Upload
salvatore-a
View
214
Download
0
Embed Size (px)
Citation preview
This article was downloaded by: [University of Newcastle (Australia)]On: 16 March 2014, At: 13:08Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Applicable Analysis: An InternationalJournalPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/gapa20
On the structure of the critical set ofnon-differentiable functions with aweak compactness conditionGabriele Bonanno a & Salvatore A. Marano ba Department of Science for Engineering and Architecture(Mathematics Section), Engineering Faculty , University ofMessina , 98166 Messina, Italyb Dipartimento di Matematica e Informatica , Università degliStudi di Catania , Viale A. Doria 6, 95125 Catania, ItalyPublished online: 19 Jan 2010.
To cite this article: Gabriele Bonanno & Salvatore A. Marano (2010) On the structure of the criticalset of non-differentiable functions with a weak compactness condition, Applicable Analysis: AnInternational Journal, 89:1, 1-10, DOI: 10.1080/00036810903397438
To link to this article: http://dx.doi.org/10.1080/00036810903397438
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.
This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
Dow
nloa
ded
by [
Uni
vers
ity o
f N
ewca
stle
(A
ustr
alia
)] a
t 13:
08 1
6 M
arch
201
4
Applicable AnalysisVol. 89, No. 1, January 2010, 1–10
On the structure of the critical set of non-differentiable functions
with a weak compactness condition
Gabriele Bonannoa* and Salvatore A. Maranob
aDepartment of Science for Engineering and Architecture (Mathematics Section),Engineering Faculty, University of Messina, 98166 Messina, Italy;
bDipartimento di Matematica e Informatica, Universita degli Studi di Catania,Viale A. Doria 6, 95125 Catania, Italy
Communicated by S. Carl
(Received 17 March 2009; final version received 9 September 2009)
The nature of critical points generated by a Ghoussoub’s type min–maxprinciple for locally Lipschitz continuous functionals fulfilling a weakPalais–Smale assumption, which contains the so-called non-smooth Ceramicondition, is investigated. Two meaningful special cases are then pointedout; see Theorems 3.5 and 3.6.
Keywords: critical set; locally lipschitz continuous functions; weak Palais–Smale condition
AMS Subject Classifications: Primary 58E05; 49J35; 49J52
1. Introduction
Very recently, in [1], the classical min–max principle of Ghoussoub [2] has beenextended to locally Lipschitz continuous functions satisfying a weak Palais–Smalehypothesis, which includes both the usual one [3, Definition 2] and the non-smoothCerami condition [4, p. 248]. This result is achieved by adapting Ghoussoub’sapproach to the new framework and, in particular, exploiting Ekeland’s variationalprinciple with an appropriate metric of geodesic type. When the functional is C1
while the compactness assumption is that of Cerami [5, Section 13.1], such an ideabasically goes back to Ekeland [6, Section IV.1].
Besides the existence of critical points, Ghoussoub’s result provides valuableinformation about their nature; see [7, Chapter 5], [6, pp. 145–148] and [5, Chapter 12].Making use of [1, Theorem 3.1], we address here the same question forlocally Lipschitz continuous functions that fulfil a weak Palais-Smale condition.Hence, non-differentiable variants of several known structure results are established.For instance, Theorem 3.3 extends Corollary 4.14 in [7], Theorem 3.4 comes out fromthe famous Pucci and Serrin’s multiple critical point result [8, Corollary 1], whileTheorem 3.7 is patterned after Theorem 8 of [9] and includes [10, Theorem 3.3].
*Corresponding author. Email: [email protected]
ISSN 0003–6811 print/ISSN 1563–504X online
� 2010 Taylor & Francis
DOI: 10.1080/00036810903397438
http://www.informaworld.com
Dow
nloa
ded
by [
Uni
vers
ity o
f N
ewca
stle
(A
ustr
alia
)] a
t 13:
08 1
6 M
arch
201
4
Similar results were previously obtained in [11] for functionals expressed as the sum of
a locally Lipschitz continuous term and of a convex, proper, lower semi-continuous
function, namely the so-called Motreanu–Panagiotopoulos non-smooth setting [12].
