Gui-Qiang Chen and Mikhail Feldman - Existence and Stability of Multidimensional Transonic Flows Through An Infinite Nozzle of Arbitrary Cross-Sections

EXISTENCE AND STABILITY OF MULTIDIMENSIONAL TRANSONIC
FLOWS THROUGH AN INFINITE NOZZLE OF ARBITRARY

CROSS-SECTIONS
GUI-QIANG CHEN AND MIKHAIL FELDMAN
Abstract. We establish the existence and stability of multidimensional steady transonic ows
with transonic shocks through an innite nozzle of arbitrary cross-sections, including a slowly
varying de Laval nozzle. The transonic ow is governed by the inviscid potential ow equation
with supersonic upstream ow at the entrance, uniform subsonic downstream ow at the exit
at innity, and the slip boundary condition on the nozzle boundary. Our results indicate that,
if the supersonic upstream ow at the entrance is suciently close to a uniform ow, there
exists a solution that consists of a C
1,
subsonic ow in the unbounded downstream region,
converging to a uniform velocity state at innity, and a C
1,
multidimensional transonic shock
dividing the subsonic ow from the supersonic upstream ow; the uniform velocity state at
the exit at innity in the downstream direction is uniquely determined by the supersonic
upstream ow; and the shock is orthogonal to the nozzle boundary at every point of their
intersection. In order to construct such a transonic ow, we reformulate the multidimensional
transonic nozzle problem into a free boundary problem for the subsonic phase, in which the
equation is elliptic and the free boundary is a transonic shock. The free boundary conditions
are determined by the Rankine-Hugoniot conditions along the shock. We further develop a
nonlinear iteration approach and employ its advantages to deal with such a free boundary
problem in the unbounded domain. We also prove that the transonic ow with a transonic
shock is unique and stable with respect to the nozzle boundary and the smooth supersonic
upstream ow at the entrance.
1. Introduction
We are concerned with the existence and stability of multidimensional steady transonic ows
with transonic shocks through multidimensional innite nozzles of arbitrary cross-sections. Such
problems naturally arise in many physical situations, especially in the de Laval nozzles which
have widely been used in the design of steam turbines and modern rocket engines (see Courant-
Friedrichs [11], Whitham [47], and the references cited therein). Since the nozzles in applications
are usually much longer with respect to their cross-sections, the problem is often formulated
mathematically as an innite nozzle problem. Correspondingly, such a multidimensional innite
nozzle problem has extensively been studied experimentally, computationally, and asymptotically
(see [11, 15, 18, 21, 22, 23, 47] and the references cited therein). The stability issue for transonic
gas ow through a nozzle for the one-dimensional model was analyzed in [40, 41]. The existence
and stability of multidimensional steady transonic ows through multidimensional innite nozzles
has remained open; see [5, 11, 12, 43, 47].
In this paper, we focus on the innity nozzle to establish the existence and stability of mul-
tidimensional transonic ows with supersonic upstream ows at the entrance, uniform subsonic
downstream ows at the exit at innity, and the slip boundary condition on the nozzle bound-
ary. The potential ow equation for the velocity potential : R
n
R is a second-order
Date: November 1, 2005.
1991 Mathematics Subject Classication. 35M10,35J65,35R35,76H05,76L05,35B45.
Key words and phrases. Elliptic-hyperbolic, nonlinear equations, second order, mixed type, multidimensional
transonic shocks, free-boundary problems, unbounded domains, innite nozzles, cross-sections, existence, unique-
ness, stability, Euler equations, inviscid potential ow.
1
2 GUI-QIANG CHEN AND MIKHAIL FELDMAN
nonlinear mixed equation of elliptic-hyperbolic type:
div ((|D|
2
)D) = 0, x R
n
, (1.1)
where the density function (q
2
) is
(q
2
) =
_
1 q
2
_ 1
2
(1.2)
and =
1
2
> 0 with adiabatic exponent > 1.
The nonlinear equation (1.1) is elliptic at D with |D| = q if
(q
2
) + 2q
2
(q
2
) > 0 (1.3)
and hyperbolic if
(q
2
) + 2q
2
(q
2
) < 0. (1.4)
For this innite nozzle problem, we rst seek a multidimensional transonic ow containing a
multidimensional transonic shock which divides a subsonic downstream ow from the supersonic
upstream ow. In order to construct such a transonic ow, we rst formulate this innite nozzle
problem into a free boundary problem: The multidimensional transonic shock is a free boundary
which divides two phases of C
1,
ow in the innite nozzle, and the equation is hyperbolic in the
supersonic upstream phase and elliptic in the subsonic downstream phase. Since the existence of
the supersonic upstream phase is a direct corollary of the standard local existence theory of the
initial-boundary value problem with the slip boundary condition for nonlinear wave equations (cf.
[27, 28, 30]), we can further formulate the free boundary problem into a one-phase free boundary
problem for a nonlinear elliptic equation by shimanization, a cut-o technique (see [4, 45]).
We further develop a nonlinear iteration approach and employ its advantages to deal with such
a free boundary problem in the unbounded domain and to solve the multidimensional transonic
nozzle problem in a direct fashion. Our results indicate that, for the transonic nozzle problem,
there exists a transonic ow such that the ow is divided into a C
1,
subsonic ow up to the nozzle
boundary in the unbounded downstream region from the supersonic upstream ow by a C
1,
multidimensional transonic shock that is orthogonal to the nozzle boundary at every intersection
point, and the uniform velocity state at the exit at innity in the downstream direction is uniquely
determined by the supersonic upstream ow at the entrance which is suciently close to a uniform
ow. The uniform velocity state at the exit can not be apriori prescribed from the corresponding
pressure for such a ow to exist. We further prove that the transonic ow with a transonic shock
is unique and stable with respect to the nozzle boundary and the smooth supersonic upstream
ow at the entrance.
One of the advantages to employ our nonlinear iteration approach is that, as long as we
know how the corresponding xed conormal problem for (1.1) can be solved and estimated, the
solution of the free boundary problem and its estimates directly follow, even for more complicated
geometry of the domain under consideration. In this sense, the iteration approach is more ecient
than the partial hodograph approach developed in [6], since we do not need to change equation
(1.1) with ne features. This indeed plays an important role to achieve the C
1,
regularity of both
the solution and the free boundary up to the nozzle boundary and to allow the nozzle boundary of
arbitrary cross-sections, by employing the features of equation (1.1). Also, this approach enables
us to deal with the multidimensional nozzle problem in a direct fashion, which especially applies
to the bounded nozzle problems with various dierent compatible boundary conditions at the
exit by solving the elliptic problems with nonsmooth xed boundaries, as discussed in Section 3
and in [9, 10, 13, 24].
Another new ingredient of this paper is to develop a C
1,
estimate framework for solutions
of the conormal problem for elliptic equations of divergence form with C
coecients in the
domains with corners and wedges. In our nozzle problem, the corners and wedges are formed in
MULTIDIMENSIONAL TRANSONIC FLOWS THROUGH AN INFINITE NOZZLE 3
the intersection of the free boundary with the xed boundary, so it is necessary to develop such
an estimate framework to deal with the new diculty of the intersection of a free boundary with
a xed boundary, especially when n 3. In [5], we considered a transonic ow in an innite
cylinder with a cross-section of the form = (0, 1)
n1
, i.e., a ow in the domain (0, 1)
n1
R,
in which, by reections, we can reduce the problem to the domain T
n1
R, where T
n1
is
the (n 1)-dimensional torus. Thus, in studying the free boundary problems in [4, 5], we were
able to avoid studying the intersection of the free boundary with the xed boundary. In the
case of arbitrary smooth cross-section in dimension n 3, this can not be done even for the
ow in the unperturbed cylinder: Indeed, if we locally atten the boundary, the equation in the
new coordinates changes its structure and the reection leads to an equation with discontinuous
coecients. When we consider a ow in the nozzle (i.e. in a perturbed cylinder), we face a similar
problem. Thus, in the present paper, we have to consider the intersection of a free boundary with
a xed boundary, that is, a nonlinear conormal problem in the domain with nonsmooth boundary
for any dimension (n 2). The problem in a xed domain with wedges was considered in [36],
where the C
1,
regularity of solutions near the wedges for a general oblique derivative problem
for linear elliptic equations was obtained. On the other hand, the results in [36] require certain
conditions on the relative geometry of the wedge and the vector elds in the oblique derivative
conditions; and, moreover, the Holder exponent of the solution depends on the geometry of
wedges, in addition to the Holder exponent of the coecients and the right-hand sides of the
equation and boundary conditions. These features present the diculties in using the results of
[36] in constructing solutions for our nonlinear problem. Thus, in this paper, we rst study the
existence and regularity of solutions, by restricting the class of equations and considering only
the conormal boundary conditions and cylindrical domains, and then we obtain sharp results
that are used for constructing the solution of the free boundary problem for the nonlinear elliptic
equation in the subsonic domain.
A further new ingredient is to develop a useful approach for obtaining uniform estimates of
solutions in the (perturbed) cylinders of length R and showing that the velocity (the
gradient of the solution) converges to an asymptotic constant state at innity. In [5], we reduced
the problem to the domain T
n1
R so that the symmetry allows us to use the pointwise
estimates of solutions to study the asymptotics at innity. In the present case, the domains are
of more general geometry, so our approach is now to use natural integral estimates of solutions to
achieve the asymptotics of solutions at innity. This approach should be useful to other problems
in unbounded domains with similar features.
The uniqueness and stability of the nozzle problem is a question of great importance to know
under what circumstances a steady transonic ow involving transonic shocks is uniquely deter-
mined by the boundary condition and the conditions at the entrance and when further conditions
at the exit are appropriate (see Courant-Friedrichs [11]). In this paper, we identify that the steady
transonic ow involving a transonic shock is uniquely determined by the Cauchy data at the en-
trance for the nonlinear wave equation and the uniform subsonic ow condition at the exit at
innity, which are natural physical conditions. Note that the pressure can not be apriori pre-
scribed at the exit at innity even for the existence; otherwise, it is clear from our results that
there is no weak solution for this problem (also see [5]).
Some eorts have been made for solving the nonlinear equation (1.1) of mixed type. In par-
ticular, Shiman [45], Bers [2], and Finn-Gilbarg [17] proved the existence and uniqueness of
solutions for the problem of subsonic ows of (1.1) past an obstacle (also see [14]). Morawetz
in [42] showed that the ows of (1.1) past the obstacle may contain transonic shocks in gen-
eral. In [4, 5, 6], we developed two approaches to deal with multidimensional transonic shocks.
Also see Canic-Keytz-Lieberman [3] for the transonic small disturbance equation and Zheng
[51] for the pressure-gradient equations. For the bounded transonic nozzle problem, see [4] for
the multidimensional channel case and [48] for the 2-dimensional case. Non-transonic shocks
(hyperbolic-hyperbolic shocks) were analyzed in [7, 8, 25, 32, 39, 44, 49] and the references cited
therein.
In Section 2, we set up the innite nozzle problem involving multidimensional transonic shocks
and present the main theorems of this paper. In Section 3, we develop a C
1,
estimate framework
up to the wedge boundary of the cylinder for the conormal problem for second-order linear
elliptic equations of divergence form with C
coecients for subsequent development. In Section

4, we reformulate the transonic nozzle problem into a free boundary problem by introducing a
multidimensional transonic shock as a free boundary which divides the upstream and downstream
phases of C
1,
ow in the innite nozzle. We introduce an iteration procedure to construct
approximate solutions of the free boundary problem in Section 4 and make uniform estimates
of the solutions on the bounded domains with wedge boundary in Section 6. In Section 7, we
establish the existence of multidimensional transonic ows with transonic shocks via solving
the corresponding free boundary problem in the innite nozzle. In Section 8, we establish the
uniqueness and stability of multidimensional transonic ows with transonic shocks in the innite
nozzle.
2. Transonic Shocks, Nozzle Problems, and Main Theorems
In this section, we rst set up the innite nozzle problem involving multidimensional transonic
shocks and present the main theorems of this paper.
A function C
0,1
() is a weak solution of (1.1) in an unbounded domain if
(i) |D(x)| 1/
a.e.
(ii) For any C
0
(),
_
(|D|
2
)D D dx = 0. (2.1)
We are interested in weak solutions with shocks. Let
+
and
be open subsets of such

that
= ,
+

= ,
and S =
+
. Let C
0,1
() be a weak solution of (1.1) and be in C
1
(
S) so that
D experiences a jump across S that is an (n 1)-dimensional smooth surface. Then satises
the following Rankine-Hugoniot conditions on S:
[]
S
= 0,
_
(|D|
2
)D
_
S
= 0, (2.2)
where is the unit normal to S from
to
+
, and the bracket denotes the dierence between
the values of the function along S on the
sides. Moreover, a function C

1
(
S), which
satises |D| 1/
, (2.2), and equation (1.1) in
respectively, is a weak solution of (1.1),

i.e., satises (2.1) in the whole domain .
Set
= |
. Then we can also write (2.2) as
+
=
on S (2.3)
and
(|D
+
|
2
)
+
= (|D
|
2
)D
on S, (2.4)
where
+
= D
+
is the normal derivative on the
+
side.
Note that the function
(p) :=
_
1 p
2
_ 1
2
p (2.5)
is continuous on
_
0,
_
1/
_
and satises
(p) > 0 for p
_
0,
_
1/
_
, (0) =
_
_
1/
_
= 0, (2.6)
0 <
(p) < 1 on (0, c
),
(p) < 0 on
_
c
,
_
1/
_
, (2.7)
(p) < 0 on (0, c
], (2.8)
where
c
=
_
1/( + 1) =
_
2/( + 1) (2.9)
is the sonic speed, for which a ow is called supersonic if |D| > c
and subsonic if |D| < c
.
Suppose that C
0,1
() C
1
(
S) is a weak solution satisfying

|D| < c
in
+
, |D| > c
in
, D
|
S
> 0. (2.10)
Then is a transonic shock solution with transonic shock S dividing into the subsonic region
+
and the supersonic region
and satisfying the physical entropy condition (see Courant-

Friedrichs [11]):
(|D
|
2
) < (|D
+
|
2
) along S.
Note that equation (1.1) is elliptic in the subsonic region and hyperbolic in the supersonic region.
Let (x
, x
n
) be the coordinates in R
n
, where x
n
R and x
= (x
1
, . . . , x
n1
) R
n1
. Let
q
0

_
c
, 1/
_
and
0
(x) := q
0
x
n
. Then
0
is a supersonic solution in . According to
(2.6)(2.7), there exists a unique q
+
0
(0, c
) such that
_
(q
+
0
)
2
_
q
+
0
=
_
(q
0
)
2
_
q
0
. (2.11)
Thus, the function
0
(x) =
_
q
0
x
n
, x
0
:= {x : x
n
< 0},
q
+
0
x
n
, x
+
0
:= {x : x
n
> 0}
(2.12)
is a plane transonic shock solution in ,
+
0
and
0
are its subsonic and supersonic regions
respectively, and S = {x
n
= 0} is a transonic shock.
Dening
+
0
(x) := q
+
0
x
n
in , we have
0
(x) = min(
+
0
(x),
0
(x)) for x . (2.13)
In this paper, we focus on the following innite nozzle with arbitrary smooth cross-sections:
= ( (, )) {x
n
1}, (2.14)
where R
n1
is an arbitrary open bounded, connected set with a smooth boundary, and
: R
n
R
n
is a smooth map, which is close to the identity map. For simplicity, we assume
that
is in C
[
n
2
]+3,
(2.15)
and
Id
[
n
2
]+3,,R
n (2.16)
for some (0, 1) and small > 0, where [s] is the integer part of s, Id : R
n
R
n
is the
identity map, and u
m,,D
is the norm in the Holder space C
m,
(D) in the domain D. Such
nozzles especially include the slowly varying de Laval nozzles [11, 47]. For concreteness, we also
assume that there exists L > L
0
(say, L
0
> 10 without loss of generality) such that
(x) = x for any x = (x
, x
n
) with x
n
> L, (2.17)
that is, the nozzle slowly varies locally in a bounded domain as the de Laval nozzles.
In the two-dimensional case, the domain dened above has the following simple form:
= {(x
1
, x
2
) : x
1
1, f
(x
2
) < x
1
< f
+
(x
2
)},
where f
4,,R
and f
on [L, ) for some constants d
satisfying d
+
> d
.
For the multidimensional case, the geometry of the nozzles is much richer.
Note that our setup implies that
=
o

l
with
l
:= [ (, )] {(x
, x
n
) : x
n
> 1},
o
:= ( (, )) {(x
, x
n
) : x
n
= 1}.
Then our transonic nozzle problem can be formulated into the following form:
Problem (TN): Transonic Nozzle Problem. Given the supersonic upstream
ow at the entrance
o
:
=
e
,
x
n
=
e
on
o
, (2.18)
the slip boundary condition on the nozzle boundary
l
:
= 0 on
l
, (2.19)
and the uniform subsonic ow condition at the exit x
n
= :
() x
n
C
1
({x
n
>R})
0 as R , for some (0, c
), (2.20)
nd a solution of problem (1.1) and (2.18)(2.20) in .
As we will see below, the supersonic upstream part
of the solution can be constructed

in
1
:= {x
n
< 1} from the data on the nozzle entrance (2.18) by using the standard
results on initial-boundary value problems for hyperbolic equations. Thus, assuming that the
C
1,
supersonic solution
is given in
1
, we can formulate the transonic nozzle problem (TN)
as the following one-phase free boundary problem.
Problem (FB): Free Boundary Problem. Given a supersonic upstream ow
, a weak solution of (1.1) in

1
, which is a C
1,
perturbation of
0
:
0

1,,
1
C
0
(2.21)
with > 0 small, for some constant C
0
, and satises (2.19), nd a multidi-
mensional subsonic ow
+
of (1.1) satisfying (2.19)(2.20) and identify a free
boundary x
n
= f(x
) dividing the subsonic ow

+
from the given supersonic
ow
so that
(x) =
_

+
(x), x
n
> f(x
),
(x), x
n
< f(x
)
(2.22)
is a transonic shock solution.
Note that, for a solution of Problem (FB), the subsonic region
+
and supersonic region
in (2.10) are of the form
+
() = {x
n
> f(x
)},
() = {x
n
< f(x
)}. (2.23)
Our main theorem for the free boundary problem, Problem (FB), is the following.
Theorem 2.1 (Existence). There exist
0
> 0, C, and

