, l)

, ϕ ∈ W 1,p (−τ, 0), L q (0, l)

∩ L p (−τ, 0), W 2,q (0, l) , γ 1 , γ 2 ∈ L p (0, T ), R

an be deduedfrom[5℄. Using thisfat to verifyontrollabilityfor a larger lassof dataand ontrol

funtions willbea partof our furtherinvestigations.

Referenes

[1℄ Adams, R.A.(1975).Sobolev spaes,Pure and Applied Mathematis, Vol. 65,New York-London:

Aademi Press

[2℄ Azizbayov, E.I., Khusainov, D.Ya. (2012), Solution to a heat equation with delay (in Russian),

BulletinofTarasShevhenko NationalUniversityof Kyiv,Series: Cybernetis, 12,pp.414.

[3℄ Bátkai, A., Piazzera, S.(2001). Semigroups and Linear Partial Dierential Equations with Delay,

Journal ofMathematial Analysisand Appliations, Vol. 264, pp.120

[4℄ Bátkai, A., Piazzera, S. (2005). Semigroups for Delay Equations, Resarh Notes in Mathematis,

10A.K.Peters: Wellesley MA

[5℄ Bátkai,A.,Shnaubelt,R.(2004).AsymptotiBehaviourofParaboli ProblemswithDelaysinthe

HighestOrder Derivatives, SemigroupForum, 69(3),pp 369399.

[6℄ Ek, Ch., Garke, H., Knabber, P. (2008). Mathematishe Modellierung. Springer-Verlag Berlin

Heidelberg

[7℄ Gopalsamy, K. (1992). Stability and Osillations in Delay Dierential Equations of Population

Dynamis, Mathematisand ItsAppliations, 74,KluwerAademiPublishers

[8℄ Hale, J.K. (1977). Theory of Funtional Dierential Equations, Applied Mathematial Sienes

Series, 3,pp.1365.

[9℄ Khusainov, D.Ya., Ivanov, A.F., Kovarzh, I.V. (2009). Solution ofone heat equationwith delay,

NonlinearOsillations, 12(2),pp.120

[10℄ Khusainov,D.Ya.,Kukharenko,A.V.(2011),ControlofSolutionofParaboliTypeLinearEqua-

tion, Proeedingsofthe Instituteof MathematisofNAS of Ukraine, 8(2)

[11℄ Khusainov, D.Ya., Pokojovy, M., Azizbayov, E. (2013), On Classial Solvability for a Linear

1D Heat Equation with Constant Delay, submitted to: Journal of Computational and Applied

Mathematis

(12)

andApproximationTheories,Enylopediaof Mathematisand itsAppliations, 74, pp.1644

[13℄ Lasieka,I., Triggiani,R.(2011). AbstratHyperboli-Like Systemsover a FiniteTime Horizon,

Enylopediaof Mathematisand its Appliations, 74,pp.1423 pp

[14℄ Okubo,A.,Levin,S.A.(2001).DiusionandEologialProblems.ModernPerspetives.Springer

Verlag,New York,Berlin,Heidelberg,pp.1467

[15℄ Prüss, J. (2002). Maximal Regularity for Abstrat Paraboli Problems with Inhomogeneous

BoundaryData in

L p

-Spaes,Mathematia Bohemia,127(2), pp.311327

[16℄ Weis,L.(2001).Operator-ValuedFourierMultiplierTheoremsandMaximal

L p

-Regularity.Math- ematisheAnnalen, 319,pp.735758

Referenzen

ÄHNLICHE DOKUMENTE

Usually it is more convenient to use the action than the Hamiltonian of a quantum system in quantum field theory. Instead of explicitly constructing the Green’s function structure

We investigate this problem for the linear heat equation, extending Altm¨uller and Gr¨une (2012) by including the case of Neumann boundary control and comparing this case with

F or a wave equation with pure delay, we study an inhomogeneous initial-boundary value problem.. in a bounded

A mathematical model subject to long wavelength and low Reynolds number approximations is presented in order to study the effects of convective boundary conditions on

For the case of variable coefficients, the uniform (boundary or internal) stabilization problems have been studied by several authors, by using or ex- tending the Riemannian

In this work, the exact bright one-soliton solution of the one-dimensional GPE with time-dependent parameters is directly obtained by using the well-known Hirota method under the

In this paper, HPM has been successfully applied to obtain the approximate analytical solutions of the space-time fractional diffusion equation governing the process of a solute

The evolution of the modulational instability in both the anomalous and normal dispersion regimes is numerically investigated and the effects of the higher-order dispersion