However, the compactness hypothesis employed there reduces to the standard one in
the locally Lipschitz continuous case and hence it is more restrictive than that taken on
below.A special mention deserve Theorems 3.5 and 3.6, achieved by combining [13,
Theorem 3.1] with Theorem 3.4, because they deal with functionals of the type
f� :¼�� ��, where �40 while � and � are locally Lipschitz continuous. In this
case, starting from the seminal paper of Ricceri [14], many multiple critical point
results have been recently established, both in the smooth and in the non-smooth
framework; see [15] for an exhaustive bibliography. Roughly speaking, here, we get a
well-determined larger interval of parameters � for which f� possesses at least three
critical points under weaker regularity and compactness conditions.
2. Preliminaries
Let (X, k � k) be a real Banach space. If U is a subset of X, we write int(U ) for the
interior of U, U for the closure of U, and @U for the boundary of U. Moreover, when
x2X and r40, we define B(x, r) :¼ {z2X : kz�xk5r}, Bðx, rÞ :¼ Bðx, rÞ. Given
x, z2X, the symbol [x, z] indicates the line segment joining x to z, i.e.
[x, z] :¼ {xþ t(z� x): t2 [0, 1]}. We denote by X � the dual space of X, while h�, �i
stands for the duality pairing between X � and X. A function f :X! IR is called
coercive provided
limkxk!þ1
f ðxÞ ¼ þ1:
If to every x2X there correspond a neighbourhood Vx of x and a constant Lx� 0
such that
j f ðzÞ � f ðwÞj � Lxkz� wk 8z,w2X,
then we say that f is locally Lipschitz continuous. In this case f 0(x; z), x, z2X,
indicates the generalized directional derivative of f at the point x along the direction
z, namely
f 0ðx; zÞ :¼ lim supw!x,t!0þ
f ðwþ tzÞ � f ðwÞ
t:
It is known [16, Proposition 2.1.1] that f 0 is upper semi-continuous on X�X. The
generalized gradient of the function f in x, denoted by @f (x), is the set
@f ðxÞ :¼ fx� 2X � : hx�, zi � f 0ðx; zÞ 8z2X g:
Proposition 2.1.2 of [16] ensures that @f (x) is nonempty, convex, in addition to weak
star compact, and that
f 0ðx; zÞ ¼ maxfhx�, zi : x� 2 @f ðxÞg:
2 G. Bonanno and S.A. Marano
Dow
nloa
ded
by [
Uni
vers
ity o
f N
ewca
stle
(A
ustr
alia
)] a
t 13:
08 1
6 M
arch
201
4
Hence, it makes sense to put
mf ðxÞ :¼ minfkx�kX � : x� 2 @f ðxÞg:
We say that x2X is a critical point of f when 02 @f (x), i.e. f 0(x; z)� 0 for all z2X.Finally, given a real number c, write
Kcð f Þ :¼ fx2X : f ðxÞ ¼ c, x is a critical point of f g,
besides f c :¼ {x2X : f (x)� c} and fc :¼ {x2X : f (x)� c}.
3. Structure of the critical set
Let (X, k � k) be a real Banach space and let h : [0,þ1[! [0,þ1[ be a continuousfunction such that Z þ1
0
1
1þ hð�Þd� ¼ þ1: ð2:1Þ
Given x, z2X, we denote by P(x, z) the family of all piecewise C1 paths p : [0, 1]!Xsatisfying p(0)¼ x and p(1)¼ z. Moreover, put
lhð pÞ :¼
Z 1
0
k p0ðtÞk
1þ hðk pðtÞkÞdt, p2Pðx, zÞ,
as well as
�hðx, zÞ :¼ infflhð pÞ : p2Pðx, zÞg: ð2:2Þ
For h(�) :¼ �, � 2 [0,þ1[, the function �h :X�X! IR defined by (2.2) coincides withthe geodesic distance introduced in [6, p. 138]. Exploiting (2.1) and the arguments of[6, p. 138] yields the next basic properties of �h.
(p1) �h(x, z)�kx� zk for all x, z2X.(p2) If U is a nonempty bounded subset of X then there exists a constant cU40
such that
�hðx, zÞ � cUkx� zk 8x, z2U:
(p3) �h turns out to be a distance on X and the metric topology derived from �hcoincides with the norm topology.
(p4) �h-bounded and norm-bounded sets in X are the same.