C, depending only on n, , , q
+
0
,
C
0
, , and L, such that, for every (0,
0
), any map satisfying (2.16)(2.17), and any
supersonic upstream ow
of (1.1) satisfying (2.19) and (2.21), there exists a solution

C
0,1
() C
(
+
) of Problem (FB) such that
D q
+
0
e
n
0,0,
+

C. (2.24)
Moreover, the solution satises the following properties:
(i) The constant in (2.20) must be q
+
:
= q
+
, (2.25)
where q
+
is the unique solution in the interval (0, c
) of the equation
((q
+
)
2
)q
+
= Q
+
(2.26)
with
Q
+
=
1
||
_
(|D
|
2
)D
dH
n1
. (2.27)
Thus, satises
q
+
x
n
C
1
({x
n
>R})
0 as R , (2.28)
and q
+
satises
|q
+
q
+
0
| C; (2.29)
(ii) The function f(x
) in (2.22) satises
f
1,,R
n1 C, (2.30)
and the surface S = {(x
, f(x
)) : x
R
n1
} is orthogonal to
l
at every
intersection point;
(iii) Furthermore, C
1,
(
+
) with
q
+
x
n
1,,
+
C. (2.31)
If the supersonic upstream ow
has a higher regularity, then we have the following unique-

ness theorem.
Theorem 2.2 (Uniqueness). There exists a constant
0
> 0 depending only on n, , , C
0
, ,
L, and q
+
0
such that, if <
0
and the supersonic solution
in Problem (FB) additionally

satises
0

2,,
1
C
0
, (2.32)
then the solution of Problem (FB) satisfying (2.24) is unique.
The standard local existence theory of smooth solutions for the initial-boundary value problem
(2.18)(2.19) for second-order quasilinear hyperbolic equations implies (see Appendix) that, as
is suciently small in (2.16) and (2.21), there exists a supersonic solution
of (1.1) in
1
,
which is a C
k+1
perturbation of
0
: For any (0, 1],
0

k,,
1
C
0
, k = 1, 2, (2.33)
for some constant C
0
> 0, and satises
= 0 on
l
1
, (2.34)
provided that (
e
,
e
) on
o
satisfying
e
q
0
x
n
H
s+k +
e
q
0

H
s+k1 , k = 1, 2, (2.35)
for some integer s > n/2 + 1 and the compatibility conditions up to the (s + k)
th
order, where
the norm
H
s is the Sobolev norm with H
s
= W
s,2
. In particular, as a direct corollary
of Theorems 2.12.2 and Proposition A.1 in Appendix, we obtain the following existence and
uniqueness result for the transonic nozzle problem, Problem (TN):
Theorem 2.3. Let q
+
0
(0, c
) and q
0

_
c
, 1/
_
satisfy (2.11), and let
0
be the transonic
shock solution (2.12). Then there exist
0
> 0, C, and

C, depending only on n, , , q
+
0
, ,
and L such that, for every (0,
0
), any map satisfying (2.16)(2.17), and any supersonic
upstream ow (
e
,
e
) on
o
satisfying
e
q
0
x
n
H
s+2 +
e
q
0

H
s+1 (2.36)
for s > n/2 + 1 and the compatibility conditions up to the (s + 2)
th
order, there exists a unique
solution C
0,1
() of Problem (TN) satisfying (2.23)(2.24) and

0

2,,
C.
Moreover, this solution satises C
0,1
() C
(
+
) and properties (i)(iii) of Theorem
2.1, where (2.27) is replaced by
Q
+
=
1
||
_
(|D
x

e
|
2
+ (
e
)
2
)
e
dH
n1
. (2.37)
Remark 2.1. The smoothness conditions (2.15)(2.16) are just for simplicity of presentation
and can be relaxed: For example, in Theorems 2.12.2, we require only C
2,
in (2.15)(2.16).
The localization condition (2.17) is not essential to achieve Theorem 2.1; In general, it can be
replaced by an appropriate decay condition.
Remark 2.2. When the initial data (
e
,
e
) (
e
,
e
) is constant and the nozzle {1
x
n
1 + } = [1, 1 + ] for some > 0 as a de Laval nozzle, then the compatibility
conditions are automatically satised. In fact, in this case,
(x) =
e
x
n
is a solution near
x
n
= 1 in the nozzle.
Remark 2.3. When n = 2, condition (2.36) for the supersonic upstream ow (
e
,
e
) on
o
in Theorem 2.3 can be replaced by the C

3
-condition:
e
q
0
x
n
C
3 +
e
q
0

C
2 . (2.38)
This can be achieved by following the arguments in Li-Yu [33].
Remark 2.4. There exist
0
and C depending only on the data such that equation (2.26) with
Q
+
dened by (2.27) has a unique solution q
+
(0, c
) satisfying (2.29).
This can be seen as follows: From (2.16), it follows that | |
o
| || | C and | e
n
| C
on
o
, and hence that | e
n
1| C on
o
. Then
|Q
+
((q
+
0
)
2
)q
+
0
|
1
||
_
(|D
|
2
)D
((q
+
0
)
2
)q
+
0
dx
1
|
o
|
||
((q
+
0
)
2
)q
+
0
=
1
||
_
(|D
|
2
)D
((q
0
)
2
)q
dx
1
|
o
|
||
((q
+
0
)
2
)q
+
0
1
||
_
(|D
|
2
)D
(|D
0
|
2
)D
+ ((q
0
)
2
)q
0
| e
n
1|
_
dx
+ C
C,
where we used (2.11) and (2.21). Thus, by (2.6)(2.7), we obtain that, if is small depending
only on the data, then there exists a unique solution q
+
(0, c
) of equation (2.26) such that

(2.29) holds.
Remark 2.5. In fact, the solution with a transonic shock of Problem (TN) in Theorem 2.3 is
also stable with respect to the nozzle boundary and the smooth supersonic upstream ow at the
entrance; see Theorem 8.1.
Remark 2.6. For the isothermal gas = 1, the same results can be obtained by following the
same arguments in this paper.
3. Conormal Problems for Linear Elliptic Equations in a Cylinder
In this section, we develop a C
1,
estimate framework up to the boundary of the cylinder for
a conormal problem for linear elliptic equations of divergence form with C
coecients.
Let R
n1
be an open bounded, connected set with C
2,
boundary.
We x the set and a number (0, 1) through this section.
For Q > 0, denote
C
Q
:= (0, Q),
1
Q
:= {x
n
= 0},
2
Q
:= (0, Q),
3
Q
:= {x
n
= Q}.
Consider the problem
n
i,j=1
(a
ij
u
x
j
)
x
i
= f in C
Q
, (3.1)
n
i,j=1
a
ij
u
x
j
i
= g
k
on
k
Q
, k = 1, 2, (3.2)
u = h on
3
Q
, (3.3)
where is the inward unit normal on
1
Q
and
2
Q
to C
Q
.
The main issue in the argument below is to show the existence and uniqueness of solutions
which are C
1,
up to the edge {0} of the cylinder. Lieberman [36] studied general oblique
derivative problems for linear elliptic equations in the domains with wedges in which the C
1,
reg-
ularity of solutions is obtained near the wedges. However, these results require certain conditions
on the relative geometry of the wedge and the vector elds in the oblique derivative conditions;
Moreover, in [36], the Holder exponent of the solution gradients depends on the geometry of
wedges and the coecients of the equation and the right-hand sides of the boundary conditions,
and < in general where is the Holder exponent of the coecients and the right-hand sides.
In this section, by restricting the class of equations and considering only conormal boundary con-
ditions and cylindrical domains, we obtain the sharp C
1,
regularity of solutions up to {0},
i.e., = . This is important for our approach developed in Sections 48.
Theorem 3.1. Let
1
, . . . ,
n
, and are the constants satisfying 0 <
i

1
for i =
1, . . . , n. Then there exist > 0 and C > 0, depending only on n, , Q, and , such that, when
a
ij
(x) satisfy
a
ij

i
i
j
0,,C
Q
for i, j = 1, . . . , n, (3.4)
and
f L
(C
Q
), h H
1
(C
Q
), (3.5)
g
1
C
0,
(
1
Q
), g
2
C
0,
(
2
Q
), (3.6)
there exists a unique weak solution u H
1
(C
Q
) of (3.1)(3.3) in the sense that (3.3) holds as the
trace and
_
C
Q
(
n
i,j=1
a
ij
u
x
j
w
x
i
+ fw)dx +
2
k=1
_
k
Q
g
k
wdH
n1
= 0 (3.7)
for any function w H
1
(C
Q
) with w = 0 on C
Q
{x
n
= Q}. Moreover, the solution u is
C
1,
(C
Q/2
) and satises
u
1,,C
Q/2
C(u
L
2
(C
Q
)
+f
0,0,C
Q
+g
1
0,,
1
Q
+g
2
0,,
2
Q
). (3.8)
Theorem 3.1 is a direct corollary of Theorem 3.2 below, stated in more technical terms. In
order to state Theorem 3.2, we introduce some notations and weighted Holder norms.
Denote
C
Q
= C
Q
\ {x
n
= Q}, (
2
Q
)
=
2
Q
\ {x
n
= Q}. (3.9)
We will use the following weighted Holder semi-norms and norms in the cylinder C
Q
, in which the
weight is the distance to the portion of the boundary
3
Q
= C
Q
\C
Q
. Denote
x
= dist(x,
3
Q
) =
Q x
n
for x = (x
, x
n
) C
Q
and
x,y
= min(
x
,
y
) for x, y C
Q
. For k R, (0, 1), and
m N (the set of nonnegative integers), dene
|u|
(k)
m;0;C
Q
=
0jm
||=m
sup
xC
Q
_
max(m+k,0)
x
|D
u(x)|
_
, (3.10)
[[u]]
(k)
m;;C
Q
=
||=m
sup
x,yC
Q
,x=y
_
max(m++k,0)
x,y
|D
u(x) D
u(y)|
|x y|
_
,
|u|
(k)
m;;C
Q
= |u|
(k)
m;0;C
Q
+ [[u]]
(k)
m;;C
Q
,
where D
1
x
1

n
x
n
, = (
1
, . . . ,
n
) is a multi-index with
j
0,
j
N, and || =
1
+ +
n
. The weighted Holder norms |u|
(k)
m;;(
2
Q
)
on the boundary part

2
Q
are dened
similarly.
Theorem 3.2. Let
1
, . . . ,
n
, and be the constants satisfying 0 <
i

1
for i =
1, . . . , n. Then there exist > 0 and C > 0, depending only on n, , Q, and , such that, when
a
ij
(x) satisfy (3.4) and
f L
2
(C
Q
) L
loc
(C
Q
), h H
1
(C
Q
), (3.11)
g
1
L
2
(
1
Q
) C
0,
(
1
Q
), g
2
L
2
(
2
Q
) C
0,
((
2
Q
)
) (3.12)
with
|f|
(2+n/2)
0,0,C
Q
+|g
2
|
(1+n/2)
0,,(
2
Q
)
< ,
there exists a unique weak solution u H
1
(C
Q
) of (3.1)(3.3) in the sense that (3.3) holds as
the trace and (3.7) holds for any w H
1
(C
Q
) with w = 0 on C
Q
{x
n
= Q}. Moreover, the
solution u is C
1,
(C
) and satises
|u|
(n/2)
1,,C
Q
C(u
L
2
(C
Q
)
+|f|
(2+n/2)
0,0,C
Q
+g
1
0,,
1
Q
+|g
2
|
(1+n/2)
0,,(
2
Q
)
). (3.13)
Now we prove Theorem 3.2. We rst consider an auxiliary problem:
n
i=1
i
u
x
i
x
i
= f +
n
i=1
i
x
i
in C
Q
, (3.14)
n
i=1
i
u
x
i
i

n
u
x
n
= g
1
on
1
Q
, (3.15)
n
i=1
i
u
x
i
i

n1
i=1
i
u
x
i
i
= g
2
+
n1
i=1
i
on
2
Q
, (3.16)
u = h on
3
Q
, (3.17)
where we used that = e
n
on
1
Q
in (3.15) and e
n
= 0 on
2
Q
in (3.16). Note that smooth
solutions of (3.14)(3.17) satisfy
_
C
Q
(
n
i=1
i
u
x
i
w
x
i

n
i=1
i
w
x
i
+ fw)dx
+
_
1
Q
(g
1
n
) wdx
2
Q
g
2
wdH
n1
= 0 (3.18)
1
(C
Q
) with w = 0 on C
Q
{x
n
= Q}.
Lemma 3.1. Let
1
, . . . ,
n
, and be the constants satisfying 0 <
i

1
for i = 1, . . . , n.
Let
f L
2
(C
Q
) L
loc
(C
Q
), = (
1
, . . . ,
n
) H
1
(C
Q
) C
(C
Q
), h H
1
(C
Q
),
g
1
L
2
(
1
Q
) C
0,
(
1
Q
), g
2
C(C
Q
) L
2
(
2
Q
) C
0,
(C
Q
)
(3.19)
with
|f|
(2+n/2)
0,0,C
Q
+||
(1+n/2)
0,,C
Q
+|g
2
|
(1+n/2)
0,,C
Q
< . (3.20)
Then there exists a unique weak solution u H
1
(C
Q
) of (3.14)(3.17) in the sense that (3.17)
holds as the trace and (3.18) holds for any w H
1
(C
Q
) with w = 0 on C
Q
{x
n
= Q}. The
solution u is C
1,
(C
Q
) and satises
u
H
1
(C
Q
)
+|u|
(n/2)
1,,C
Q
C(f
L
2
(C
Q
)
+|f|
(2+n/2)
0,0,C
Q
+
L
2
(C
Q
)
+||
(1+n/2)
0,,C
Q
+g
1
0,,
1
Q
+g
2
L
2
(
2
Q
)
+|g
2
|
(1+n/2)
0,,(
2
Q
)
+h
H
1
(C
Q
)
) (3.21)
and
|u|
(n/2)
1,,C
Q
C(u
L
2
(C
Q
)
+|f|
(2+n/2)
0,0,C
Q
+||
(1+n/2)
0,,C
Q
+g
1
0,,
1
Q
+|g
2
|
(1+n/2)
0,,(
2
Q
)
), (3.22)
where C depends only on n, , Q, and .
Proof. We prove this lemma in six steps. The constant C in the argument below is a universal
constant depending only on n, , Q, and , and may be dierent at each occurrence.
Step 1. First we show the uniqueness of weak solutions. Suppose that u
1
and u
2
are two
weak solutions of (3.14)(3.17). Then u
1
u
2
= 0 on
3
Q
. Thus, writing the weak forms (3.18)
for u
1
and u
2
respectively and subtracting the expressions with w = u
1
u
2
yield
0 =
_
C
Q
n
i,j=1
i
|
x
i
(u
1
u
2
)|
2
dx.
Since u
1
u
2
= 0 on
3
Q
and
i
> 0, we conclude u
1
= u
2
a.e. in C
Q
.
Step 2. We now focus on the existence of solutions and estimate (3.21). First we show that the
problem can be reduced to the case where the right-hand side of the boundary condition (3.15)
vanishes. We can extend g
1
from
1
Q
= {0} to the whole hyperplane {(x
, 0) : x
R
n1
}
such that the extension g
1
C
0
(R
n1
) satises g
0,,R
n1 Cg
1
0,,
. Now, as in [20, page
124-125], we choose a nonnegative C
2
0
(R
n1
) with
_
R
n1
(y
)dy
= 1 and dene
G(x
, x
n
) =
1
n
x
n
_
R
n1
g
1
(x
x
n
y
)(y
)dy
for x
n
> 0.
Then G C
1,
(R
n
+
) with
G
1,,C
Q
Cg
0,,R
n1 Cg
1
0,,
(3.23)
and
n
G
x
n
= g
1
= g
1
on {x
n
= 0}.
Thus, u is a weak solution of (3.14)(3.17) if and only if v = uG is a weak solution of the same
problem with modied right-hand sides: f, , g
1
, and g
2
are replaced by

f = f,
i
=
i
i
G
x
i
,
g
1
= 0, and g
2
= g
2
. Using (3.23), it is easy to check that (3.21)(3.22) for v in terms of

f,

, g
1
,
and g
2
implies (3.21)(3.22) for u in terms of f, , g
1
, and g
2
.
Thus, from now on, we assume
g
1
= 0 on
1
Q
. (3.24)
Step 3. We also note that we can assume without loss of generality that
n
= 0 on
1
Q
. (3.25)
Indeed, if u is a weak solution of (3.14)(3.17), then u is also a weak solution of (3.14)(3.17)
with
n
replaced by

n
:

n
(x
, x
n
) =
n
(x
, x
n
)
n
(x
, 0) for (x
, x
n
) C
Q
. Since
L
2
(C
Q
)
+|
|
(1+n/2)
0,,C
Q
C(
n
L
2
(C
Q
)
+||
(1+n/2)
0,,C
Q
),
it follows that (3.21)(3.22) in terms of the right-hand sides with
n
replaced by

n
implies
(3.21)(3.22) in terms of the original functions.
Clearly,

n
satises (3.25). Thus we can assume (3.25) from now on.
Step 4. With (3.24)(3.25), we extend problem (3.14)(3.17) to the cylinder C
(Q,Q)
=
(Q, Q). Set
2
(Q,Q)
:= (Q, Q).
Note that, with (3.24)(3.25), u is a weak solution of (3.14)(3.17) if (3.17) holds as the trace
and, for any w H
1
(C
Q
) with w = 0 on C
Q
{x
n
= Q},
_
C
Q
(
n
i=1
i
u
x
i
w
x
i

n
i=1
i
w
x
i
+ fw)dx
_
2
Q
g
2
wdH
n1
= 0. (3.26)
We use the even reection to extend u, f, h, g
2
,
1
, . . . ,
n1
to C
(Q,Q)
and the odd reection
to extend
n
to C
(Q,Q)
. That is, for x
and x
n
(0, Q), we use the same notations for the
original and extended functions to dene
(x
, x
n
) = (x
, x
n
) for {u, f, h, g
2
,
1
, . . . ,
n1
},
n
(x
, x
n
) =
n
(x
, x
n
).
(3.27)
Obviously, for u H
1
(C
Q
) and f, , h, and g
2
as in Lemma 3.1 with (3.25) for
n
, the extended
functions satisfy
u H
1
(C
(Q,Q)
), f L
2
(C
(Q,Q)
) L
loc
(C
(Q,Q)
), H
1
(C
(Q,Q)
) C
(C
(Q,Q)
),
h H
1
(C
(Q,Q)
), g
2
C
(
2
(Q,Q)
) L
2
(
2
(Q,Q)
),
where C
(Q,Q)
= C
(Q,Q)

