Professional Documents
Culture Documents
Sound Propagation in Stratified Fluids (Applied Mathematical Sciences) - Calvin H. Wilcox
Sound Propagation in Stratified Fluids (Applied Mathematical Sciences) - Calvin H. Wilcox
eR
eo she ten cone [Ly a} some ero nen
with initial values f, g that satisfy (1.6), (1.7). The reduction of the
principle is shown below.2.1, Seamary as
the integral
9) EQ e) = f, Cyucew |? + FG) [Dpuces”) FF oT) ax,
where df = dx,dxpdy, may be interpreted as the energy of the acoustic field
win the ser KER at thne t. Te Le known that aoluttona of (1.8) satiety
the conservation Lav
(1-10) E(u,R 8)» ECOR?,0)
whore E(u,R!,0) <4". Only solutions with finite coal energy are consi-
dored below. A necessary and sufficient condition for u re have this
property ie that the initial state have finite energy [22]
aay J. {90 |? + &*G) gC0 [73 PEG) axe ®
The ‘nietal value problem in {ts classical formulation (1.6), (1.8)
Witt have & solution only ££ 0(3), 69), F(X) and g(X) are sutflctentty
saooth. fowever, for arbitrary p(y), c(y) saristying (14) the problem say
be shown to have a unique solution with finite energy vhonever the initial
state fy has thie property. A proof 4s outlined in §3 below. A fornsh
construction of the solution may be based on the operational calculus for
the Linear operator &, defined by the differential operator
aa) aun et) oo 0+ [Eds tu]
acting in the ilbert epace € with sealar product
i
Ry
TE the dousin of A is defined to be the set of al u @ euch that Vu €
and 7 + tw) € chen A is a aeifadjoint non-negative operator. Horeover,
a4) aCe) © eos Cw EM sin ea Dg
4 the solution of the initial valve problem
piu + A 0, > 0,
aus)
WO) =f, uO) = ge16
THE PROPAGATION PROBLEMS AND THEIR SOLUTIONS
Tr can be shown chat (1.14) is the solution with Finite energy whenever the
Antetal values have finite energy.
‘The Snitiel value problem for (1.2) where P satisfies (1.5) and
ute)
0 for & < T can be reformlated as
ceR,
( pha + a0 + Flee
(1.16)
ec
aay ween a [EM sin Oe) ae ea
:
Ft dg clear that for t > 0 this solution satisfies (1.15) with
wo) oof 0M eas AAT ate
F
18)
ayutoy = [? feos aS Fe) ae
2
‘The renaindor of the analysis will be based on the representation
(2.14), Te will be convenient to rewrite it in the form
a9) wCe) = Re (62,7
where v(t,X) is the complex-valued wave function defined by
ve
2.20) weet) = EA
and
a.) press ales
This representation is valid if f and g are real-valued and al £, £, g and
AVE g are in. A rigorous interpretation of (1.14)-(1.21) may be based
fon the calculus of selfadjeine operators in Hilbert spaces.
Lt ig noteworthy that the sane formalism is applicable to the cases of
semi~infinite and finite layers. For the senf-lafinite layer
A} = (ky > Ob, Xe the Kilbort space with scaler product2.2, The Acoustic Propagater aw
a.22) ads | BGT vay Pay oO) ax
Be
sed the doin of 4 is the set of all w EH auch Mae TUE %
4 Ju) €H and u satisfies the Dirichlet or Neumann condition at
°.
‘The relations (1.14)-(1.21) thea hold as Defore. The analogous statenents
hold for eho Finite layer RE = (XO < y Ch).
§2. THE ACOUSTIC PROPAGATOR
In this section a preciee definition of the acoustic propagator A is
siven and a proof of its celfadjoineness in H is outlined. Here and
throughout this work p(y) and e(y) are assumed to be Lebosgue measurable
functions that satisfy the boundedness conditions (1.4). It fellows that
ey 00 = [cg oon a
F
defines a measure in 4? thas 4s equivalent to Lebesgue measure (Lie. m 4s
absolutely continuous with respect to dX and de/dk = &°%(y) 0G). Hence
she Lebesgue space
2.2) Hm EetR,e FG) PAG) a
{a a Wilbert apace with scalar product (1.13). Note that (1-4) inplics
that ICs equivalent as a norsed space to the usual Lebesgue space
Le(R4,dX), although they axe dlatince as Hilbert spaces.
A selfadjoint realization of the differential operator A in
be denoted by A, is obtained by defining the donain of A to be
ea De) = AER) 9 fw > GOV Hye)
where Lz(R') = Ly(R,dX) and
aay LEY) = Let) 9 fu 6 Ga € EECRPDT
4s tho usual first Sobolev space [1]. ALl the differential operations ia
(2.3), @.4) are to bo enderstoed in the sense of the theory of diacesbue
tions, Te follows that rhe Linear operator A in dofined by (1.12).
(2.3) sacteties
ey Asaheo18
{THE PROLAGHETON PROBLEMS AND THETR SOLUTIONS
where AY is the adjoint of A with respect to the sealar product (1.13). &
proof of (2.5) may be given by the nethod employed in [26]. Alternatively,
@
space (11, p. 322]. Indeed, if one defines a sesguilincar form A In Xt by
5) may be derived from Kato's theory of sesquilinear forms in Hilbert
6) Ba) = UR) Cx
and
an Aww =| G+ ov oon ax
then 4t ds easy to verify that A is closed and non-negative, and chat A Le
the unique selfadfoine noa-negative operator in H associated with A, As an
sddiefonal dividend 4¢ follows fro Kato's second represencation theoren
[a2 p. 332) chat D(A) © LEC) and for alt ve D(A") one has.
8 Ia? as dean = [| el? oo ax
where [slg £4 the norm in X
83. SOLUTIONS WrTe FINETE ENERGY
Te was shown {a [22] that the initial value problem (1.6), (1-8) hae
a unique generalized solution with finite exergy (+ solution ¥FE) whenever
the initial state £, g has finite energy; 4.e., (L-ll) holds. Taese
conditions way be weieten, By (2.7),
on ACE f) + llgl? = JAX of? + Il? <
or, equivalently,
2) FeO), sex,
Ie follows that the function u(c,X) defined by (1.14) is the unique solution
FE and
6.3) we CRB 9 CR AERD),
(ere CK(1,I0 is the set of k tines continuously differentioble X-valued
function on I.)2.3, Solutions with Finies Enexoy co
the acoustic fields studied below are defined by (1.14) where £, g
setinfy (3.2). Te will be convenient to specialize further by writing u
in the form (1.19)-(1.21), This Se correct provided £ and g are real
valued, (3-2) holds and g © D(A). Of course, D¢s”"*) # H, in general.
However, D(A!) is dense in H, by the spectral theorem. This fact can
often be used to extend results derived from (1.19)-(1.21) to general
solutions «FE.Chapter 3
Spectral Analysis of Sound
Propagation in Stratified Fluids
‘the purpose of thia chapcer 4e to construct families of normal mode
functions end te derive corresponding normal mode expansions for the
acoustic propagator A, The basic physical concepts and results are sun~
narized in §1. A aore detatled formulation and complete proofs of ail che
results are developed in tho remaining sections. This rather long chapter
provides the analytical foundation for the solutions of the propagation
problems that are developed in Chapters 4 and 5.
Si. surmany
‘Throughout thie chapter ir ts assumed chat p(y) and e(y) saristy che
boundedness conditions (2.1.4) and the four conditions:
© Lo) = 891 ay <#y 2f" Jeon = ecm! ay