Through (p1), (p2) and (p4) one easily verifies that the metric space (X, �h) iscomplete.
Now, let B be a nonempty closed subset of X and let F be a class of nonemptycompact sets in X. According to [2, Definition 1], we say that F is a homotopy-stablefamily with extended boundary B when for every A2F , �2C0([0, 1]�X,X ) such that�(t, x)¼ x on ({0}�X )[ ([0, 1]�B) one has �({1}�A)2F . Some meaningfulsituations are special cases of this notion.
Example 3.1 If Q denotes a compact set in X, Q0 is a nonempty closed subset of Q,�02C
0(Q0,X ), � :¼ {� 2C0(Q,X ) : �jQ0¼ �0}, and F :¼ {�(Q): � 2�}, then F is a
Applicable Analysis 3
Dow
nloa
ded
by [
Uni
vers
ity o
f N
ewca
stle
(A
ustr
alia
)] a
t 13:
08 1
6 M
arch
201
4
homotopy-stable family with extended boundary B :¼ �0(Q0). In particular, this
occurs when Q indicates a compact topological manifold in X having a nonempty
boundary Q0 while �0 :¼ idjQ0.
The following assumptions will be posited in the sequel.
(a1) f :X! IR is a locally Lipschitz continuous function.(a2) F denotes a homotopy-stable family with extended boundary B.(a3) There exists a nonempty closed subset F of X such that
ðA \ F Þ n B 6¼ ; 8A2F ð2:3Þ
and, moreover,
supx2B
f ðxÞ � infx2F
f ðxÞ: ð2:4Þ
(a4) h : [0,þ1[! [0,þ1[ is a continuous function fulfilling (2.1), while �h indicatesthe metric defined by (2.2).
Set, as usual,
c :¼ infA2F
maxx2A
f ðxÞ: ð2:5Þ
Thanks to (2.3) one has
infx2F
f ðxÞ � c: ð2:6Þ
The next result, patterned after the seminal min–max principle of Ghoussoub [2], has
been very recently established in [1]; see [1, Theorem 3.1].
THEOREM 3.1 Let (a1)–(a4) be satisfied. Then to every sequence {An}�F such that
limn!þ1maxx2Anf (x)¼ c there corresponds a sequence {xn}�X nB having the
following properties:
(i1) limn!þ1 f (xn)¼ c.(i2) (1þ h(kxnk))f
0(xn; z)��"nkzk for all n2 IN, z2X,where "n! 0þ.(i3) limn!þ1 �h(xn, F )¼ 0 provided infx2F f (x)¼ c.(i4) limn!þ1 �h(xn,An)¼ 0.
We say that f satisfies a weak Palais-Smale condition at the level a2 IR when for
some h as in (a4) one has
ðPSÞha Every sequence {xn}�X such that
limn!þ1
f ðxnÞ ¼ a and limn!þ1
ð1þ hðkxnkÞÞmf ðxnÞ ¼ 0 ð2:7Þ
possesses a convergent subsequence.
Remark 3.1 If h(�) 0 then ðPSÞha reduces to the usual Palais–Smale condition for
locally Lipschitz continuous functions [3]. Setting h(�) :¼ �, �2 [0,þ1[, we obtain a
non-smooth version, previously introduced in [4], of the so-called Cerami
compactness assumption [5, Section 13.1].The following critical point result is an almost direct but meaningful consequence
of Theorem 3.1.
4 G. Bonanno and S.A. Marano
Dow
nloa
ded
by [
Uni
vers
ity o
f N
ewca
stle
(A
ustr
alia
)] a
t 13:
08 1
6 M
arch
201
4
THEOREM 3.2 Suppose (a1)–(a4) and ðPSÞhc , with c given by (2.5), hold true. Then
Kc( f ) 6¼ ;. If, moreover, infx2F f (x)¼ c then Kc( f )\F 6¼ ;.