2
(Q,Q)
. Moreover, dening the weighted Holder semi-norms and
norms | |
(k)
m;;C
(Q,Q)
and | |
(k)
m;;
2
(Q,Q)
by (3.10) with the supremums taken over the domains
C
(Q,Q)
and
2
(Q,Q)
respectively and
x
= dist(x, C
(Q,Q)
\
2
(Q,Q)
) = min(Qx
n
, Q + x
n
) for x = (x
, x
n
) C
(Q,Q)
,
we obtain
|f|
(2+n/2)
0,0,C
(Q,Q)
C|f|
(2+n/2)
0,0,C
Q
, ||
(1+n/2)
0,,C
(Q,Q)
C||
(1+n/2)
0,,C
Q
,
|g
2
|
(1+n/2)
0,,
2
(Q,Q)
C|g
2
|
(1+n/2)
0,,C
Q
,
(3.28)
where the norms on the right-hand sides are nite by (3.20).
Then it follows from (3.26)(3.27) that the extended functions satisfy
_
C
(Q,Q)
_
n
i=1
i
u
x
i
w
x
i

n
i=1
i
w
x
i
+ fw
_
dx
_
2
(Q,Q)
g
2
wdH
n1
= 0 (3.29)
for any w H
1
(C
(Q,Q)
) satisfying w = 0 on {x
n
= Q}. Indeed, (3.29) holds separately when
the integration is over subdomains {x
n
> 0} and {x
n
< 0} (the last follows by the change of
variable x
n
x
n
). Thus, the extended function u is a weak solution of
n
i=1
i
u
x
i
x
i
= f +
n
i=1
i
x
i
in C
(Q,Q)
, (3.30)
n
i=1
i
u
x
i
i

n1
i=1
i
u
x
i
i
= g
2
+
n1
i=1
i
on
2
(Q,Q)
, (3.31)
u = h on ( {Q}) ( {Q}). (3.32)
Conversely, if u is a weak solution of (3.30)(3.32) (i.e. (3.29) holds for w as above), then u
satises u(x
, x
n
) = u(x
, x
n
) for x
and x
n
(0, Q). To see this, we rst note that a weak
solution of (3.29) is unique: The proof similarly follows the proof of uniqueness for (3.1)(3.3)
by using the fact that, if u
1
and u
2
are two weak solutions of (3.30)(3.32), then we can use
w = u
1
u
2
in (3.29). Now, the properties in (3.27) for f, h, , and g
2
imply that, if u(x
, x
n
) is a
solution of (3.29), then u(x
, x
n
) is also a solution. Thus, by uniqueness, u(x
, x
n
) = u(x
, x
n
).
Now, let w H
1
(C
Q
) with w = 0 on {x
n
= Q}. Extend w by the even reection to C
(Q,Q)
. We
can use this extended w in (3.29). Then, using the properties in (3.27) for u, f, h, , and g
2
, we
see that, for all the integrals in (3.29), the integrals over subdomains {x
n
> 0} and {x
n
< 0}
are equal. Thus, (3.29) implies (3.26). Then, in order to solve (3.14)(3.17), it suces to solve
(3.30)(3.32) for the extended right-hand sides.
Step 5. We now derive some estimates for the weak solution u H
1
(C
(Q,Q)
) of (3.30)(3.32).
We rst derive the energy estimate. We can use w = u h in (3.29) and then use the ellipticity
and the Holder inequality to obtain
u
H
1
(C
(Q,Q)
)
C((f, )
L
2
(C
(Q,Q)
)
+h
H
1
(C
(Q,Q)
)
+g
2
L
2
(
2
(Q,Q)
)
). (3.33)
Note that, by the standard local regularity results for the conormal problem for elliptic equa-
tions in domains with C
1,
boundary (e.g., [35]), we have u C
1,
(C
(Q,Q)
).
Now we derive the estimates for the Holder norms of u. We rst recall the standard estimates
for the conormal problem for elliptic equations in a unit ball B
1
= B
1
(0) and a half-ball B
+
1
=
B
1
(0) {x
n
> 0}. If u H
1
(B
1
) is a weak solution of the equation
n
i,j=1
(a
ij
u
x
j
)
x
i
= f +
n
i=1
i
x
i
in B
1
(3.34)
with a
ij
C
(B
1
) satisfying the ellipticity condition: there exists some > 0 such that, for any
x B
1
,
||
2
i,j=1
a
ij
(x)
i
j

1
||
2
for any R
n
, (3.35)
and, if f L
(B
1
) and C
(B
1
), then u C
1,
(B
1/2
) with
u
1,,B
1/2
C(u
L
2
(B
1
)
+f
L
(B
1
)
+
0,,B
1
), (3.36)
where C depends only on n, , and a
ij
0,,B
1
. This follows from [20, Theorem 8.32], combined
with [20, Theorem 8.17]. Similar estimates can be obtained for the conormal problem, in which
u H
1
(B
+
1
) is a weak solution of (3.34) in B
+
1
with the conormal condition
n
j=1
a
nj
u
x
j
= g +
n
on
B
+
1
:= B
+
1
{x
n
= 0}, (3.37)
where g C
(B
+
). Then u C
1,
(B
+
1/2
) with
u
1,,B
+
1/2
C(u
L
2
(B
+
1
)
+f
L
(B
+
1
)
+
0,,B
+
1
+g
0,,
B
+
1
), (3.38)
where C depends only on n, , and a
ij
0,,B
+
1
. This can be inferred by simplifying the estimates
of [35] to the case of linear elliptic equations.
Let (0, 1/2). By scaling, we obtain the corresponding local estimates for weak solutions of
(3.34) in B
and for weak solutions of the conormal problem (3.34) and (3.37) in B
+
. We write
only the estimate for the conormal problem in B
+
, since the interior estimate in B
diers only
in that it does not have the terms with the norms of g on the right-hand side. Thus, if u is a
weak solution of (3.34) in B
+
/2
and satises (3.37) in B
+
{x
n
= 0}, then u C
1,
(B
+
/2
) with
u
0,0,B
+
/2
+ Du
0,0,B
+
/2
+
1+
[Du]
0,,B
+
/2
C(
n/2
u
L
2
(B
+
)
+
2
f
L
(B
+
)
+
0,0,B
+
+
1+
[]
0,,B
+
+ g
0,0,
B
+
+
1+
[g]
0,,
B
+
).
(3.39)
This estimate is obtained by introducing the function v(x) = u(x) in B
+
1
, writing the equa-
tion and the conormal condition satised by u in terms of v, using estimate (3.38) for v with
corresponding right-hand sides of the equation and conormal condition satised by v, and then
rewriting that estimate in terms of u.
Now we get similar estimates for weak solutions of (3.30)(3.32). Using that is C
1,
and compact, there exits r > 0 and M > 0 such that, for any z
2
(Q,Q)
and any positive
< min(r,
z
), there is a C
1,
dieomorphism F that attens
2
(Q,Q)
in B
(z) C
(Q,Q)
so that
B
+
/M
F
_
B
(z) C
(Q,Q)
_
B
+
M
and (F, F
1
)
C
1, depends only on . The transformed
function v(y) := u(F
1
(y)) satises a conormal problem of the form (3.34) and (3.37) in B
+
/M
,
where the equation is elliptic (with ellipticity constant /2 and the C
norms of the coecients

a
ij
depending only on the C
1,
norms of ) and the appropriate norms of right-hand sides of
the original and transformed problems are estimated by one in terms of another with a constant
depending only on the C
1,
norms of . This can be seen by choosing the function w supported
in B
(z) in (3.29) and changing the variables x y = F(x) in (3.29). Now we obtain estimate
(3.39) for v, which implies the following estimate for u:
u
0,0,
/2
(z)
+ Du
0,0,
/2
(z)
+
1+
[Du]
0,,
/2
(z)
C(
n/2
u
L
2
(
(z))
+
2
f
L
(z))
+
0,0,
(z)
+
1+
[]
0,,
(z)
+g
2
0,0,
(z)
+
1+
[g
2
]
0,,
(z)
)
(3.40)
for any z
2
(Q,Q)
, 0 < < min(r,
z
), with C depending only on n, , L, and , where
(z) := B
(z) C
(Q,Q)
and
(z) := B
(z)
2
(Q,Q)
. We also have the corresponding
interior estimates: For any z C
(Q,Q)
and 0 < < dist(z, C
(Q,Q)
), we obtain (3.40) for
(z) = B
(z) without the terms involving the norms of g

2
on the right-hand side. Multiplying
the interior and boundary estimates (3.40) by
n/2
and using a standard argument (e.g., [20,
Theorem 4.8]) yields
|u|
(n/2)
1,,C
(Q,Q)
C(u
L
2
(C
(Q,Q)
)
+|f|
(2+n/2)
0,0,C
(Q,Q)
+||
(1+n/2)
0,,C
(Q,Q)
+|g
2
|
(1+n/2)
0,,
2
(Q,Q)
). (3.41)
Obviously, estimates (3.33) and (3.41) imply (3.21). Also, (3.41) implies (3.22).
Step 6. It remains to prove the existence of a weak solution u H
1
(C
(Q,Q)
) of (3.30)(3.32).
We rst assume that f, C
(C
(Q,Q)
) and g
2
C
(
2
(Q,Q)
). Then, since
2
(Q,Q)
is a C
2,
surface and g
2
+
n1
i=1

i
i
C
1,
(
2
(Q,Q)
), we can nd G C
2,
(C
(Q,Q)
) such that
n1
i=1
i
G
x
i
i
= g
2
+
n1
i=1
i
on
2
(Q,Q)
. (3.42)
To construct such a function G, we rst extend the right-hand side of (3.42) to
2
(2Q,2Q)
=
(2Q, 2Q) so that g C
1,
(
2
(2Q,2Q)
) with g = g
2
+
n1
i=1

i
i
on
2
(Q,Q)
. Then, for
every x
(3Q/2, 3Q/2), we can locally atten

2
(2Q,2Q)
by a C
2,
dieomorphism
: R
n
R
n
, i.e., for some r > 0, (B
r
(x
) ( R)) = (B
r
(x
)) {x
n
> 0} with
(B
r
(x
) ( R)) = (B
r
(x
)) {x
n
= 0}. Then condition (3.42) is transformed into
n
i=1
c
i
(x
)

G
x
i
= g on {x
n
= 0}, (3.43)
where c
i
C
1,
(R
n1
) with c
n
min(
1
, . . . ,
n
)/2 > 0 if r is chosen suciently small, and g
C
1,
(R
n1
) with compact support is obtained by transforming g by and extending to R
n1
.
Now, by the argument similar to [20, page 124-125], we choose a nonnegative C
2
0
(R
n1
) with
_
R
n1
(y
)dy
= 1, dene the function
G(x
, x
n
) = x
n
_
R
n1
g(x
x
n
y
)
c
n
(x
x
n
y
)
(y
)dy
,
and show that

G is C
2,
(R
n
+
) and satises (3.43). Transforming

G back by
1
, we obtain
G G
x
C
2,
(( R) B
r
(x
)) satisfying (3.42) and G 0 on

2
(2Q,2Q)
B
r
(x
). Gluing
these local functions by using a partition of unity, and taking into account that the local functions
G vanish on
2
(2Q,2Q)
, we obtain a C
2,
function G in a neighborhood of the boundary part
2
(Q,Q)
in C, which satises (3.42) on
2
(Q,Q)
. Finally, we extend G to the whole cylinder
C
(Q,Q)
with G C
2,
(C
(Q,Q)
).
Now, denoting

f = f +
n
i=1
(
i
x
i

i
G
x
i
x
i
), we have

f C(C
(Q,Q)
). Thus we can solve the
variational problem of minimizing
I[v] =
_
C
(Q,Q)
(
1
2
n
i=1
i
v
2
x
i
+

fv)dx
on the set {v H
1
(C
(Q,Q)
) : v = h G on {x
n
= Q}}. If v is a minimizer, then
u = v + G is a weak solution of (3.30)(3.32).
For general f, L
2
(C
(Q,Q)
) and g
2
L
2
(
2
(Q,Q)
), we approximate by f
l
,
l
C
(C
(Q,Q)
)
and g
l
2
C
(
2
(Q,Q)
) such that (f
l
,
l
) (f, ) in L
2
(C
(Q,Q)
) and g
l
2
g
2
in L
2
(
2
(Q,Q)
).
Then, for each l, we can nd a weak solution u
l
H
1
(C
(Q,Q)
) of (3.30)(3.32) with functions
f
l
,
l
, and g
l
2
on the right-hand side. Since u
k
u
l
satises (3.30)(3.32) with f
k
f
l
,
k

l
,
g
k
2
g
l
2
, and h = 0 on the right-hand side, we apply (3.33) to see that {u
l
} is a Cauchy sequence
in H
1
(C
(Q,Q)
) and its limit u is a solution of (3.30)(3.32) with the original right-hand sides.
Now Lemma 3.1 is proved.
Proof of Theorem 3.2. If is a small constant depending only on , then (3.4) implies the
ellipticity (3.35) with
3
4
instead of . We choose such below.
First we show the uniqueness of weak solutions. Suppose that u
1
and u
2
are two weak solutions
of (3.1)(3.3). Then u
1
u
2
= 0 on
3
Q
. Writing the weak form (3.7) for u
1
and u
2
respectively,
subtracting the expressions with w = u
1
u
2
, and then using the ellipticity yields
0 =
_
C
Q
n
i,j=1
a
ij
x
i
(u
1
u
2
)
x
j
(u
1
u
2
)dx
3
4
_
C
Q
|D(u
1
u
2
)|
2
dx.
Since u
1
u
2
= 0 on
3
Q
, we conclude u
1
= u
2
a.e. in C
Q
.
We now apply the Banach contraction xed point theorem to prove the existence and regularity
of the weak solution u H
1
(C
Q
) of (3.1)(3.3) for suciently small > 0 in the space:
K = {v H
1
(C
Q
) C
1,
(C
Q
) : v
K
:= v
H
1
(C
Q
)
+|v|
(n/2)
1,,C
Q
< }.
We now dene a mapping J : K K. For v K, consider problem (3.14)(3.17) with functions
f,

h, g
1
, g
2
, and

on the right-hand sides, where
f = f,

h = h, g
2
= g
2
, (3.44)
g
1
(x) = g
1
(x) +
n
i,j=1
(
i
j
i
a
ij
(x))v
x
j
(x)
i
(x) for x
1
Q
= {0}, (3.45)
i
(x) =
n
j=1
(
i
j
i
a
ij
(x))v
x
j
(x) for x C
Q
, i = 1, . . . , n, (3.46)
with f, g
1
, g
2
, and h from (3.1)(3.3). Since = e
n
on
1
Q
, we have
g
1
= g
1
+
n
j=1
(
n
j
n
a
nj
)v
x
j
g
1
+

n
on x
1
Q
. (3.47)
By (3.4) and v K, the functions g
1
and satisfy the conditions of Lemma 3.1. Thus, by Lemma
3.1, there exists a unique weak solution u of (3.14)(3.17) with the right-hand sides described
above, which satises (3.21) and thus implies u K. We dene the mapping J : K K by
setting Jv = u.
Now we show that J is a contraction mapping in the norm
K
if > 0 is small. Let
v
1
, v
2
K and u
k
= Jv
k
for k = 1, 2. Then u
1
u
2
is a weak solution of (3.14)(3.17) with the
following functions

f,

h, g
1
, g
2
, and

on the right-hand sides:
f = 0, g
2
= 0,

h = 0,
i
(x) =
n
j=1
(
i
j
i
a
ij
(x))(v
1
x
j
(x) v
2
x
j
(x)) for x C
Q
, i = 1, . . . , n,
g
1
(x) =
n
j=1
(
n
j
n
a
nj
(x))(v
1
x
j
(x) v
2
x
j
(x)) for x
1
Q
.
By (3.4), we have
L
2
(C
Q
)
CDv
1
Dv
2
L
2
(C
Q
)
Cv
1
v
2
K
,
|
|
(1+n/2)
0,,C
Q
C|Dv
1
Dv
2
|
(1+n/2)
0,,C
Q
C|v
1
v
2
|
(n/2)
1,,C
Q
Cv
1
v
2
K
,
g
1
0,,
1
Q
CDv
1
Dv
2
0,,
1
Q
C|Dv
1
Dv
2
|
(1+n/2)
0,,C
Q
C|v
1
v
2
|
(n/2)
1,,C
Q
Cv
1
v
2
K
.
Thus, from (3.21),
u
1
u
2
K
Cv
1
v
2
K
.
Therefore, the mapping J : K K is a contraction mapping in the norm
K
if < 1/C, i.e.,
is small depending only on n, , Q, and .
For such , there exists a xed point u K satisfying Ju = u. Then u is a weak solution of
(3.1)(3.3). Indeed, since Ju = u, then u satises (3.18) with right-hand sides given by (3.44)
(3.46) computed with v = u. Rearranging and taking into account the last expression in (3.47)
yields (3.7). Also, since u satises (3.18) with right-hand sides given by (3.44)(3.46), u satises
(3.22) with these right-hand sides, which implies
|u|
(n/2)
1,,C
Q
C(u
L
2
(C
Q
)
+|f|
(2+n/2)
0,0,C
Q
+g
1
0,,
1
Q
+|g
2
|
(1+n/2)
0,,(
2
Q
)
) + C|u|
(n/2)
1,,C
Q
,
by estimating

and g
1
similar to the estimates of

and g
1
above and with v = u. If is small,
this implies (3.13). Then Theorem 3.2 is proved.
Proposition 3.2. Let
1
, . . . ,
n
, and be as in Theorem 3.2. Let L > 0. Then there exist
> 0, depending only on n, , , and L, such that for any Q L, if a
ij
(x) satisfy (3.4) and
f, g
1
, g
2
, h satisfy (3.5), (3.6), there exists a unique weak solution u H
1
(C
Q
) of (3.1)(3.3).
Proof. By Theorem 3.2, if > 0 is small depending only on n, , , and L, there exists u
1