Proof Let {xn}�X nB fulfil (i1)–(i4) of Theorem 3.1. Conclusion (i2) actually means
limn!þ1
ð1þ hðkxnkÞÞmf ðxnÞ ¼ 0: ð2:8Þ
In fact, due to [17, Lemma 1.3], for any n2 IN there exists a z�n 2X� such that
kz�nkX � � 1 and
"�1n ð1þ hðkxnkÞÞ f0ðxn; zÞ � hz
�n, zi 8z2X:
Hence,
"nð1þ hðkxnkÞÞ�1z�n 2 @f ðxnÞ,
which gives
ð1þ hðkxnkÞÞmf ðxnÞ � "nkz�nkX � � "n, n2 IN:
Now (2.8) immediately comes out from "n! 0þ. Thanks to ðPSÞhc we may thus
assume that xn! x in X, where a subsequence is considered when necessary. At this
point, (i2) and the upper semi-continuity of f 0 yield f 0(x; z)� 0 for all z2X, namely
x2Kc( f ), because f (x)¼ c by (i1). Next, suppose that infx2F f (x)¼ c. On account of
( p3) the set F turns out to be �h-closed. So, (i3) forces x2F, i.e. Kc( f )\F 6¼ ;. g
We are in a position now to establish our first structure result, which extends
Corollary 4.14 of [7] to the framework of this article.
THEOREM 3.3 Let (a1), (a2), (a4) and ðPSÞhc , with c as in (2.5), be satisfied. Assume
also that
(a5) the members of F are path-wise connected and contain B,(a6) supx2B f (x)5c.
Then Kc( f ) possesses a nonlocal minimum point.
Proof Choose F :¼ @f c. For every A2F one has A\ f c 6¼ ;, besides
ðA \ F Þ n B 6¼ ;: ð2:9Þ
Indeed, (2.5) implies maxx2A f (x)� c. Thus, the first claim is obvious. To verify (2.9)
we first observe that (A\F ) nB¼A\F, because B�X n f c on account of (a6).
If A\F¼; for some A2F then A\ int( f c) 6¼ ; and A� int( f c)[ (X n f c). Since A is
path-wise connected we get A� f c, which implies A\B¼;. However, this
contradicts (a5). We next have
c � infx2f c
f ðxÞ � infx2F
f ðxÞ � infx2A\F
f ðxÞ � maxx2A
f ðxÞ 8A2F ,
namely, c¼ infx2F f (x). Hence, by (a6), inequality (2.4) holds, and hypothesis (a3) is
satisfied. Consequently, thanks to Theorem 3.2, Kc( f )\F 6¼ ;. Pick x2Kc( f )\F.
Bearing in mind the choice of F, for every �40 there exists an x�2B(x, �) such that
f (x�)5c¼ f (x), i.e. x is not a local minimum for f. g
Applicable Analysis 5
Dow
nloa
ded
by [
Uni
vers
ity o
f N
ewca
stle
(A
ustr
alia
)] a
t 13:
08 1
6 M
arch
201
4
Theorem 3.2 produces also the following non-smooth version of the famous
Corollary 1 in [8].
THEOREM 3.4 Suppose (a1), (a4) and ðPSÞha, a2 IR, hold true. If x0, x12X are two
local minima of f then it has at least three critical points.
Proof Obviously, x0 and x1 turn out to be critical points for f. It is not restrictive to
assume that f (x1)� f (x0)¼ 0. Let r40 satisfy
r5 kx1 � x0k, infx2Bðx0, rÞ
f ðxÞ � f ðx0Þ ¼ 0: ð2:10Þ
Define Q :¼ [x0, x1], Q0 :¼ {x0,x1}, �0 :¼ idjQ0, � :¼ {� 2C0(Q,X ) : �jQ0
¼ �0},F :¼ {�(Q) : � 2�} and B :¼ {x0, x1}. Hypothesis (a2) holds, because F is a
homotopy-stable family with extended boundary B; see Example 3.1. Setting F :¼
@B(x0, r) one evidently has (A\F ) nB 6¼ ; for all A2F as well as
supx2B
f ðxÞ ¼ maxf f ðx0Þ, f ðx1Þg ¼ 0 � infx2F
f ðxÞ, ð2:11Þ
which yields (a3). Since ðPSÞhc , with
c :¼ inf�2�
maxx2�ð½x0,x1Þ
f ðxÞ, ð2:12Þ
is true too, all the assumptions of Theorem 3.2 are satisfied. So, Kc( f ) 6¼ ; and
Kc( f )\F 6¼ ; provided infx2F f (x)¼ c. On account of (2.11) the conclusion is obvious
when infx2F f (x)5c. Otherwise, it easily follows from Kc( f )\F 6¼ ;, the choice of F,
and (2.10). g
Combining [13, Theorem 3.1] with Theorem 3.4 we obtain the two multiplicity
results below, where X is supposed to be reflexive, �, � :X! IR are locally Lipschitz
continuous,
f� :¼ �� ��, �2 IR, ð2:13Þ
and, for any r, r1, r24infx2X �(x) such that r15r2,
’ð1ÞðrÞ :¼ infx2Xn�r
supz2Xn�r�ðzÞ ��ðxÞ
r��ðxÞ,
’ð2ÞðrÞ :¼ supx2Xn�r
�ðxÞ � supz2�r�ðzÞ
�ðxÞ � ras soon as r5 sup
x2X�ðxÞ,
’1ðr1, r2Þ :¼ maxf’ð1Þðr1Þ, ’ð1Þðr2Þg,
’2ðr1, r2Þ :¼ infx2Xn�r1
supz2�r1 n�r2
�ðzÞ ��ðxÞ
�ðzÞ ��ðxÞ:
THEOREM 3.5 Let (a4) be fulfilled. If, moreover,
(a7) � is sequentially weakly lower semi-continuous and coercive,(a8) � is sequentially weakly upper semi-continuous,(a9) there exist r1, r24infx2X �(x) such that r15r2 and ’1(r1, r2)5’2(r1, r2),
6 G. Bonanno and S.A. Marano
Dow
nloa
ded
by [
Uni
vers
ity o
f N
ewca
stle
(A
ustr
alia
)] a
t 13:
08 1
6 M
arch
201
4
(a10) the function f� given by (2.13) satisfies ðPSÞha for all �40, a2 IR,
then f� possesses at least three critical points (two local minima and a nonlocal
minimum point) provided 1/’2(r1, r2)5�51/’1(r1, r2).
Remark 3.2 When
infx2X
�ðxÞ ¼ �ð0Þ ¼ �ð0Þ ¼ 0,
the same reasoningmade in the proof of [13, Corollary 3.1] (see also [14, Remark 3.11])
shows that the condition
(a9)0 for some r1, r240, x12X one has r15�(x1)5r2 as well as
supz2�r1
�ðzÞ5r1�ðx1Þ
2�ðx1Þ, sup
z2�r2
�ðzÞ5r2�ðx1Þ
2�ðx1Þ
forces (a9). Since
’1ðr1, r2Þ � max1
r1supz2�r1
�ðzÞ,1
r2supz2�r2
�ðzÞ
( ), ’2ðr1, r2Þ �
�ðx1Þ
2�ðx1Þ,
by Theorem 3.5 the function �� �� possesses at least three critical points provided
2�ðx1Þ
�ðx1Þ5 �5 min
r1supz2�r1
�ðzÞ,
r2supz2�r2
�ðzÞ
( ): ð2:14Þ
THEOREM 3.6 Assume that (a4), (a7), (a8), (a10) hold, and that
(a11) there exists r2 ]infx2X �(x), supx2X �(x)[ fulfilling ’(1)(r)5’(2)(r),(a12) f� turns out to be coercive for every �2 ]1/’(2)(r), 1/’(1)(r)[.
Then the conclusion of Theorem 3.5 is true whenever 1/’(2)(r)5�51/’(1)(r).
Proof Theorem 3.1 and Remark 3.3 in [13] yield a global minimum x0 of f�jXn�r,
because �51/’(1)(r). We next claim that f�jXn�rhas a global minimum. In fact, from
�41/’(2)(r) it follows
�ð �xÞ � ��ð �xÞ5 r� � supz2�r
�ðzÞ ð2:15Þ
for some �x2X n�r. Let �� : X! IR defined by setting, for every x2X,
��ðxÞ :¼r if �ðxÞ � r ,�ðxÞ otherwise.
�
Through (a7) we easily see that �� is sequentially weakly lower semi-continuous and
coercive. Thanks to (a8) and (a12), the function ��� �� enjoys the same properties.