H
1
(C
L
) C
1,
(C
L
) which is a weak solution of (3.1) in C
L
, (3.2) on
k
L
, and u
1
= 0 on
3
L
, in
the sense of (3.7) with Q replaced by L. Then we can extend u
1
from C
L/2
to C
Q
so that the
extended function u
1
satises u
1
C
1,
(C
Q
) and u
1
= u
1
on C
L/2
.
Denote
g
2
:= g
2
i,j=1
a
ij
( u
1
)
x
j
i
on
2
Q
.
Then g
2
C
0,
(
2
Q
), and g
2
0 on
2
L/2
. Now, repeating the construction of Step 6 of proof of
Lemma 3.1, we show existence of G C
1,
(C
Q
) satisfying
n
i,j=1
a
ij
G
x
j
i
= g
2
and G = 0 on
2
Q
. (3.48)
Also, choosing by C
(R
1
) satisfying 0 on (, L/4) and 1 on (L/2, ), and using
that g
2
0 on
2
L/2
and G 0 on
2
Q
, we see that the function

G(x) := (x
n
)G(x) also satises
(3.48). Thus, replacing G by

G, we get a function G C
1,
(C
Q
) satisfying (3.48) and
n
i,j=1
a
ij
G
x
j
i
= 0 on
1
Q
.
Now, for f,
i
C
0,
(C
Q
) for i = 1, . . . , n, and

h H
1
(C
Q
), we construct a weak solution
v H
1
(C
Q
) of the problem
n
i,j=1
(a
ij
v
x
j
)
x
i
= f +
n
i=1
i
x
i
in C
Q
,
n
i,j=1
a
ij
v
x
j
i
= 0 on
1
Q
,
2
Q
, v =

h on
3
Q
in the sense that v =

h on
3
Q
holds as the trace and
_
C
Q
(
n
i,j=1
a
ij
v
x
j
w
x
i

n
i=1
i
w
x
i
+ fw)dx +
2
k=1
_
k
Q
n
i=1
i
wdH
n1
= 0 (3.49)
1
(C
Q
) with w = 0 on
3
Q
. First we assume
i
C
1,
(C
Q
) for i = 1, . . . , n.
Then, denoting

f = f +
n
i=1

i
x
i
, we have

f C(C
Q
). Then we nd a weak solution v of (3.49)
as a minimizer of
I[v] =
_
C
Q
(
1
2
n
i,i=1
a
ij
v
x
i
v
x
j
+

fv)dx
on the set {v H
1
(C
Q
) : v =

h on
3
Q
}. In general case, assuming that
i
C
0,
(C
Q
) and
using w = v h in (3.49), and using ellipticity, we obtain
v
H
1
(C
(Q,Q)
)
C(f
L
2
(C
(Q,Q)
)
+
C
0
(C
(Q,Q)
)
+h
H
1
(C
(Q,Q)
)
), (3.50)
where C depends only on n, , Q, , . Then approximating = (
1
, . . . ,
n
) by
l
C
1,
(C
Q
)
such that
l
in C
0,
(C
Q
), denoting by v
l
a weak solution of (3.49) with
l
instead of ,
and using (3.50) and the linearity of the problem, we nd that {v
l
}
l=1
is a Cauchy sequence in
H
1
(C
Q
), and its limit is a weak solution of (3.49) with C
0,
(C
Q
).
Now let v H
1
(C
Q
) be a weak solution of (3.49) with f as in Proposition, and
i
=
n
j=1
a
ij
( u
1
+ G)
x
j
,

h = h u
1
G,
where h is as in Proposition. Note that
i
C
0,
(C
Q
) and

h H
1
(C
(Q,Q)
), thus such v exists,
Then function u = u
1
+ G + v is a weak solution of (3.1)(3.3)
Next we prove some regularity properties of weak solutions near
3
Q
, more specically near
the wedge W := {x
n
= Q}. For a < b, denote C
(a,b)
= (a, b) and
2
(a,b)
= (a, b).
Proposition 3.3. Let > 0 and b > 0 be constants. Let the coecients a
ij
(x) satisfy the
ellipticity condition (3.35) and let a
ij
0,,(Qb,Q)

1
. Let u H
1
(C
(Qb,Q)
) be a weak
solution of the conormal problem
n
i,j=1
(a
ij
u
x
j
)
x
i
= 0 in C
(Qb,Q)
,
n
i,j=1
a
ij
u
x
j
i
= 0 on
2
(Qb,Q)
, (3.51)
u = 0 on {x
n
= Q} (3.52)
in the sense that (3.52) holds as the trace and
_
C
(Qb,Q)
n
i,j=1
a
ij
u
x
j
w
x
i
dx = 0 (3.53)
1
(C
(Qb,Q)
) with w = 0 on C
(Qb,Q)
\
2
(Qb,Q)
. Then there exists
(0, 1) and C > 0, depending only on the data (i.e., on n, , b, , , ) such that u
C
1,
(C
(Qb/2,Q)
\ W) and
|Du(x)| Cu
L
2
((Qb,Q))
(dist(x, W))
1
for any x C
(Qb/2,Q)
, (3.54)
where W := {x
n
= Q}.
Proof. The regularity u C
1,
(C
(Qb/2,Q)
\ W) follows from the standard interior estimates for
linear elliptic equations and the estimates for the Dirichlet and conormal problems. Thus it
suces to prove (3.54).
Denote
D
(Qb,Q)
:= C
(Qb,Q)
\
2
(Qb,Q)
.
We rst note that any weak solution u of (3.51) (without requiring (3.52)) satises the following
local estimates: For any B
2R
= B
2R
(y) R
n
with R (0, b/10), denote M = sup
B
2R
D
(Qb,Q)
u
+
and m = inf
B
2R
D
(Qb,Q)
u, where in the case B
2R

D
(Qb,Q)
= we set m = and M = .
Dene the functions u
+
M
and u
m
in C
(,)
by
u
+
M
(x) :=
_
sup(u(x), M), x C
(Qb,Q)
,
M, x C
(,)
\ C
(Qb,Q)
,
u
m
(x) :=
_
inf(u(x), m), x C
(Qb,Q)
,
m, x C
(,)
\ C
(Qb,Q)
.
Then, for any p > 1,
sup
B
R
(y)C
(,)
u
+
M
CR
n/p
u
+
M
L
p
(B
2R
(y)C
(,)
)
.
If u is nonnegative in B
4R
(y) C
(Qb,Q)
, then, for any 1 p n/(n 2), the weak Harnack
inequality holds:
inf
B
R
(y)C
(,)
u
m
CR
n/p
u
L
p
(B
2R
(y)C
(,)
)
,
where the constant C depends only on , b, , and p. The proof follows the proofs of [20, Thm
8.17, 8.18, 8.25, 8.26] since, in the weak formulation (3.53), we can use the same test functions
as in [20, Thm 8.17, 8.18, 8.25, 8.26] and obtain the same expressions in the result, with the only
dierence that we integrate over C
(,)
instead of R
n
. Also, in our case, the test functions
may be nonzero on the boundary part
2
(Qb,Q)
, and thus we use the Sobolev inequalities for the
functions that vanish only on a part of the boundary (cf. [34, Thm. 6.37]).
Once the weak Harnack inequality is obtained, we follow the proofs of [20, Thm 8.22, 8.27]
to obtain the existence of (0, ) and C > 0 depending only on the data such that, if u is a
weak solution of (3.51)(3.52), then
|u(x)| Cu
L
(C
(Q3b/4,Q)
)
|Qx
n
|
for any x C
(Qb/2,Q)
.
We also obtain u
L
(C
(Q3b/4,Q)
)
Cu
L
2
((Qb,Q))
by applying the local estimates to u
with p = 2 and using (3.52). Then, using 0 Qx
n
dist(x, W) for x C
(Qb/2,Q)
, we have
|u(x)| Cu
L
2
((Qb,Q))
(dist(x, W))
for all x C
(Qb/2,Q)
. (3.55)
Now we follow the argument of Step 5 of the proof of Lemma 3.1. For x R
n
and > 0,
denote
(x) := C
(Qb,Q)
B
(x). Let x {x
n
= Q} and = min(b, dist(x, W))/10. Then
B
(x) C
(Qb,Q)
{x
n
= Q}. Thus, applying the standard estimates for the Dirichlet
problem [20, Thm 8.32, 8.33], properly scaled, we obtain
u
0,0,
/2
(x)
+ Du
0,0,
/2
(x)
+
1+
[Du]
0,,
/2
(x)
C
n/2
u
L
2
(
(x))
, (3.56)
where C depends only on the data. Now let x
2
(Qb/2,Q)
and = min(b, dist(x, W))/10. Then
B
(x) C
(Qb,Q)

2
(Qb,Q)
. Thus the proof of (3.40) yields (3.56) in the present case. Finally,
if x C
(Qb/2,Q)
, = min(b, dist(x, W))/40, and B
(x) C
(Qb,Q)
, then the interior estimates
[20, Thm 8.32] also yield (3.56). Note that, if z C
(Qb/2,Q)
\ W, 0 < r min(b, dist(x, W))/10,
and y
r
(z), then
1
2
dist(z, W) dist(y, W) 2dist(z, W). Then (3.55) and (3.56) imply
Du
0,0,
/2
(x)
Cu
L
2
((Qb,Q))
(dist(x, W))
1
(3.57)
for all choices of (x, ) discussed above. Moreover, the sets
/2
(x), with (x, ) specied above,
cover C
(Qb/2,Q)
. Then we use again that
1
2
dist(x, W) dist(y, W) 2dist(x, W) for all y
/2
(x) to have (3.54).
4. Free boundary problems in the infinite nozzle
In order to solve Problem (FB), we rst reformulate it into a free boundary problem for the
subsonic part of the solution, since Problem (FB) is originally hyperbolic-elliptic mixed.
We rst modify equation (1.1) to make it uniformly elliptic so that it coincides with the original
equation in the range of D in the subsonic region
+
for satisfying (2.31) with suciently
small . The details of the truncation procedure are in [4, Section 4.2].
Let = (c
q
+
0
)/2. Then there exist C
1,1
([0, )) and c
j
> 0, j = 1, 2, 3, depending only
on q
+
0
and , such that
(q
2
) = (q
2
) if 0 q < c
, (4.1)
(q
2
) = c
0
+
c
1
q
if q > c
, (4.2)
0 < c
0

_
(q
2
)q
_
c
2
for q (0, ). (4.3)
Then the equation
L := div ( (|D|
2
)D) = 0 (4.4)
is uniformly elliptic, whose ellipticity constants depend only on q
+
0
and .
To formulate the problem into a free boundary problem, the main point is to replace the
pointwise gradient condition {|D(x)| < c
} dening
+
() by a condition stated in terms of
so that our problem is formulated into the framework of free boundary problems.
To do that, we rst need to construct a local C
1,
supersonic solution
in the domain
1
:= {1 < x
n
< 1} of the initial-boundary value problem (2.18)(2.19) for the nonlinear
hyperbolic equation (1.1) such that
0

1,,
1
C
1
(4.5)
when is suciently small. This is ensured by the local existence theorem, Proposition A.1 in
Appendix. By the standard extension argument (see [5] and Section 7.2), we can extend
to
the whole innite nozzle such that
0

1,,
C
0
,
= 0. (4.6)
Then the following heuristic observation motivates our formulation: By (2.13) and q
0
> q
+
0
,
we have
+
(
0
) = {x :
0
(x) <
0
(x)}. Since
is a small C
1,
perturbation of
0
and q
0
> q
+
0
, then we expect that
+
is close to
+
0
in C
1,
(
+
()) so that we can expect that
+
() = {x : (x) <
(x)}.
We also perform the corresponding truncation of the free boundary condition (2.4):
(|D|
2
)
= (|D
|
2
)D
on S. (4.7)
On the right-hand side of (4.7), we use the original function since = on the range of
|D
|
2
. Note that (4.7), with the right-hand side considered as a known function, is the conormal
boundary condition for the uniformly elliptic equation (4.4).
We rst solve the following free boundary problem (TFB), which is a truncated version of
Problem (FB), in which the gradient condition that determines
+
in Problem (FB) is also
replaced by the condition
in with the denition

+
:= { <
} .
Problem (TFB): Truncated Free Boundary Problem. Given a supersonic
upstream ow
C
1,
(
1
) of (1.1) satisfying (4.5) and (2.18)(2.19) for small
> 0 and some constant C
0
, nd C() such that
(i) satises

in (4.8)
and conditions (2.18)(2.19) on the boundary and (2.20) at innity;
(ii) C
1,
(
+
) C
2
(
+
) is a solution of (4.4) in
+
:= { <
} , the
non-coincidence set;
(iii) the free boundary S =
+
is given by the equation x
n
= f(x
) so that
+
= {x
n
> f(x
)} with f C
1,
;
(iv) the free boundary condition (4.7) holds on S.
In the next sections we rst solve Problem (TFB) by an iteration procedure using uniform
estimates at innity and the regularity estimates near the nonsmooth boundary. Finally, we use
an estimate for |D| to conclude that the solution of the truncated problem, Problem (TFB), is
actually a solution of Problem (FB) and thus Problem (TN).
5. Iteration Procedure and Uniform Estimates
We now introduce an iteration procedure to construct approximate solutions of Problem (TFB)
in the domain and make uniform estimates of the solutions on the unbounded domain with
nonsmooth boundary.
5.1. Iteration set. Let M 1. We set
K
M
:=
_
C
1,
() : q
+
x
n
1,,
M
_
, (5.1)
where q
+
is dened by equation (2.26) (see Remark 2.4) and (0, 1). By denition, K
M
is
convex. Now we will show that K
M
is compact in the weighted Holder space C
1,/2
(1)
() as dened
below.
For (0, 1), m N, and k R, we dene the following spaces:
C
m,
(k)
() := {u C
m,
() : u
(k)
m,,
< }, (5.2)
where u
(k)
m;;
= u
(k)
m;0;
+ [u]
(k)
m;;
are the norms with
u
(k)
m;0;
=
0jm
(
||=j
sup
x
_
k
x
|D
u(x)|
_
),
[u]
(k)
m;;
=
||=m
sup
x,y,x=y
_
k
x,y
|D
u(x) D
u(y)|
|x y|
_
,
x
= |x
n
| + 1 for x = (x
, x
n
) D, and
x,y
= min(
x
,
y
) for x, y D.
Lemma 5.1. If > > 0, then K
M
is a compact subset of C
1,
(1)
().
Proof. Let
j
K
M
for j = 1, 2, . . . . By a standard argument, we can extract a subsequence (still
denoted by)
j
, which converges in C
1,
on every compact subset of to the limit C
1,
().
Then q
+
x
n
1,,
M and thus K
M
. It remains to show that
j

(1)
1,,
0 as
j .
Fix 0 < < 1. Then
j
(1)
1,,{x
n
>1/}
(M + q
+
), and the same estimate holds
for . Also, since
j
in C
1,
( {x
n
2/}), there exists j
0
such that, for j > j
0
,
j

(1)
1,,{x
n
2/}
. Then, for j > j
0
, we have
j

(1)
1,,
C, and the assertion is
proved.
5.2. Construction of the iteration scheme. Let K
M
. Since q
0
> q
+
0
, it follows that, if

q
0
q
+
0
C(M + 1)
(5.3)
with large C depending only on n, then (4.5) implies
(
)
x
n
(x)
q
0
q
+
0
2
> 0 in (5.4)
and
> on {x
n
> 1}. Then the set
+
() := { <
} has the form:
+
() = {x
n
> f(x
)} with f
1,,R
n1 CM (5.5)
for C depending only on q
0
q
+
0
. The inward unit normal on S
:= {x
n
= f(x
)} to
+
is
(x) =
D
(x) D(x)
|D
(x) D(x)|
for x S
. (5.6)
By the denition of K
M
and (5.3), formula (5.6) also denes
(x) on
1
and
e
n
0,,
1
CM with C = C(q
+
0
, q
0
). (5.7)
Motivated by (4.7), we dene the function
G
(x) := (|D
(x)|
2
)D
(x)
(x) on
1
. (5.8)
Then we consider the following problem in the domain
+
():
div ( (|D|
2
)D) = 0 in
+
(), (5.9)
(|D|
2
)
= G
(x) on S
:= {x
n
= f(x
)}, (5.10)
= 0 on
l
+
:=
+
()
l
, (5.11)
lim
R
q
+
x
n
(
+
(){x
n
>R})
= 0, (5.12)
show that there exists a unique solution (x), and extend the solution to the whole domain so
that K
M
.
We rewrite problem (5.9)(5.12) in terms of the function u(x) = (x) q
+
x
n
. We rst note
that the boundary condition (5.11) is equivalent to
(|D|
2
)
= 0 on
l
+
.
Now u(x) is a solution of the following problem:
div A(Du) = 0 in
+
(), (5.13)
A(Du) = g
(x) on S
, (5.14)
A(Du) = ((q
+
)
2
)q
+
e
n
on
l
+
, (5.15)
lim
R
u
L
(
+
(){x
n
>R})
= 0, (5.16)
where
A(P) = (|P + q
+
e
n
|
2
)(P + q
+
e
n
) ((q
+
)
2
)q
+
e
n
for P R
n
, (5.17)
g
(x) = G
(x) ((q
+
)
2
)q
+
e
n
. (5.18)
Thus, u(x) satises the uniformly elliptic equation with the same ellipticity constants as in (4.4):
||
2
i,j=1
A
i
P
j
(P)
i
j