Since X is reflexive, there exists an x12X satisfying
��ðx1Þ � ��ðx1Þ � ��ðxÞ � ��ðxÞ 8x2X: ð2:16Þ
Thus, the assertion is verified once we show that x12X n�r. If, on the contrary,
x12�r then, in view of (2.16),
r� ��ðx1Þ � �ð �xÞ � ��ð �xÞ: ð2:17Þ
Applicable Analysis 7
Dow
nloa
ded
by [
Uni
vers
ity o
f N
ewca
stle
(A
ustr
alia
)] a
t 13:
08 1
6 M
arch
201
4
Due to (2.15) one has
�ð �xÞ � ��ð �xÞ5 r� ��ðx1Þ: ð2:18Þ
Gathering (2.17) and (2.18) together gives
r� ��ðx1Þ5 r� ��ðx1Þ,
which is evidently impossible. Hence, the function f� possesses two local minima, x0and x1. Since it is locally Lipschitz continuous, the conclusion directly follows from
Theorem 3.4. g
Remark 3.3 Simple calculations guarantee that the condition(a11)
0 There exist r2 IR, x0,x12X fulfilling �(x0)5r5�(x1) and, moreover,
supz2�r
�ðzÞ5ð�ðx1Þ � rÞ�ðx0Þ þ ðr��ðx0ÞÞ�ðx1Þ
�ðx1Þ ��ðx0Þ
implies (a11). In particular, if �(0)¼�(0)¼ 0 then (a11)0 obviously reduces to
(a11)00 For some r40, x12X one has r5�(x1) as well as
supz2�r
�ðzÞ5 r�ðx1Þ
�ðx1Þ: ð2:19Þ
In this case, by Theorem 3.6, the function �� �� possesses at least three critical
points provided
�ðx1Þ
�ðx1Þ5 �5
r
supz2�r�ðzÞ
: ð2:20Þ
Indeed, since ’ð1ÞðrÞ � 1r supz2�r
�ðzÞ while, due to (2.19),
’ð2ÞðrÞ ��ðx1Þ � supz2�r
�ðzÞ
�ðx1Þ � r�
�ðx1Þ � r �ðx1Þ�ðx1Þ
�ðx1Þ � r¼
�ðx1Þ
�ðx1Þ,
we obtain
1
’ð2ÞðrÞ�
�ðx1Þ
�ðx1Þ5
r
supz2�r�ðzÞ�
1
’ð1ÞðrÞ:
Remark 3.4 Suppose
infx2X
�ðxÞ ¼ �ð0Þ ¼ �ð0Þ ¼ 0:
Then inequalities (2.20) written for r¼ r1 evidently are less restrictive than (2.14), i.e.
Theorem 3.6 allows a wider range of parameters �. However, in Theorem 3.5 the
coercivity of f� is not assumed.
Remark 3.5 By [18, Corollary 2.4], hypothesis (a12) holds true whenever the
function f� turns out to be bounded from below and (a10) is fulfilled for some non-
decreasing h.
Remark 3.6 Theorem 3.6 generalizes [13, Theorem 3.2] and, in the smooth case, [19,
Theorem 2.1]; see also [20, Theorem 2.1]. In fact, besides requiring weaker regularity
8 G. Bonanno and S.A. Marano
Dow
nloa
ded
by [
Uni
vers
ity o
f N
ewca
stle
(A
ustr
alia
)] a
t 13:
08 1
6 M
arch
201
4
and compactness conditions on the involved functionals, it exhibits a well-determined interval of parameters �, which contains those of the above-mentionedresults. Accordingly, their applications to nonlinear differential problems (as forinstance Dirichlet, Neumann, mixed, Hamiltonian, p-Laplacian problems) might beimproved through Theorem 3.6.
Remark 3.7 Let us explicitly point out a useful sufficient condition in order that �
and � be weakly continuous on X:
There is a Banach space ~X such that X compactly embeds in ~X and �,� : ~X! IRare locally Lipschitz continuous on the whole ~X.
Finally, a critical point x2X is called a saddle point for f when to every �40there correspond x0, x00 2B(x, �) such that f (x0)5f (x)5f (x00). Arguing as in theproofs of Theorem 2 and Corollary 3 of [21], but with Theorem 3.2 in place ofGhoussoub–Preiss min–max principle (see also [11, Theorem 4.3]), yields the nextresult.
THEOREM 3.7 Suppose X is infinite dimensional and x0, x12X. If (a1), (a4) and ðPSÞhc ,
with c as in (2.12), hold true, while max{ f (x0), f (x1)}5c, then Kc( f ) contains asaddle point.