1
||
2
for any P, R
n
. (5.19)
Moreover, from the denition of A(P) and (4.1)(4.2), A(P) satises
A(0) = 0, (1 +|P|)|D
P
A
i
P
j
(P)| C. (5.20)
Finally, we state a linear problem corresponding to (5.13)(5.16). Namely, we use (5.20) to
nd that, for i = 1, . . . , n,
A
i
(Du(x)) =
n
j=1
a
ij
(x)u
x
j
(x), a
ij
(x) =
_
1
0
A
i
p
j
(tDu(x))dt.
We replace u q
+
x
n
in the denition of the coecients a
ij
by q
+
x
n
for K
M
to
dene
a
ij
(x) = a
()
ij
(x) =
_
1
0
A
i
p
j
_
t(D(x) q
+
e
n
)
_
dt for x , i, j = 1, . . . , n. (5.21)
The ellipticity (5.19) of A(P) implies that the coecients a
()
ij
(x) satisfy the ellipticity condition
(3.35) for any x .
Also, from (5.17) and (5.21),
a
()
ij
(x) =
_
1
0
_
(|tD(x) + (1 t)q
+
e
n
|
2
)
j
i
(5.22)
+2
(|tD(x) + (1 t)q
+
e
n
|
2
)(t
x
i
(x) + (1 t)q
+
n
i
)(t
x
j
(x) + (1 t)q
+
n
j
)
_
dt
for x , where
j
i
= 1 if i = j and
j
i
= 0 if i = j. In particular, we have a
ij
= a
ji
.
We note that, for

0
= q
+
x
n
, the corresponding coecients a
ij
dened by (5.22) are constants
and satisfy
a
ij
=
i
j
i
for i, j = 1, . . . , n, (5.23)
where
i
are
i
=
_
(q
+
)
2
_
for i = 1, . . . , n 1, and
n
=
(q
+
) for (s) = (s
2
)s. By
(4.1)(4.3), we have

i

1
for i = 1, . . . , n.
Now, for K
M
, we use (5.22) and (4.1)(4.3) to obtain
a
()
ij
a
ij
0,,
CM. (5.24)
Thus, we formulate the following conormal xed boundary elliptic problem:
n
i,j=1
(a
()
ij
u
x
j
)
x
i
= 0 in
+
(), (5.25)
n
i,j=1
a
()
ij
u
x
j
i
= g
(x) on S
, (5.26)
n
i,j=1
a
()
ij
u
x
j
i
= ((q
+
)
2
)q
+
e
n
on
l
+
(), (5.27)
lim
R
u
C(
+
(){x
n
>R})
= 0. (5.28)
Since the coecients are only C
, we can expect to nd only a weak solution u C

1,
(
+
())
of (5.25)(5.28) in the sense that u(x) satises (5.28) and, for any w C
1
c
(R
n
),
_
+
()
n
i,j=1
a
()
ij
u
x
j
w
x
i
dx +
_
S
wdH
n1
(5.29)
+
()
((q
+
)
2
)q
+
e
n
wdH
n1
= 0.
We will determine the iteration map J() = by solving (5.25)(5.28) for u, extending u
from
+
() to so that u +q
+
x
n
K
M
, and dening = u +q
+
x
n
. A xed point of this map
is obviously a solution of Problem (TFB) in . In Sections 67, we prove the existence of such
a map J, as well as its xed point.
6. Fixed Boundary Problems in a bounded domain
+
R
()
In order to nd a solution of (5.25)(5.28) in the unbounded domain
+
(), we rst solve the
corresponding problem in the bounded domain
+
R
() :=
+
() {x
n
< R}, R > L,
and then pass to the limit as R . Thus we consider the following problem:
n
i,j=1
(a
()
ij
u
x
j
)
x
i
= 0 in
+
R
(), (6.1)
n
i,j=1
a
()
ij
u
x
j
i
= g
(x) on S
, (6.2)
n
i,j=1
a
()
ij
u
x
j
i
= ((q
+
)
2
)q
+
e
n
on
l
+
R
(), (6.3)
u = 0 on
+
R
() {x
n
= R}, (6.4)
where a
()
ij
are dened by (5.21) and
l
+
R
() =
l
+

+
R
().
We rst note several properties of g
, which will be used later.

Lemma 6.1. There exists C > 0, depending only on the data and independent of M and R,
such that
g
0,,
1
C. (6.5)
This is proved in [5, (4.48), page 334]. Another property of g
is
Lemma 6.2.
_
S
f
g
dH
n1
= ((q
+
)
2
)q
+
_
+
R
e
n
dH
n1
.
Proof. Note that
0 =
_
R
(x
n
) =
_
{x
n
=R}
e
n
dx
_
S
f
e
n
dH
n1
+
R
e
n
dH
n1
,
where is the inward unit normal. Since = e
n
on {x
n
= R}, then, from the last equality,
we have
_
S
f
e
n
dH
n1
+
_
+
R
e
n
dH
n1
= ||. (6.6)
Also, since G
is dened by (5.8) and
C
1,
( \
+
R
()) is a weak solution of (1.1) in
\
+
R
() and satises the boundary condition (2.19), we nd
_
S
f
G
dH
n1
=
_
(|D
|
2
)D
dH
n1
. (6.7)
Then we have
_
S
f
g
dH
n1
((q
+
)
2
)q
+
_
+
R
e
n
dH
n1
=
_
S
f
G
dH
n1
((q
+
)
2
)q
+
_
_
S
f
e
n
dH
n1
+
_
+
R
e
n
dH
n1
_
=
_
S
f
G
dH
n1
((q
+
)
2
)q
+
|| = 0,
where we used (6.7) and the fact that q
+
satises (2.26)(2.27).
We now x K
M
and write
+
R
for
+
R
() and a
ij
for a
()
ij
to simplify the presentation.
Also, C denotes a universal constant, which may be dierent at each occurrence, and depends
only on the data and is independent of M and R. Furthermore, there exists
0
> 0 depending
only on M and the data such that, when
0
, S
{1/10 < x
n
< 1/10} by (5.5); we always
assume
0
below.
Lemma 6.3. Let (0,
0
) be suciently small, depending only on the data and M. Let
u H
1
(
+
R
), R > L, be a weak solution of (6.1)(6.4) in the sense that
(i) u = 0 on
+
R
{x
n
= R} as the trace;
(ii) For any w H
1
(
+
R
) satisfying w = 0 on
+
R
{x
n
= R} in the sense of traces,
_
+
R
n
i,j=1
a
ij
u
x
j
w
x
i
dx +
_
S
f
g
wdH
n1
+
R
((q
+
)
2
)q
+
e
n
wdH
n1
= 0. (6.8)
Then
Du
L
2
(
+
R
)
C. (6.9)
Proof. We use the approach in the proof of [5, Lemma 4.2]. We rst choose w = u in (6.8) to
obtain
_
+
R
n
i,j=1
a
ij
u
x
j
u
x
i
dx =
_
S
f
g
udH
n1
+
_
+
R
((q
+
)
2
)q
+
e
n
udH
n1
.
Now, if Q is a constant which will be chosen below, we use Lemma 6.2 to obtain
_
+
R
n
i,j=1
a
ij
u
x
j
u
x
i
dx =
_
S
f
g
(u Q) dH
n1
+
L
((q
+
)
2
)q
+
e
n
(u Q) dH
n1
,
where we used that e
n
= 0 on
l
+
R
\
l
+
L
by (2.14) and (2.17).
Choosing Q = (u)
L
:=
1
|
+
L
()|
_
+
L
()
u(x)dx and using the L
2
estimate of the boundary
traces of functions in the Sobolev space H
1
(
+
L
()), (6.5), and | e
n
| C on
l
+
L
by (2.16),
we nd
_
+
R
n
i,j=1
a
ij
u
x
j
u
x
i
dx
_
_
S
f
g
2
dH
n1
_
1/2
_
_
S
f
(u (u)
L
)
2
dH
n1
_
1/2
+ ((q
+
)
2
)q
+
_
_
+
L
| e
n
|
2
dH
n1
_
1/2
_
_
+
L
(u (u)
L
)
2
dH
n1
_
1/2
C
_
_
+
L
()
_
(u (u)
L
)
2
+|Du|
2
_
dx
_
1/2
C
_
_
+
L
()
|Du|
2
dx
_
1/2
,
where, in the last estimate, we used the Poincare inequality in the domain
+
L
(), and thus the
constant C depends only on n, L, the norms in (2.15)(2.16), and f
1,,
. On the other hand,
it follows from (5.5) that if < 1/M, then f
1,,
C with C > 0 independent of M, R, and
K
M
. Now, using (3.35), we have
_
+
R
|Du|
2
dx
1
+
R
n
i,j=1
a
ij
u
x
j
u
x
i
dx
C

_
_
+
R
|Du|
2
dx
_
1/2
.
This completes the proof.
By [35], the weak solution u of (6.1)(6.4) is in C
1,
_
+
R
\ (S
{x
n
= R})
_
, since
l
+
R
is C
1,
.
Lemma 6.4. Let and u be as in Lemma 6.3. Then, for any x
n
(L, R), there exists x

such that u(x
, x
n
) = 0.
Proof. We follow the scheme of the proof [5, Lemma 4.3, Step 2]. Fix z (L, R). Suppose that
there is no x
such that u(x
, z) = 0. Then we can assume u(x
, z) > 0 for all x
, since
the case u(x
, z) < 0 for all x
can be handled similarly. From the compactness of and

continuity of u, there exists a constant > 0 such that
u(x
, z) > 0 for all x
.
Consider the domain D := (z, R). Then, using Proposition 3.3, we get u C
1,
(D \ W),
with W =
l
{x
n
= R}. Moreover, u is a weak solution of
n
i,j=1
(a
ij
u
x
j
)
x
i
= 0 in D,
n
i,j=1
a
ij
u
x
j
i
= 0 on
l
D = D
l
{x
n
< R},
u > 0 on D {x
n
= z},
u = 0 on D {x
n
= R}.
Thus, by the maximum principle, u 0 in D. In particular,
u
= u
x
n
0 on (D {x
n
= R}) \ W. (6.10)
By the strong maximum principle [20, Theorem 8.19], u > 0 in the interior of D. Also, denoting
D
= {x D : dist(x,
l
) > }, where > 0 is xed to be small so that D
= , we have
a
ij
C
0,
(D
/2
), u C
1,
(D
/2
).
Thus, by [5, Lemma A.1],
u
= u
x
n
> 0 on D
{x
n
= R}. (6.11)
Since u = 0 on {x
n
= R} and a
nn
> 0 from the ellipticity, we conclude from (6.10)(6.11)
that
n
i,j=1
a
ij
u
x
j
i
= a
nn
u
x
n
_
_
_
0 on (D {x
n
= R}) \ W,
> 0 on D
{x
n
= R}.
Since a
nn
and u
x
n
are continuous on D
{x
n
= R}, we conclude
_
{x
n
=R}
n
i,j=1
a
ij
u
x
j
i
dH
n1
> 0. (6.12)
On the other hand, we show that
_
{x
n
=R}
n
i,j=1
a
ij
u
x
j
i
dH
n1
= 0, (6.13)
thus contradicting (6.12). To prove (6.13), we choose (L, R), and use in (6.8) the test function
w(x) = (( x
n
)/), where > 0, C
(R) with 0 on R, 1 on (1, ) and 0 on

(, 1/2). Note that w C
(
R
) and w 0 on {x
n
= R}. Thus we use this w in (6.8) and
u C
1,
(D \ W) and send 0+ to obtain
_
{x
n
=}
n
i,j=1
a
nj
u
x
j
dH
n1
+
_
S
f
g
dH
n1
((q
+
)
2
)q
+
e
n
dH
n1
= 0. (6.14)
Since > L and e
n
= 0 on
l
{x
n
> L}, we nd from (6.14) and Lemma 6.2 that
_
{x
n
=}
n
i,j=1
a
nj
u
x
j
dH
n1
= 0. (6.15)
From u C
1,
(D\W) and a
ij
C
(), we have
n
i,j=1
(a
nj
u
x
j
)(x
, )
n
i,j=1
(a
nj
u
x
j
)(x
, R)
as R for every x
. Moreover, from (3.54) of Proposition 3.3, there exist C > 0 and

(0, 1) such that, for each ((R + L)/2, R) and x
, we have
|Du(x
, )| C(dist((x
, ), W))
1
C(dist(x
, ))
1
,
where the last inequality holds since dist(x, W) dist(x
, ) for any x = (x
, x
n
)
R
. Since
(dist(, ))
1
L
1
(), we use the dominated convergence theorem to send to the limit as
R in (6.15) and use that = e
n
on D {x
n
= R} to obtain (6.13).
Thus, we conclude a contradiction with (6.12), which completes the proof.
From (2.14)(2.17), there exist r > 0 and C > 0 with the following properties: For any
x
0

l
{x
n
> 1/2}, there exist an orthonormal coordinate system (y
1
, . . . , y
n
) with y
n
= x
n
and a function : R
n1
R such that
B
10r
(x
0
) = {y
1
> (y
2
, . . . , y
n
)} B
10r
(x
0
),
1,,R
n1 C, D(x
0
) = 0. (6.16)
Proposition 6.5. Let u(x) be as in Lemma 6.3. If is suciently small, depending only on the
data and M, then, for any x
0
= (x
0
1
, . . . , x
0
n
)
+
R10r
with L + 1 x
0
n
< R 10r,
Du
0,,B
r
(x
0
)
+
R
CDu
L
2
(B
4r
(x
0
)
+
R
)
. (6.17)
Proof. We use the fact that u K also satises equation (6.1) and the conormal boundary
conditions (6.2)(6.3) for any constant K R.
When x
0
satises B
2r
(x
0
)
+
R
, we apply [26, Theorem 3.13] to u K and obtain
Du
0,,B
r
(x
0
)
+
R
Cu K
L
2
(B
4r
(x
0
)
+
R
)
. (6.18)
When x
0

l
+
with L+1 x
0
n
< R10r, then, since the right-hand side of (6.3) vanishes
on
l
+
{x
n
> L}, we use (5.24) and (a linear version of) the estimates for the conormal
derivative problem [35], applied to u K, to obtain
Du
0,,B
r
(x
0
)
+
R
Cu K
C
0
(B
2r
(x
0
)
+
R
)
. (6.19)
Now we use the standard estimates [20, Theorem 8.17] for the equations of divergence form
(extended to the case of local estimates near the boundary for the conormal problem, see e.g. [37,
Chapter 4, Section 10] where these modications are indicated in the parabolic case; note that
the Lipschitz regularity of the boundary in our case allows to use the weak form of the conormal
problem) to obtain
u K
C
0
(B
2r
(x
0
)
+
R
)
Cu K
L
2
(B
4r
(x
0
)
+
R
)
.
That is, (6.18) is satised for the case x
0

l
+
R
.
Now choosing
K =
1
|B
4r
(x
0
)
+
R
|
_
B
4r
(x
0
)
+
R
udx,
using the Poincare inequality, and noting that the constant in the Poincare inequality can be
chosen independent of x
0

+
R
by (6.16), we complete the proof.
Proposition 6.6. Let and u(x) be as in Lemma 6.3. Then
u
C
0
(
+
R
)
C.
Proof. Lemma 6.3 and Proposition 6.5 imply that
|Du| C on
+
R
{L + 1 < x
n
< R 10r}.
Combining this with Lemma 6.4, we have
|u| C on
+
R
{x
n
(L + 1, R 10r)}.
To extend this bound to the domains
+
R
{R 10r < x
n
< R} = (R 10r, R) and
+
L+1
,
we note that these domains are of the xed size and structure (the cylinder (a, b) and its
small perturbation, respectively). Note that, on the boundary parts {x
n
= L+1}
+
L+1
and
{x
n
= R 10r} ( (R 10r, R)), we have |u| C, as we proved above. Moreover, the
right-hand sides of (6.2)(6.3) are estimated in L
by C from (2.16) and (6.5), respectively.

Thus we can use the standard estimates [20, Theorem 8.15] for the equations of divergence form
(extended to the case when we have the conormal boundary conditions on a part of the boundary,
see e.g. [37, Chapter 4, Section 10]).
Proposition 6.7. If is suciently small depending only on the data and M, there exists a
unique weak solution u H
1
(
+
R
) of (6.1)(6.4). Furthermore, u C
1,
(
+
R10r
) satises (6.9)
and
u
1,,
+
R10r
C. (6.20)
Proof. Since the map in (2.14) satises (2.16), then, if is suciently small, there exists the
inverse map : R
n
R
n
of . The map satises
Id
2,,R
n C. (6.21)
Then, from (2.14),
(
l
) R. (6.22)
Note that, if is small, then, by (2.16), (4.5), and (5.4), we conclude that the set (
+
()) =
{ <
} ( R) has the form

(
+
()) = {x
n
>

f(x
)} ( R) with
f
1,,R
n1 CM. (6.23)
Now we consider a map
=
1
: R
n
R
n
for
1
(x
, x
n
) = (x
, x
n
(x
n
)

f(x
)), (6.24)
where C
(R) is nonincreasing and satises 1 on (, 1/8) and 0 on (1/4, ).

From this denition, we have
Id
1,,R
n CM.
Thus, for small , the map

is invertible and
1
Id
1,,R
n CM.
In particular, for small ,
Id
1,,R
n +
1
Id
1,,R
n 1/10. (6.25)
Moreover, if is small, then
(0, ) =

+
R
{x
n
< }
_
for > L/4. (6.26)
Dene the function v(y) := u(x) = u(
1
(y)) for y C
R
:= (0, R). Then u is a weak
solution of the conormal problem (6.1)(6.4) if and only if v H
1
(C
R
) is a weak solution of the
corresponding problem of the form (3.1)(3.3) in C
R
. To obtain the expressions of the coecients
a
ij
and the right-hand sides f, g
1
, and g
2
of problem (3.1)(3.2) for v, in terms of the data of
problem (6.1)(6.3) and the map

, we make the change of variables y =

(x) in (6.8). Then
we have
a
ij

i
j
i
0,,C
R
CM for i, j = 1, . . . , n, (6.27)
with
i
dened by (5.23), where we used (5.24) and (6.25). In this case, the right-hand side in
(3.1) is f = 0; to estimate g
1
and g
2
in (3.2), we use (2.16), (6.5), and (6.25) to obtain
g
k
0,,
k
C, k = 1, 2. (6.28)
Also, from (6.4) and (6.26), we see that h = 0 in (3.3) for v.
Now the existence a solution v(y) follows from Proposition 3.2 if > 0 is suciently small
depending only on the data and L, so that the right-hand side of (6.27) is smaller than in
Proposition 3.2.
Thus u(x) := v(
(x)) is a weak solution of (6.1)(6.4).