References
[1] R. Livrea and S.A. Marano, A min–max principle for non-differentiable functions with a
weak compactness condition, Commun. Pure Appl. Anal. 8 (2009), pp. 1019–1029.[2] N. Ghoussoub, A min–max principle with a relaxed boundary condition, Proc. Amer.
Math. Soc. 117 (1993), pp. 439–447.[3] K.-C. Chang, Variational methods for nondifferentiable functions and their applications to
partial differential equations, J. Math. Anal. Appl. 80 (1981), pp. 102–129.
[4] N.C. Kourogenis and N.S. Papageorgiou, Nonsmooth critical point theory and nonlinear
elliptic equations at resonance, J. Aust. Math. Soc. Ser. A 69 (2000), pp. 245–271.
[5] Y. Jabri, The Mountain Pass Theorem: Variants, Generalizations and some Applications,
Encyclopedia of Mathematics and its Applications, Cambridge University Press,
Cambridge, 2003.
[6] I. Ekeland, Convexity Methods in Hamiltonian Mechanics, Ergeb. Math. Grenzgeb. 3,
Vol. 19, Springer, Berlin, 1990.
[7] N. Ghoussoub, Duality and Perturbation Methods in Critical Point Theory, Cambridge
Tracts in Mathematics, Vol. 107, Cambridge University Press, Cambridge, 1993.[8] P. Pucci and J. Serrin, A mountain pass theorem, J. Differ. Equ. 63 (1985), pp. 142–149.[9] P. Pucci and J. Serrin, Extensions of the mountain pass theorem, J. Funct. Anal. 59 (1984),
pp. 185–210.[10] G. Barletta and S.A. Marano, The structure of the critical set in the mountain-pass theorem
for nondifferentiable functions, Differ. Integral Equ. 16 (2003), pp. 1001–1012.[11] R. Livrea and S.A. Marano, Existence and classification of critical points for
nondifferentiable functions, Adv. Differ. Equ. 9 (2004), pp. 961–978.
[12] D. Motreanu and P.D. Panagiotopoulos,Minimax Theorems and Qualitative Properties of
the Solutions of Hemivariational Inequalities, Nonconvex Optimization and its
Applications, Vol. 29, Kluwer, Dordrecht, 1998.[13] G. Bonanno and P. Candito, Non-differentiable functionals and applications to elliptic
problems with discontinuous nonlinearities, J. Differ. Equ. 224 (2008), pp. 3031–3959.
Applicable Analysis 9
Dow
nloa
ded
by [
Uni
vers
ity o
f N
ewca
stle
(A
ustr
alia
)] a
t 13:
08 1
6 M
arch
201
4
[14] B. Ricceri, On a three critical points theorem, Arch. Math. (Basel) 75 (2000), pp. 220–226.[15] B. Ricceri, A three critical points theorem revisited, Nonlinear Anal. 70 (2009),
pp. 3084–3089.[16] F.H. Clarke, Optimization and Nonsmooth Analysis, Classics in Applied Mathematics,
Vol. 5, SIAM, Philadelphia, 1990.[17] A. Szulkin, Minimax principles for lower semicontinuous functions and applications to
nonlinear boundary value problems, Ann. Inst. Henri Poincare 3 (1986), pp. 77–109.
[18] D. Motreanu, V.V. Motreanu, and D. Pasca, A version of Zhong’s coercivity result fora general class of nonsmooth functionals, Abstract and Applied Analysis 7 (2002),pp. 601–612.
[19] D. Averna and G. Bonanno, A three critical points theorem and its applications to ordinaryDirichlet problems, Topol. Methods Nonlinear Anal. 22 (2003), pp. 93–103.
[20] G. Bonanno, Some remarks on a three critical points theorem, Nonlinear Anal. 54 (2003),
pp. 651–665.[21] G. Fang, The structure of the critical set in the general mountain pass principle, Ann. Fac.
Sci. Toulouse Math. 3 (1994), pp. 345–362.
10 G. Bonanno and S.A. Marano
Dow
nloa
ded
by [
Uni
vers
ity o
f N
ewca
stle
(A
ustr
alia
)] a
t 13:
08 1
6 M
arch
201
4