Since a
ij
satisfy ellipticity condition, uniqueness of weak solution u of problem (6.1)(6.4)
follows from the uniqueness argument in the proof of Theorem 3.2.
Estimate (6.9) for u(x) follows from Lemma 6.3.
It remains to prove (6.20). For any x
+
R
{L/2 x
n
R 10r}, the standard interior
estimates [20, Theorem 8.32] and the estimates for oblique derivative problems [35] imply
u
1,,B
r
(x
0
)
+
R
C(u
C
0
(B
4r
(x
0
)
+
R
)
+ ((q
+
)
2
)q
+
e
n
0,,B
r
(x
0
)
+
R
) C, (6.29)
where we used (2.16) and Proposition 6.6. Thus it suces to obtain an estimate in
+
R
{x
n
L}.
By (6.25)(6.26), it suces to show that
v
1,,(0,L)
C, (6.30)
since the similar estimate then holds for u
1,,
+
R
{x
n
L}
.
To show (6.30), we note that using (6.25)(6.26) and Proposition 6.6 yields
v
L
2
((0,2L))
2u
L
2
(
+
R
{x
n
<2L})
C.
Now estimate (3.8), applied to v in the cylinder (0, 2L), with the use of (6.28), implies (6.30),
if is small depending only on M and the data.
7. Existence of Solutions of the Free Boundary Problem
We now show the existence of solutions of the free boundary problem. In this section, C
denotes a universal constant, which may be dierent at each occurrence and depends only on the
data.
7.1. Solutions of the xed boundary problem in the innite nozzle. We rst have
Proposition 7.1. If is suciently small depending only on the data and M, then there exists
a unique weak solution u C
1,
(
+
()) of (5.25)(5.28). This solution satises
Du
L
2
(
+
())
C, (7.1)
u
1,,
+
()
C, (7.2)
lim
u
1,,{x
n
>}
= 0. (7.3)
Moreover, we have
(i) u = lim
R
u
R
, where u
R
is the unique solution of (6.1)(6.4) for each R > 0 and the
convergence is in C
1,
on any compact subset of
+
() for any 0 < ;
(ii) For every R L, there exists x
such that u(x
, R) = 0.
Proof. We assume that is suciently small to satisfy the conditions stated in Sections 56.
We rst show that any weak solution u C
1,
(
+
()) of (5.25)(5.28) satises (7.3). From
(5.28), for any > 0, there exists > L + 2r such that
u
C({x
n
>})
.
Note that (2.17) implies that the right-hand side of (5.27) vanishes on
l
+
() {x
n
> }.
Then the standard estimates [20, Theorem 6.2, Lemma 6.29] imply that, for x = (x
, x
n
) with
x
n
> L + 2r,
u
1,,B
r
(x)
Cu
0,0,B
2r
(x)
C.
Thus, (7.3) follows.
Now we prove the uniqueness of weak solutions of (5.25)(5.28) in C
1,
(
+
()). Let u, v
C
1,
(
+
()) be two solutions of (5.25)(5.28). Then, for any R > L + diam(), dene
(x) :=
R
(x)(u(x) v(x)),
where
R
(x) = (|x|/R) and C
c
(R
+
) is a nonnegative function such that 1 on [0, 1].
Then C
1
c
(
+
). Using as a test function in (5.29) for u and v respectively and subtracting
them, we have
0 =
_
+
n
i,j=1
a
()
ij
_
(u
x
i
v
x
i
)(u
x
j
v
x
j
)
R
+ (u v)(u
x
j
v
x
j
)
x
i
R
_
dx.
Thus,
_
+
n
i,j=1
a
()
ij
(u
x
i
v
x
i
)(u
x
j
v
x
j
)
R
dx
C
R
_
+
(B
2R
(0)\B
R
(0))
(|Du| +|Dv|)(|u| +|v|)dx.
Note that |
+
(B
2R
(0) \ B
R
(0))| 2R||. Thus, using the ellipticity of a
()
ij
, we nd
+
B
R
(0)
|Du Dv|
2
dx C(u
1,0,\B
R
(0)
+v
1,0,\B
R
(0)
)
for large R. Now we send R and use (7.3) to obtain
+
R
|Du Dv|
2
dx = 0.
Thus, u v = const., which yields u = v by using (7.3).
It remains to prove the existence of a weak solution of (5.25)(5.28) satisfying (7.1)(7.2) and
properties (i)(ii) of Proposition 7.1.
The existence of a unique solution u
R
C
1,
(
+
R10r
) H
1
(
+
R
) of (6.1)(6.4) for any R > L
follows from Proposition 6.7.
Now, let R
j
. Using estimates (6.9) and (6.20) for u
R
, which hold by Proposition
6.7, we can extract from {u
R
j
} a subsequence converging in C
1,/2
on compact subsets of
to a function u C
1,
() that satises (7.1). Indeed, this can be achieved by extracting a
subsequence converging in C
1,/2
(
+
L+1
), a further subsequence converging in C
1,/2
(
+
L+2
),
etc., and by using the diagonal procedure. Obviously, u is a weak solution of (5.25)(5.27), since
we can pass to the limit in (5.29) for any xed w C
1
c
(
+
).
Moreover, this solution u satises property (ii): Indeed, for xed R
0
L, R
j
R
0
for
suciently large j. For such j, by Lemma 6.4, there exists x
j
such that u
R
j
(x
j
, R
0
) = 0.
Then there exists a subsequence of {x
j
} converging to some point x
. Since u
R
j
(, R
0
)
u(, R
0
) uniformly in , we have u(x
, R
0
) = 0. Thus, the solution u satises properties (i)(ii).
Also, since each u
R
satises (6.9) and (6.20) with C independent of R, then (7.1)(7.2) hold.
Now we prove that u satises (5.28). From (7.1), for any > 0, there exists > L such that
Du
L
2
(
+
{x
n
>})
.
Then, by Proposition 6.5 (which holds for u as u
R
),
Du
0,,
+
{x
n
>+1}
C.
From this estimate and property (ii) proved above, we have
u
C
0
(
+
{x
n
>+1})
C.
Now (5.28) is proved.
7.2. Iteration map. We rst dene an extension operator P
, similar to [4, Proposition 4.5]

and [5, Proposition 5.1], to obtain
Proposition 7.2. Let > 0 be suciently small depending only on the data and M, and
satisfy (4.5). Then, for any K

M
, there exists an extension operator P
: C
1,
(
+
())
C
1,
() satisfying the following two properties:
(i) There exists C depending only on n, , q
+
0
, and (but independent of M, R, , and )
such that, for the solution u C
1,
(
+
()) of problem (5.25)(5.28) and = u +q
+
x
n
,
P
q
+
x
n
1,,
C; (7.4)
(ii) Let (0, ). If a sequence
k
K
M
converges to C
1,
() in the norm
(1)
1,,
dened in (5.2), then K
M
; Furthermore, if u
k
C
1,
(
+
(
k
)) and u C
1,
(
+
())
are the solutions of problems (5.25)(5.28) for
k
and , respectively, and if
k
= u
k
+
q
+
x
n
and = u + q
+
x
n
, then P
k
P
in the norm
(1)
1,,
.
Proof. Let K
M
. We rst dene an appropriate extension operator P
: C
1,
(
+
())
C
1,
().
Let C
1,
(
+
(
j
)) and u = q
+
x
n
. Then v := u

1
satises v C
1,
(C
(0,)
) with
C
(a,b)
= (a, b), where

is dened by (6.24).
We rst dene the extension operator E
1
: C
1,
(C
(0,)
) C
1,
(C
(2,)
) for any (0, 1).
Let v C
1,
(C
(0,)
). Dene E
1
v = v in C
(0,)
. For (x
, x
n
) C
(1,0)
, dene
E
1
v(x
, x
n
) =
2
i=1
c
i
v(x
,
x
n
i
),
where c
1
= 3 and c
2
= 4, which are determined by

2
i=1
c
i
_
1
i
_
m
= 1 for m = 0, 1. Note that,
for suciently small , we have

(C
(2,)
) by (6.25). Thus, (E
1
v)

C
1,
().
Finally, we dene
P
()(x) = (E
1
v)

(x) + q
+
x
n
.
Obviously, P
() C
1,
(). Also, since

(
+
()) = C
(0,)
, we have
P
()(x) = v

(x) + q
+
x
n
= u(x) + q
+
x
n
= (x) for x
+
(),
i.e., P
: C
1,
(
+
()) C
1,
() is an extension operator.
Now, using (6.25), we obtain the following estimate as in the proof of [4, Proposition 4.5]:
P
q
+
x
n
1,,
C q
+
x
n
1,,
+
()
. (7.5)
Then, if u and are those dened in assertion (i), we have (7.2) for u = q
+
x
n
by Proposition
7.1, and thus (7.5) implies (7.4). Assertion (i) is proved.
Now we prove assertion (ii). First, by Lemma 5.1, K
M
.
By (7.4) applied to
j
and by Lemma 5.1, there exists a subsequence (still denoted by)
j
(x)
such that
P
j
in C
1,
(1)
() (7.6)
for some C
1,
().
Denote

k
:=
k
and

:= . Then, by (2.16),

k

in C
1,
([1, 1]). Denote by
f
k
(x
) the function from (6.23) for

k
(x). Then, by (2.16), (4.5), and (5.4),

f
k

f in C
1,
().
Now let

k
and

be the maps from (6.24) for

f
k
and

f, respectively. Then

k
and

and their
inverse maps satisfy (6.25), and

k

and

1
k

1
in C
1,
on compact subsets of R
n
.
Note v
k
:= u
k
1
k
C
1,
([0, )). Recall that u
k
is a weak solution of (5.25)(5.28) in the
domain
+
(
k
), with coecients a
(
k
)
ij
dened by (5.22) for
k
and corresponding right-hand
sides (5.18) and (5.27) for
k
. Then, by Proposition 7.1, each u
k
satises (7.1)(7.2). Therefore,
Dv
k
L
2
((0,))
C, (7.7)
v
k
1,,(0,)
C for k = 1, . . . . (7.8)
Moreover, v
k
is a weak solution of a conormal problem
n
i,j=1
(b
(
k
)
ij

x
j
v
k
)
x
i
= 0 in (0, ), (7.9)
n
i,j=1
b
(
k
)
ij

x
j
v
k
i
= g
k
(x) on {0}, (7.10)
n
i,j=1
b
(
k
)
ij

x
j
v
k
i
= h
k
(x) on (0, ), (7.11)
where b
(
k
)
ij
, g
k
, and h
k
are dened by the coecients and right-hand sides of (5.25)(5.27) for
k
and by

1
k
. In particular, since a
(
k
)
ij
a
()
ij
in C
1,
on compact subsets of as k ,
which follows from (5.22), and since

1
k

1
in C
1,
on compact subsets of R
n
), it is easy
to show that, for any compact K [0, ),
b
(
k
)
ij
b
()
ij
in C
1,
(K) and ( g
k
, h
k
) ( g
, h
) in C
(K),
where b
()
ij
, g
, and h
are the coecients and right-hand sides of (7.9)(7.11) for . By (7.8),

there exists a subsequence v
k
m
which converges in C
1,
on the compact subsets of [0, ) to
some v C
1,
( [0, )). Then, choosing w C
1
c
(R
n
) in the weak formulation (5.29) for the
conormal problem (7.9)(7.11):
_
(0,)
n
i,j=1
b
(
k
)
ij
v
x
k
w
x
i
dx +
_
{0}
g
k
wdx
_
(0,)
h
k
wdH
n1
= 0,
we pass to the limit for the subsequence {k
m
} and conclude that v is a weak solution of problem
(7.9)(7.11) with coecients b
()
ij
and right-hand sides g
and h
. Note also that v satises (7.7)

since each v
k
m
satises (7.7). Transforming back, we conclude that u := v

is a weak solution
of problem (5.25)(5.27) for and satises (7.1). Now the proof of Proposition 7.1 implies that u
satises (5.28). Then the uniqueness assertion of Proposition 7.1 implies u = u on
+
(). Thus,
P
() = (E
1
v)

+ q
+
x
n
. (7.12)
Finally,
P
k
m
(
k
m
) = (E
1
v
k
m
)

k
m
+ q
+
x
n
(E
1
v)

+ q
+
x
n
in C
1,
(K)
for any compact K , since (
k
, v
k
m
) (
, v) in C
1,
(K) and E
1
: C
1,
(C
(0,)
) C
1,
(C
(2,)
)
is continuous. From this, using (7.6) and (7.12), we obtain that P
k
m
(
k
m
) P
() in C
1,
(1)
().
Moreover, by the same argument, from any subsequence of P
k
(
k
), we can extract a further
subsequence converging in C
1,
(1)
() to the same limit P
(). Thus the whole sequence P
k
(
k
)
converges to this limit. Proposition 7.2 is proved.
Now we can dene the iteration map J : K
M
C
1,
() by
J := P
(7.13)
with (x) = u(x)+q
+
x
n
, where u C
1,
(
+
()) is the unique solution of problem (5.25)(5.28)
with K
M
. By Proposition 7.2(ii), J is continuous in C
1,
() for any (0, ).
7.3. Existence of solutions of the free boundary problem. Now we can prove the existence
of solutions of the free boundary problem. We denote by both the function in
+
() and
its extension P
.
Choose M to be the constant C in (7.4). Then, for K
M
, we have from Proposition 7.2(i)
that := J K
M
if > 0 is suciently small and depends only on n, , q
+
, and , since
M > 0 is now xed. Thus, (7.13) denes the iteration map J : K
M
K
M
and, from Proposition
7.2(ii), J is continuous on K
M
in C
1,/2
(1)
().
In order to nd a solution of Problem (TFB), we seek a xed point of the map J. We use the
Schauder xed point theorem (cf. [20, Theorem 11.1]) in the following setting:
Let > 0 satisfy the conditions of Proposition 7.2. By Lemma 5.1, the set K
M
is a compact
convex subset of C
1,/2
(1)
(). We have shown that J(K
M
) K
M
and J is continuous in C
1,/2
(1)
().
Then, by the Schauder xed point theorem, J has a xed point K
M
.
If is such a xed point, then
(x) := min(
(x), (x))
is a classical solution of Problem (TFB) and S
is its free boundary.

It follows that is a solution of Problem (FB), thus Problem (TN), provided that is small
enough so that (7.4) implies that |D(x)| < c
, where = (c
q
+
0
)/2. Indeed, then
(4.1) implies that lies in the untruncated region for equation (4.4). Also, |D | < c
on
+
( ) := { (x) <
(x)} and |D | > c
on \
+
since = on
+
( ) and =
on
\
+
.
8. Uniqueness and Stability
We now prove the uniqueness of the velocity at innity in Problem (TN) and, when
C
2,
,
prove the uniqueness and stability of solutions of Problem (TN). We rst prove the following
lemma.
Lemma 8.1. Let
satisfy (4.5). Let C

0,1
() be a solution of Problem (TN) with (0, c
)
satisfying (2.24). Then there exists
0
> 0 depending only on n, , , and q
+
0
such that, if
(0,
0
), satises (2.23), (2.30), and
D e
n
C
0
({x
n
>R})
0 as R . (8.1)
Proof. The proof that satises (2.23) and (2.30) is the same as in [5, Lemma 7.1].
Now we prove (8.1). Setting w(x) := (x) x
n
, noting that e
n
= 0 on
l
+
{x
n
> L}
by (2.17), and using (2.23) and (2.30), we see that w satises
n
i,j=1
(a
ij
w
x
j
)
x
i
= 0 in {x
n
> L},
n
i,j=1
a
ij
w
x
j
i
= 0 on
l
+
{x
n
> L},
lim
R
w
C(){x
n
>R})
= 0,
where
a
ij
(x) =
_
1
0
_

_
|tD(x) + (1 t)e
n
|
2
_
j
i
+2
_
|tD(x) + (1 t)e
n
|
2
_
(t
x
i
(x) + (1 t)
n
i
)
_
t
x
j
(x) + (1 t)
n
j
_
_
dt.
By (2.24), the equation is elliptic if is small. Then the proof of (8.1) is similar to the proof of
(7.3) in Proposition 7.1.
Now we show that in Problem (TN) is uniquely determined by
.
Proposition 8.2. There exists
0
depending only on the data as in Lemma 8.1 such that, if
satises (4.5) and C

0,1
() is a solution of Problem (TN) with (0, c
) satisfying (2.24),
then = q
+
in (2.20), with q
+
dened by (2.26) and (x) satises (2.28).
This can be achieved by following the proof of [5, Proposition 7.2] and using Lemma 8.1 instead
of [5, Lemma 7.1].
It remains to prove the uniqueness of solutions of Problem (TN) satisfying (2.24) when
satises (2.32).
Proposition 8.3. There exists
0
depending only on the data as in Lemma 8.1 such that, if
satises (2.32), then the solution C

0,1
() of Problem (TN) with = q
+
satisfying (2.24) is
unique.
The rest of this section is to prove Proposition 8.3 by using a version of the partial hodograph
transform developed in [6].
8.1. Extension of
to the domain . Since
satises (2.19) and (2.32) in the domain
1
:= {x
n
< 1}, then we can extend
to so that the extension (still denoted by)
is
in C
2,
(), coincides with the original
in {x
n
< 7/8}, and
0

2,,
C(n, ), (8.2)
supp(
0
) {x
n
< L}, (8.3)
= 0 on
l
, (8.4)
where
0
(x) = q
0
x
n
. To construct such an extension, we rst note that we can modify the
mapping and replace it by a mapping
1
C
2,
(R
n
, R
n
) such that
1
satises (2.14),
1
Id
2,,R
n C, (2.17), and
D
1
(x)
C
(x) = (x)
(
1
(x)) for any x C,
where C = (, ),
C
() and
() are the inward unit normals to C and

l
respectively,
and 1
2,,C
C. To construct such
1
, we rst dene the corresponding C
2,
mapping
1
between suciently small neighborhoods (depending only on and ) of C and
l
by
1
(z + s
C
(z)) = (z) + s
((z))
for z C and s (0, ), where > 0 depends only on and L. Then we let
1
(x) =
(d
(x))
1
(x) + (1
(d
(x)))(x),
where d
() is the distance function to

l
and
: R R is a smooth function satisfying

0
1 on R,
1 on [, 7/8), and
0 on (, ). Now we dene
(x) =
(
1
(x))
on
1
1
(
1
). Then, if is small,
is dened at least on (7/8, 7/8) and satises D
C
= 0
on (7/8, 7/8). Now we dene

(x) =
1
(x
n
)
(x) + (1
1
(x
n
))
0
(x) on (7/8, ),
and then the function
(x) =
(
1
1
(x)) is a C
2,
extension of
satisfying (8.2)(8.4).
Denote
g = div ((|D
|
2
)D
) in . (8.5)
Since
and
0
satisfy (1.1) in
1
and , respectively, then we nd from (8.2)(8.3) that g
satises
g C
(), g
0,,
C,
g 0 on ( {x
n
< 7/8}) ( {x
n
> 1}) .
(8.6)
From now on, we use the extended function
(x) and C may denote a dierent constant

at each occurrence, depending only on the data, unless otherwise is specied.
8.2. Partial hodograph transform. Let be a solution of Problem (TN) with = q
+
and
satisfy (2.24). Then
+
:= {|D| < c
} has the form (2.23) and (2.30). Dene a function u in
+
by
u(x) :=
(x) (x).
Then (2.31) and (8.2) imply
u (q
0
q
+
)x
n
1,,
+ C. (8.7)
In particular, if is suciently small, we use (2.29) to obtain
0 < (q
0
q
+
0
)/2 u
x
n
(x) 2(q
0
q
+
0
) for any x
+
. (8.8)
Furthermore, (2.28) and (8.3) imply
u (q
0
q
+
)x
n
C
1
({x
n
>R})
0 as R . (8.9)
Since is a solution of Problem (TN), satises (1.1) in
+
and (2.4) on S. Then, since
S
1
and
satises (8.4) on
l
, and also using (8.5), we see that u is a solution of the
following problem:
div (A(x, Du)) = g in
+
,
A(x, Du) = 0 on S,
A(x, Du) = 0 on
l
+
,
(8.10)
where, for x
+
and P R
n
,
A(x, P) = (|D
(x) P|
2
)(D
(x) P) (|D
(x)|
2
)D
(x). (8.11)
Note that u satises the weak form of problem (8.10):
_
+
(A(x, Du) D g) dx = 0 for any C
1
0
(R
n
). (8.12)
Since =
on S, it follows that
u = 0 on S. (8.13)
Now we make a change of variables z = (x) to rewrite problem (8.12) into a cylindric domain
C
+
:= (
+
), where and its inverse are the maps in (2.14), (2.16), and (6.21). It follows
that, if is suciently small, then
C
+
= ( R) {z
n
>

f(z
)},
f
1,,R
n1 CM, (8.14)
and C
+
=
l
C
+

S, where
l
C
+
:= ( R) {z
n
>

f(z
)} = (
l
+
),
S := ( R) {z
n
=

f(z
)} = (S).
Also,
+
= (C
+
). Consider the function
u(z) = u((z)) for z C
+
.
Then (2.16)(2.17) and (8.7) imply
u (q
0
q
+
)x
n
1,,C
+ C. (8.15)
In particular, u satises (8.8) if is small. By (8.9) and (2.17),
u (q
0
q
+
)x
n
C
1
((R,))
0 as R . (8.16)
From (8.13), u satises
u = 0 on

S. (8.17)
Moreover, u is a weak solution of a xed boundary problem of the form (8.10) in the domain
C
+
with modied functions

A(x, P) and g(x) satisfying

A C
1,
( (1, ) R
n
) and g
C
( (1, )). Precisely, u satises

_
C
+
(

A(x, D u) D g) dx = 0 for any C
1
0
(R
n
). (8.18)
This is obtained by using the test function , making the change of variables x (x) in
(8.12), and obtaining the expressions for

A in C
+
R
n
and g in C
+
in terms of A, g, and the
map . We do not write the explicit expressions of

A and g, but only point out some properties
of

A and g, which can be obtained readily by using these expressions. First, we emphasize that
A and g are given in terms of the original functions A, g, and the map , which are independent
of the solution of Problem (TN). Next, using (2.17), we see that, for any R > 0, there exists
C
R
such that
A

A
1,,C
+
B
R
C
R
C
2,
(R
n
)
. (8.19)
Furthermore, from (2.17), it follows that A =

A for x
n
> L, i.e.,
A(x, P) =

A(x, P) for any x (L, ), P R
n
. (8.20)
Similarly, from (2.16)(2.17) and (8.6), we conclude that g satises
g C
( [1, )), g
0,,(1,)
C,
g 0 on ( {x
n
< 1/2}) ( {x
n
> L})
(8.21)
if L > 1 and is suciently small. Dene
F(x
, x
n
) =
_
x
n
0
g(x
, s)ds for (x
, x
n
) (1, ). (8.22)
Then, from (8.21)(8.22), we have
F,

F
x
n
C
( [1, )), (8.23)
F
0,,(1,)
C,
F
x
n
0,,(1,)
C, (8.24)
F 0 in (1, 1/2), (8.25)
F is independent of x
n
(L, ). (8.26)
Note that, from (8.14), for suciently small , the free boundary

S lies within the domain
(1, 1/2). Then, by (8.25), the function

F vanishes on

S. Also, using (8.14) and (8.23), we
can rewrite (8.18) as
_
C
+
(

A(x, D u) +

F(x)e
n
) D dx = 0 for any C
1
0
(R
n
). (8.27)
Now we make the partial hodograph transform. Dene a mapping F : C
+
R
n
by
(x
, x
n
) (y
, y
n
) = (x
, u(x
, x
n
)).
The nondegeneracy property (8.8) of u implies that the mapping F is one-to-one on C
+
and,
from (8.8) for u, (8.14), and (8.17)
F(C
+
) = (0, ), F(
S) = {0},
i.e., the free boundary

S is mapped to the xed boundary {0}. Also, by (8.8) for u, there
exists a function v C
1,
( (0, )) such that, for (x
, x
n
) C
+
and y
n
0,
u(x
, x
n
) = y
n
if and only if v(x
, y
n
) = x
n
. (8.28)
Thus,
F
1
(y
, y
n
) = (y
, v(y
, y
n
)).
Dierentiating the identity u(x
, v(x
, y
n
)) = y
n
, which holds for any (x
, y
n
) (0, ), we
nd
v
y
n
> 0 in (0, )
and
D
x
u =
1
v
y
n
D
y
v, u
x
n
=
1
v
y
n
, (8.29)
where the left-hand and right-hand sides are taken at the points (x
, x
n
) and F(x
, x
n
), respec-
tively. In particular, property (8.8) of u implies
0 <
1
2(q
0
q
+
0
)
v
y
n
(y)
2
q
0
q
+
0
for any y C
+
. (8.30)
From this and (8.15), we obtain
v v
0
1,,(0,)
C, (8.31)
where v
0
(y) = y
n
/(q
0
q
+
). From (8.16) and (8.30),
v v
0
C
1
((R,))
0 as R . (8.32)
Now, since u(x) is a solution of the conormal problem (8.27) in C
+
, then v(y) is a solution
of the corresponding problem in (0, ). In order to show that this problem has also a
conormal structure, we make the change of variables x y = F(x) in (8.27). In order to do
that, we especially need to change the variables in the test function . For that, we note that
the function (y) := F
1
(y) = (y
, v(y
, y
n
)) satises C
1
( (0, )) and, by (8.30), if
0 on R
n
\ B
R
, then 0 on ( (0, )) \ B
R
1
for some R
1
, i.e., =

|
(0,)
for some
C
1
0
(R
n
). Similarly, for any C
1
0
(R
n
), there exists C
1
0
(R
n
) such that |
R
n
+
= F
1
and (x) = F(x) for x C
+
. We dierentiate the identity (x) = (x
, u(x
, x
n
)) and use
(8.29) to obtain
D
x
= D
y

y
n
v
y
n
D
y
v,
y
n
=

y
n
v
y
n
. (8.33)
Now, in (8.27), we make the change of variables x y = F(x), use (8.29) and (8.33), note that
the Jacobian of F
1
is JF
1
(y) = v
y
n
(y), and write

A(x
, x
n
, p
, p
n
) for

A(x, P) and

F(x
, x
n
)
for

F(x) to obtain
_
(0,)
_
B(y
, v, Dv) +

F(y
, v)e
n
_
Ddy = 0 for any C
1
0
(R
n
), (8.34)
where, for y
, z R, P = (p
, p
n
) R
n1
R
+
,
B
i
(y
, z, P) =

A
i
(y
, z,
p
p
n
,
1
p
n
)p
n
for i = 1, . . . , n 1,
B
n
(y
, z, P) =

A
n
(y
, z,
p
p
n
,
1
p
n
)
n1
i=1
A
i
(y
, z,
p
p
n
,
1
p
n
)p
i
.
(8.35)
Thus, v(y) is a weak solution of the corresponding conormal problem in (0, ).
Note that the functions B(y
, z, P) and

F(y
, z) and the domain (0, ) in (8.34) are

independent of the solution of Problem (TN).
Conversely, let v(y) is a solution of (8.34) satisfying (8.31)(8.32) with suciently small
depending only on the data so that (8.30) holds. Then the corresponding solution of Problem
(TN), satisfying (2.24), can be determined. Indeed, from (8.30),
C
+
:= {(x
, v(x
, y
n
)) : x
, y
n
> 0} (8.36)
has the form (8.14), and a function u C
1,
(C
+
) can be dened such that (8.28) holds. Then
(8.27) holds. Moreover, (8.31) implies (8.15). Next, we conclude that
+
:= (C
+
) has the form
(2.23) and (2.30), and u(x) := u((x)) satises (8.12). Moreover, (8.15) implies (8.7). If is
suciently small, then (8.7) and (8.15) imply (8.8) for u and u, respectively. Now, (8.32) implies
(8.16) and further (8.9). Then it follows that , dened by =
u in
+
and =
in
\
+
, is a weak solution of Problem (TN) satisfying (2.24).
Thus, the uniqueness of solutions of the conormal problem (8.34) satisfying (8.31)(8.32)
implies the uniqueness of solutions of Problem (TN) satisfying (2.24).
We also note that, since satises equation (1.1) which is uniformly elliptic in
+
and the
function in (1.1) is smooth in the range of |D|
2
, and since satises (2.19) on
l
+
and
l
is a C
2,
boundary, then the standard regularity results for elliptic equations and oblique
derivative problems imply C
2,
(
+
\ S). Then it follows that v C
2,
( (0, )).
8.3. Uniqueness of weak solutions of the conormal problem (8.34) satisfying (8.31)
(8.32). In the argument of this section, the constant C depends only on the data, unless other-
wise is specied.
Suppose that v
1
and v
2
are two solutions of (8.34) in (0, ) satisfying (8.31)(8.32). We
assume that is small so that (8.30) holds for v
1
and v
2
. Consider the function w = v
2
v
1
.
Then w satises
w
1,,(0,)
2C (8.37)
and
w
1,0,(R,)
0 as R . (8.38)
Moreover, w C
2,
((0, )) since v
1
and v
2
have this regularity. Also, subtracting equations
(8.34) for v
1
and v
2
, we obtain
_
(0,)
_
n
i,j=1
a
ij
(y)w
y
j
y
i
+
n
i=1
b
i
(y)w
y
i
_
dy = 0 (8.39)
for any C
1
c
(R
n
), where a
ij
and b
i
are dened by
a
ij
(y) =
_
1
0
B
i
p
j
(y
, (1 t)v
1
(y) + tv
2
(y), (1 t)Dv
1
(y) + tDv
2
(y))dt,
b
i
(y) =
_
1
0
_
B
i
z
(y
, (1 t)v
1
(y) + tv
2
(y), (1 t)Dv
1
(y) + tDv
2
(y))
+
n
i

n

F(y
, (1 t)v
1
(y) + tv
2
(y))
_
dt.
(8.40)
Note that a
ij
, b
i
C
( [0, )).
Let (a
0
ij
, b
0
i
) be the coecients (a
ij
, b
i
) for the background solution: Id,
(x) = q
0
x
n
,
v
1
(y) = v
2
(y) = v
0
y
n
/(q
0
q
+
0
). Then, using (2.11), (8.11), and (8.35), we nd
a
0
ij
=
i
j
i
, b
0
i
= 0 for i, j = 1, . . . , n, (8.41)
where
i
are the following constants:
i
=
_
(q
+
0
)
2
_
for i = 1, . . . , n1, and
n
=
(q
+
0
)/(q
0

q
+
0
) for (s) dened by (2.5). Then, since q
+
0
(0, c
), we obtain from (2.5) and (2.7) that
i
> 0 for i = 1, . . . , n.
From (8.41) and using (2.16), (8.2), (8.11), (8.19), (8.24), (8.30)(8.31), and (8.35), we have
a
ij

i
j
i
0,,(0,)
+b
i
0,,(0,)
C for i, j = 1, . . . , n. (8.42)
Moreover, using (8.3), (8.20), and (8.26), we see that
b
i
0 on {y
n
> L
1
} for i = 1, . . . , n, (8.43)
where L
1
= 2L/(q
0
q
+
0
).
Note that, by (8.42) with suciently small depending only on the data, the coecients
a
ij
(y) satisfy the ellipticity condition (3.35) with constant

> 0 depending only on the data for
any y (0, ).
Now we prove that w 0. We will do this in three steps.
Step 1. We show that there exist R
1
2L
1
and a nonempty open set
such that
w(x
, R
1
) 0 for all x
.
Denote
g :=
n
i=1
a
ni
w
x
i

n
i,j=1
a
ij
w
x
j
(e
n
)
i
on {2L
1
}. (8.44)
From (8.37) and (8.42), we have
g
0,,
C. (8.45)
Now we use (8.43) and w C
2
( (0, )) to conclude that w, restricted to (2L
1
, ), is a
C
2
solution of the conormal problem
n
i,j=1
(a
ij
w
x
j
)
x
i
= 0 in (2L
1
, ),
n
i,j=1
a
ij
w
x
j
i
= 0 on (2L
1
, ),
n
i,j=1
a
ij
w
x
j
i
= g on {2L
1
}.
(8.46)
The uniqueness argument in the proof of Proposition 7.1 can show that w is the unique weak
solution in C
1,
( [2L
1
, )) of problem (8.46) satisfying (8.38).
Moreover,
_
{2L
1
}
g dx
= 0. (8.47)
Indeed, we use in (8.39) the test function (x) = (x)((2L
1
x
n
)/), where > 0, C
0
(R
n
)
with 1 on (0, 4L
1
), and C
(R) with 0 on R, 1 on (1, ) and 0 on

(, 1/2), and then we send 0+ using that w C
2
( (0, )) to get (8.47).
Let R > 2L
1
. Consider the conormal problem
n
i,j=1
(a
ij
w
(R)
x
j
)
x
i
= 0 in (2L
1
, ),
n
i,j=1
a
ij
w
(R)
x
j

i
= 0 on (2L
1
, ),
n
i,j=1
a
ij
w
(R)
x
j

i
= g on {2L
1
},
w
(R)
= 0 on {R}.
(8.48)
Now we can follow the argument of Section 6. Let w
(R)
H
1
( (2L
1
, R)) be a weak solution
of problem (8.48). Using (8.45) and (8.47) instead of (6.5) and (6.2) respectively, we follow the
proofs of Lemmas 6.36.4 to show that w
(R)
satises
Dw
(R)
L
2
((2L
1
,R))
C
with C independent of R and that, for every x
n
(2L
1
, R), there exists x
such that
w
(R)
(x
, x
n
) = 0. Then we follow the proofs of Propositions 6.56.7 to show that if > 0 is
small depending only on the data, there exists a unique weak solution w
(R)
H
1
( (2L
1
, R))
of problem (8.48), and it satises
w
(R)
1,,(2L
1
,R10r)
C,
with C independent of R. Now we follow the proof of Proposition 7.1 to show that there exists
a sequence R
j
such that w
(R
j
)
converge in C
1,/2
on any compact subset of [2L
1
, )
to a solution w C
1,
( [2L
1
, )) of problem (8.46) satisfying (8.38). Moreover, for any
x
n
[2L
1
, ) there exists x
such that w(x
, x
n
) = 0. From the uniqueness of a weak
solution in C
1,
([2L
1
, )) of problem (8.46) satisfying (8.38), we conclude that w = w. Thus
there exists x
0
such that w(x
0
, 3L
1
) = 0.
If w is constant on (2L
1
, ), then w 0 on (2L
1
, ) by (8.38), and the assertion of
Step 1 is proved. Thus we assume that w is not constant on (2L
1
, ). Also, w is bounded
on (2L
1
, ) by (8.37).
Now, we use that w C
2,
([L
1
, )) and satises
n
i,j=1
(a
ij
w
x
j
)
x
i
= 0 in (L
1
, ) and
n
i,j=1
a
ij
w
x
j
i
= 0 on (L
1
, ). We also note that, since v
1
, v
2
C
2,
( [L
1
, )), we
have a
ij
C
1,
([L
1
, )) so that the equation in (8.46) rewritten in the non-divergence form
has continuous coecients in [L
1
, ). Then it follows that inf
x[2L
1
,)
w(x) cannot be
achieved at any x [2L
1
, ) by the strong maximum principle and at any x [2L
1
, )
since Hopfs lemma contradicts the conormal boundary condition at such point. Thus, from
w(x
0
, 3L
1
) = 0, we conclude that, if w 0 on (2L
1
, ), then w 0 on (2L
1
, ).
Thus, if w is not identically zero on (2L
1
, ), then there exists x
[2L
1
, ) such
that w(x
) < 0. Since w is continuous, the assertion of Step 1 follows.

Step 2. We use that w 0 on
{R
1
} with R
1
2L
1
by Step 1 to prove that w 0 on
(0, 2L
1
).
Choose the following test function in (8.39): For xed > 0 and R > 0,
=
(w m
R
)
+
(w m
R
)
+
+
, where m
R
:= max
_
0, sup
y(R,)
w(y)
_
<
by (8.37). Then C
0,1
( [0, )) and 0 on [R, ). Thus we can substitute this
into (8.39) to obtain
_
(0,R)
_
a
ij
w
+
y
i
((w m
R
)
+
)
y
j
((w m
R
)
+
+ )
2
+
b
j
w
+
((w m
R
)
+
)
y
j
((w m
R
)
+
+ )
2
_
dy = 0.
Then using the ellipticity of (8.39) with constant

> 0, (8.37), and (8.42)(8.43) yields
_
(0,R)
Dlog
_
1 + (w m
R
)
+
/
_
2
dy =

_
(0,R)
|D(w m
R
)
+
|
2
((w m
R
)
+
+ )
2
dy

_
(0,R)
b
j
w
+
((w m
R
)
+
)
y
j
((w m
R
)
+
+ )
2
dy
=
_
(0,L
1
)
b
j
_(w m
R
)
+
((w m
R
)
+
)
y
j
((w m
R
)
+
+ )
2
+ m
R
((w m
R
)
+
)
y
j
((w m
R
)
+
+ )
2
_
dy
C
_
(0,L
1
)
_
Dlog
_
1 + (w m
R
)
+
/
_
+
m
R
2
_
dy
C
_
_
(0,L
1
)
Dlog
_
1 + (w m
R
)
+
/
_
2
dy
_
1/2
+ C
m
R
2
.
Thus, for R > L
1
,
_
(0,R)
Dlog
_
1 + (w m
R
)
+
/
_
2
dy C + C
m
R
2
.
Letting R and noting that lim
R
m
R
= 0 by (8.38), we obtain
_
(0,)
Dlog
_
1 + w
+
/
_
2
dy C.
Since log
_
1 + w
+
/
_
= 0 on
{x
n
= R
1
} where
is nonempty and open, it follows from

the Poincare inequality that
_
(0,R
1
)
log
_
1 + w
+
/
_
2
dy C,
where C depends only on the data,
, and R
1
, but is independent of . Letting 0, we
conclude w
+
0 in (0, R
1
), which implies w 0 in (0, 2L
1
), since R
1
2L
1
.
Step 3. Similarly, we prove w 0 in (0, 2L
1
), thus w 0 in (0, 2L
1
). Also,
w C
1
([0, )). Then g 0 by (8.44). Now w 0 in (2L
1
, ) by uniqueness of a weak
solution in C
1,
([2L
1
, )) of problem (8.46) satisfying (8.38). Thus w 0 in (0, ), and
the uniqueness of weak solutions of the conormal problem (8.34) satisfying (8.31)(8.32) follows.
8.4. Stability. As a consequence of the uniqueness, nondegeneracy, and regularity of solutions
of the free boundary problem, we have the following stability theorem.
Theorem 8.1. There exist a constant
0
> 0 and a nonnegative nondecreasing function G
C([0, )) with G(0) = 0, depending only on n, , , C
0
, , L, and q
+
0
, such that, if
(i) 0 < < <
0
;
(ii) : R
n
R
n
satises (2.16)(2.17), and

: R
n
R
n
satises (2.17) and
3,,(
1
2
,
1
2
)
; (8.49)
(iii)
(x) satises (2.32) and
(x) satises
2,,(
1
2
,
1
2
)
, (8.50)
then the unique solutions (x) and (x) of Problem (TN) for the nozzles dened by the mappings
and

in (2.14) and supersonic upstream ows
(x) and
(x), respectively, satisfy
1,,
G(), (8.51)
where

f
(x
) and

f

(x
) are the free boundary functions of the cylindric domains C

+
() :=
(
+
()) and C
+
( (x)) :=

(
+
( )) in (8.14), respectively, where =
1
and

=

1
.
Its proof repeats the one of [4, Theorem 6.1]. See also the remarks on the proof of the stability
part of [5, Theorem 2.3]. In the present case, as an additional step, the mapping from to C is
involved.
Appendix A. Existence of Smooth Supersonic Upstream Flows in the Nozzle
In this paper, we require a local C
1,
supersonic solution
in Sections 37 and C
2,
super-
sonic solution
in Section 8 in the domain

1
:= {1 < x
n
1} of the initial-boundary
value problem (2.18)(2.19) for the nonlinear wave equation (1.1) such that
0

k,,
1
C
0
, k = 1 or 2, (A.1)
for some C
0
> 0 independent of , when is suciently small. This is a direct corollary of the
standard local theory of smooth solutions of the initial-boundary value problem for the nonlinear
wave equations.
Proposition A.1. Let the nozzle boundary satisfy (2.15)(2.16). Let the Cauchy data (
e
,
e
)
on
o
satisfy
e
q
0
x
n
H
s+k +
e
q
0

H
s+k1 , k = 1 or 2, (A.2)
for some integer s > n/2 + 1 and the compatibility conditions on the intersection
l

o
up to the (s + k)
th
order. Then there exist
0
> 0 and C
0
> 0 such that there is a unique
classical solution
C
k+1
(
1
) of problem (2.18)(2.19) for the nonlinear wave equation (1.1),
respectively for k = 1 or 2, such that, when
0
,
0

k+1,0,
1
C
0
. (A.3)
If, in addition, conditions (2.15)(2.16) and (A.2), as well as the compatibility conditions, hold
for higher-order derivatives, then the solution
has the corresponding higher regularity.

As in Section 8, we set
w(x) :=
(x) q
0
x
n
.
Then w satises the following initial-boundary value problem:
div(A(Dw)) = 0, (A.4)
(w, w
x
n
)|
x
n
=1
= (
e
+ q
0
,
e
q
0
), (A.5)
w
= q
0

n
, (A.6)
as well as the compatibility conditions up to the (s + k)
th
order, where
A(P) = (|P + q
0
e
n
|
2
)(P + q
0
e
n
) ((q
0
)
2
)q
0
e
n
.
Since q
0
> c
, then (A.4) is strictly hyperbolic, nonlinear wave equations with respect to the
x
n
-direction (whose role is as the time direction) for a small perturbation of the trivial solution
u = 0. By the standard energy estimates and the iteration procedure (cf. [27, 28, 30] and the
references cited therein), for small > 0, there exists a unique solution
w(x
, x
n
)
s+k
j=0
C
j
([1, 1]; H
s+1j
(
))
such that
s+k
j=0
w
C
j
([1,1];H
s+1j
(
))
C,
for k = 1 or 2, respectively. Since s > n/2 + 1, the Sobolev embedding theorem implies (A.3).
Acknowledgments. Gui-Qiang Chens research was supported in part by the National Science
Foundation under Grants DMS-0244473, DMS-0204225, DMS-0204455, and INT-9987378, and
an Alexandre von Humboldt Foundation Fellowship. Mikhail Feldmans research was supported
in part by the National Science Foundation under Grants DMS-0200644 and DMS-0074037.
References
[1] S. Agmon, A. Douglis, and L. Nirenberg, Estimates near the boundary for solutions of elliptic partial dier-
ential equations satisfying general boundary conditions I, Comm. Pure Appl. Math. 12 (1959), 623727; II,
Comm. Pure Appl. Math. 17 (1964), 3592.
[2] L. Bers, Existence and uniqueness of subsonic ows past a given prole, Comm. Pure Appl. Math. 7 (1954),
441-504.
[3] S. Canic, B. L. Keytz, and G. Lieberman, A proof of existence of perturbed steady transonic shocks via a
free boundary problem, Comm. Pure Appl. Math. 53 (2000), 484-511.
[4] G.-Q. Chen and M. Feldman, Multidimensional transonic shocks and free boundary problems for nonlinear
equations of mixed type, J. Amer. Math. Soc. 16 (2003), 461-494.
[5] G.-Q. Chen and M. Feldman, Steady transonic shocks and free boundary problems in innite cylinders for
the Euler equations, Comm. Pure Appl. Math. 57 (2004), 310356.
[6] G.-Q. Chen and M. Feldman, Free boundary problems and transonic shocks for the Euler equations in
unbounded domains, Ann. Scuola Norm. Sup. Pisa Cl. Sci. (5), 3 (2004), 827869.
[7] S.-X. Chen, Existence of stationary supersonic ows past a point body, Arch. Rational Mech. Anal. 156
(2001), 141-181.
[8] S.-X. Chen, Asymptotic behavior of supersonic ow past a convex combined wedge, Chinese Ann. Math.
19B (1998), 255-264.
[9] M. Costabel and M. Dauge, Construction of corner singularities for Agmon-Douglis-Nirenberg elliptic systems,
Math. Nachr. 162 (1993), 209237.
[10] M. Costabel and M. Dauge, Stable asymptotics for elliptic systems on plane domains with corners, Commun.
Partial Di. Eqs. 19 (1994), 16771726.
[11] R. Courant and K. O. Friedrichs, Supersonic Flow and Shock Waves, Springer-Verlag: New York, 1948.
[12] C. M. Dafermos, Hyperbolic Conservation Laws in Continuum Physics, 2nd Ed., Springer-Verlag: Berlin,
2005.
[13] M. Dauge, Elliptic Boundary Value Problems in Corner DomainsSmoothness and Asymptotics of Solutions,
Lecture Notes in Math. 1341, Springer-Verlag: Berlin, 1988.
[14] G. Dong, Nonlinear Partial Dierential Equations of Second Order, Transl. Math. Monographs, 95, AMS:
Providence, RI, 1991.
[15] P. Embid, J. Goodman, and A. Majda, Multiple steady state for 1-D transonic ow, SIAM J. Sci. Stat.
Comp. 5 (1984), 21-41.
[16] R. Finn and D. Gilbarg, Asymptotic behavior and uniqueness of plane subsonic ows, Comm. Pure Appl.
Math. 10 (1957), 2363.
[17] R. Finn and D. Gilbarg, Three-dimensional subsonic ows, and asymptotic estimates for elliptic partial
dierential equations, Acta Math. 98 (1957), 265-296.
[18] C. A. J. Fletcher, Computational Techniques for Fluid Dynamics, Vols. I-II, Springer-Verlag: Berlin, 1991.
[19] D. Gilbarg and L. Hormander, Intermediate Schauder estimates, Arch. Rational Mech. Anal. 74 (1980),
297318.
[20] D. Gilbarg and N. Trudinger, Elliptic Partial Dierential Equations of Second Order, 2nd Ed., Springer-
Verlag: Berlin, 1983.
[21] H. Glaz and T.-P. Liu, The asymptotic analysis of wave interactions and numerical calculations of transonic
nozzle ow, Adv. Appl. Math. 5 (1984), 114146.
[22] J. Glimm and A. Majda, Multidimensional Hyperbolic Problems and Computations, Springer-Verlag: New
York, 1991.
[23] J. Glimm, G. Marshall, and B. Plohr, A generalized Riemann problem for quasi-one-dimensional gas ow,
Adv. Appl. Math. 5 (1984), 130.
[24] P. Grisvard, Elliptic Problems in Nonsmooth Domains, Monographs and Studies in Math. 24, Pitman: Lon-
don, 1985.
[25] C.-H. Gu, A method for solving the supersonic ow past a curved wedge (in Chinese), Fudan J. 7 (1962),
11-14.
[26] Q. Han and F. Lin, Elliptic Partial Dierential Equations, Courant Lecture Notes in Mathematics 1. Courant
Institute/AMS: New York/Providence, 1997.
[27] F. John, Nonlinear Wave Equations, Formation of Singularities, University Lecture Series 2, AMS: Provi-
dence, 1990.
[28] M. Kikuchi and Y. Shibata, On the mixed problem for some quasilinear hyperbolic system with fully nonlinear
boundary condition, J. Di. Eqs. 80 (1989), 154197.
[29] D. Kinderlehrer and L. Nirenberg, Regularity in free boundary problems, Ann. Scuola Norm. Sup. Pisa Cl.
Sci. (4), 4 (1977), 373391.
[30] H. Koch, Mixed problems for fully nonlinear hyperbolic equations, Math. Z. 214 (1993), 942.
[31] P. D. Lax, Hyperbolic Systems of Conservation Laws and the Mathematical Theory of Shock Waves, CBMS-
RCSM, SIAM: Philiadelphia, 1973.
[32] T.-T. Li, On a free boundary problem, Chinese Ann. Math. 1 (1980), 351-358.
[33] T.-T. Li and W.-C. Yu, Boundary Value Problems for Quasilinear Hyperbolic Systems, Duke University
Mathematics Series, V. Durham, NC 27706: Duke University, Mathematics Department. X, 1985.
[34] G. Lieberman, Mixed boundary value problems for elliptic and parabolic dierential equations of second
order, J. Math. Anal. Appl. 113 (1986), 422440.
[35] G. Lieberman, Holder continuity of the gradient of solutions of uniformly parabolic equations with conormal
boundary conditions, Ann. Mat. Pura Appl. (4), 148 (1987), 7799.
[36] G. Lieberman, Oblique derivative problems in Lipschitz domains II: Discontinuous boundary data. J. Reine
Angew. Math. 389 (1988), 121.
[37] G. Lieberman, Second Order Parabolic Dierential Equations, World Scientic Publishing Co. Inc.: River
Edge, NJ, 1996.
[38] G. Lieberman and N. Trudinger, Nonlinear oblique boundary value problems for nonlinear elliptic equations,
Trans. Amer. Math. Soc. 295 (1986), 509546.
[39] W.-C. Lien and T.-P. Liu, Nonlinear stability of a self-similar 3-dimensional gas ow, Commun. Math. Phys.
204 (1999), 525549.
[40] T.-P. Liu, Nonlinear stability and instability of transonic gas ow through a nozzle, Comm. Math. Phys. 83
(1983), 243-260.
[41] T.-P. Liu, Nonlinear resonance for quasilinear hyperbolic equation, J. Math. Phys. 28 (1987), 25932602.
[42] C. S. Morawetz, On the non-existence of continuous transonic ows past proles IIII, Comm. Pure Appl.
Math. 9 (1956), 4568; 10 (1957), 107131; 11 (1958), 129144.
[43] C. S. Morawetz, On a weak solution for a transonic ow problem, Comm. Pure Appl. Math. 38 (1985),
797817.
[44] D. G. Schaeer, Supersonic ow past a nearly straight wedge, Duke Math. J. 43 (1976), 637-670.
[45] M. Shiman, On the existence of subsonic ows of a compressible uid, J. Rational Mech. Anal. 1 (1952),
605652.
[46] E. M. Stein, Singular Integrals and Dierentiability Properties of Functions, Princeton Univ. Press: Princeton,
NJ, 1970.
[47] G. B. Whitham, Linear and Nonlinear Waves, John Wiley & Sons: New York, 1974.
[48] P. Xin and H. Yin, Transonic shock in a nozzle: two-dimensional case, Comm. Pure Appl. Math. 58 (2005),
9991050.
[49] Y. Zhang, Global existence of steady supersonic potential ow past a curved wedge with a piecewise smooth
boundary, SIAM J. Math. Anal. 31 (1999), 166-183.
[50] Y. Zheng, Systems of Conservation Laws: Two-dimensional Riemann Problems, Birkhauser Boston, Inc.:
Boston, MA, 2001.
[51] Y. Zheng, A Global solution to a two-dimensional Riemann problem involving shocks as free boundaries,
Acta Math. Appl. Sinica (English Ser.) 19 (2003), 559572.
Department of Mathematics, Northwestern University, Evanston, IL 60208-2730
E-mail address: gqchen@math.northwestern.edu
Department of Mathematics, University of Wisconsin, Madison, WI 53706
E-mail address: feldman@math.wisc.edu

Gui-Qiang Chen and Mikhail Feldman - Existence and Stability of Multidimensional Transonic Flows Through An Infinite Nozzle of Arbitrary Cross-Sections

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Gui-Qiang Chen and Mikhail Feldman - Existence and Stability of Multidimensional Transonic Flows Through An Infinite Nozzle of Arbitrary Cross-Sections

Uploaded by

Copyright:

Available Formats

EXISTENCE AND STABILITY OF MULTIDIMENSIONAL TRANSONIC

FLOWS THROUGH AN INFINITE NOZZLE OF ARBITRARY

coecients for subsequent development. In Section

be open subsets of such

sides. Moreover, a function C

, (2.2), and equation (1.1) in

respectively, is a weak solution of (1.1),

. Then we can also write (2.2) as

(p) < 1 on (0, c

(p) < 0 on (0, c

and subsonic if |D| < c

S) is a weak solution satisfying

and satisfying the physical entropy condition (see Courant-

on [L, ) for some constants d

of the solution can be constructed

, a weak solution of (1.1) in

) dividing the subsonic ow

in (2.10) are of the form

of (1.1) satisfying (2.19) and (2.21), there exists a solution

has a higher regularity, then we have the following unique-

in Problem (FB) additionally

in Theorem 2.3 can be replaced by the C

) of equation (2.26) such that

on the boundary part

, since the interior estimate in B

norms of the coecients

(z) without the terms involving the norms of g

(3Q/2, 3Q/2), we can locally atten

= 1, dene the function

)) satisfying (3.42) and G 0 on

in with the denition

} has the form:

, we can expect to nd only a weak solution u C

, which will be used later.

is dened by (5.8) and

such that u(x

, z) = 0. Then we can assume u(x

, z) > 0 for all x

, z) < 0 for all x

can be handled similarly. From the compactness of and

, z) > 0 for all x

(R) with 0 on R, 1 on (1, ) and 0 on

. Moreover, from (3.54) of Proposition 3.3, there exist C > 0 and

by C from (2.16) and (6.5), respectively.

} ( R) has the form

(R) is nonincreasing and satises 1 on (, 1/8) and 0 on (1/4, ).

(x)) is a weak solution of (6.1)(6.4).

such that u(x

, similar to [4, Proposition 4.5]

satisfy (4.5). Then, for any K

) the function from (6.23) for

are the coecients and right-hand sides of (7.9)(7.11) for . By (7.8),

. Note also that v satises (7.7)

(). Thus the whole sequence P

is its free boundary.

(x)} and |D | > c

satisfy (4.5). Let C

satises (4.5) and C

satises (2.32), then the solution C

to the domain . Since

satises (2.19) and (2.32) in the domain

to so that the extension (still denoted by)