Volume 2009, Article ID 439176,22pages doi:10.1155/2009/439176
Research Article
Fixed Point Theory for Admissible Type Maps with Applications
Ravi P. Agarwal
1and Donal O’Regan
21Department of Mathematical Sciences, Florida Institute of Technology, Melbourne, FL 32901, USA
2Department of Mathematics, National University of Ireland, Galway, Ireland
Correspondence should be addressed to Ravi P. Agarwal,agarwal@fit.edu Received 8 December 2008; Accepted 18 June 2009
Recommended by Marlene Frigon
We present new Leray-Schauder alternatives, Krasnoselskii and Lefschetz fixed point theory for multivalued maps between Fr´echet spaces. As an application we show that our results are directly applicable to establish the existence of integral equations over infinite intervals.
Copyrightq2009 R. P. Agarwal and D. O’Regan. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
1. Introduction
In this paper, assuming a natural sequentially compact condition we establish new fixed point theorems for Urysohn type maps between Fr´echet spaces. InSection 2we present new Leray- Schauder alternatives, Krasnoselskii and Lefschetz fixed point theory for admissible type maps. The proofs rely on fixed point theory in Banach spaces and viewing a Fr´echet space as the projective limit of a sequence of Banach spaces. Our theory is partly motivated by a variety of authors in the literaturesee1–6and the references therein.
Existence inSection 2is based on a Leray-Schauder alternative for Kakutani mapssee 4,5,7for the history of this resultwhich we state here for the convenience of the reader.
Theorem 1.1. LetBbe a Banach space,Uan open subset ofB,and 0∈U. SupposeT :U → CKB is an upper semicontinuous compact (or countably condensing) map (hereCKBdenotes the family of nonempty convex compact subsets ofB). Then either
A1T has a fixed point inUor
A2there existsu∈∂U(the boundary ofUinB) andλ∈0,1withu∈λTu.
Existence inSection 2will also be based on the topological transversality theoremsee 5,7for the history of this resultwhich we now state here for the convenience of the reader.
LetBbe a Banach space andUan open subset ofB.
Definition 1.2. We letF ∈ KU, Bdenote the set of all upper semicontinuous compact or countably condensingmapsF:U → CKE.
Definition 1.3. We letF∈K∂UU, BifF∈KU, Bwithx /∈Fxforx∈∂U.
Definition 1.4. A mapF∈K∂UU, Bis essential inK∂UU, Bif for everyG∈K∂UU, Bwith G|∂U F|∂Uthere existsx∈Uwithx∈G x. OtherwiseF is inessential inK∂UU, B i.e., there exists a fixed point freeG∈K∂UU, BwithG|∂UF|∂U.
Definition 1.5. F, G ∈ K∂UU, B are homotopic inK∂UU, B, written F ∼ G inK∂UU, B, if there exists an upper semicontinuous compactor countably condensingmapN : U× 0,1 → CKBsuch thatNtu Nu, t : U → CKB belongs toK∂UU, Bfor each t∈0,1andN0FwithN1G.
Theorem 1.6. LetBandUbe as above and letF ∈ K∂UU, B. Then the following conditions are equivalent:
iFis inessential inK∂UU, B;
iithere exists a mapG∈K∂UU, Bwithx /∈Gxforx∈UandF ∼GinK∂UU, B.
Theorem 1.6immediately yields the topological transversality theorem for Kakutani maps.
Theorem 1.7. LetBand Ube as above. Suppose thatF andG are two maps inK∂UU, Bwith F∼GinK∂UU, B. ThenFis essential inK∂UU, Bif and only ifGis essential inK∂UU, B.
Also existence in Section 2 will be based on the following result of Petryshyn 8, Theorem 3.
Theorem 1.8. LetEbe a Banach space and letC⊆Ebe a closed cone. LetUandVbe bounded open subsets inEsuch that 0 ∈ U ⊆ U ⊆ V and let F : W → CKCbe an upper semicontinuous, k-set contractive (countably) map; here 0≤k <1,W V∩CandWdenotes the closure ofWinC.
Assume that
1y ≥ x ∀y∈Fxandx∈∂Ωandy ≤ x ∀y∈Fxandx∈∂W(hereΩ U∩C and∂W denotes the boundary ofWinC) or
2y ≤ x ∀y∈F xandx∈∂Ωandy ≥ x ∀y∈Fxandx∈∂W.
ThenFhas a fixed point inW\Ω.
Also inSection 2we consider a class of maps which contain the Kakutani maps.
Suppose that X and Y are Hausdorfftopological spaces. Given a class X of maps, XX, Ydenotes the set of mapsF :X → 2Y nonempty subsets ofYbelonging toX, and Xcthe set of finite compositions of maps inX. A classUof maps is defined by the following properties:
iUcontains the classCof single-valued continuous functions;
iieachF∈ Ucis upper semicontinuous and compact valued;
iiifor any polytopeP,F ∈ UcP, Phas a fixed point, where the intermediate spaces of composites are suitably chosen for eachU.
Definition 1.9. F ∈ UκcX, Yif for any compact subsetKof X, there is aG ∈ UcK, Ywith Gx⊆Fxfor eachx∈K.
The classUκc is due to Park9and his papers include many examples in this class.
Examples ofUκc maps are the Kakutani maps, the acyclic maps, the approximable maps, and the maps admissible in the sense of Gorniewicz.
Existence in Section 2is based on a Leray-Schauder alternative10which we state here for the convenience of the reader.
Theorem 1.10. Let E be a Banach space, Uan open convex subset of E,and 0 ∈ U. Suppose F ∈ UκcU, Eis an upper semicontinuous countably condensing map withx /∈λFxforx∈∂Uand λ∈0,1. ThenFhas a fixed point inU.
Also existence inSection 2will be based on some Lefschetz type fixed point theory. Let X, Y,and Γbe Hausdorfftopological spaces. A continuous single-valued mapp:Γ → Xis called a Vietoris mapwrittenp:Γ⇒Xif the following two conditions are satisfied:
ifor eachx∈X, the setp−1xis acyclic,
iipis a proper map, that is, for every compactA⊆Xone has thatp−1Ais compact.
LetDX, Y be the set of all pairsX ⇐p Γ →q Y wherepis a Vietoris map and qis continuous. We will denote every such diagram byp, q. Given two diagramsp, qand p, q, where X ⇐p Γ →q Y, we write p, q ∼ p, qif there are maps f : Γ → Γ and g:Γ → Γsuch thatq◦fq,p◦fp,q◦gq,andp◦gp. The equivalence class of a diagramp, q∈DX, Ywith respect to∼is denoted by
φ
X⇐p Γ−→q Y
:X−→Y 1.1
orφ p, qand is called a morphism fromX toY. We letMX, Ybe the set of all such morphisms. For anyφ ∈MX, Ya setφx qp−1xwhereφ p, qis called an image ofxunder a morphismφ.
Consider vector spaces over a fieldK. Let Ebe a vector space andf : E → E an endomorphism. Now letNf {x∈E:fnx 0 for some n}wherefnis thenth iterate of f, and letE E\Nf. SincefNf ⊆ Nfone has the induced endomorphism f: E → E. We call f admissible if dimE < ∞; for suchf we define the generalized trace Trfoffby putting Trf trf where tr stands for the ordinary trace.
Letf {fq} : E → Ebe an endomorphism of degree zero of a graded vector space E {Eq}. We callf a Leray endomorphism ifiallfqare admissible andiialmost allEq are trivial. For suchfwe define the generalized Lefschetz numberΛfby
Λ f
q
−1qTr fq
. 1.2
LetHbe the ˘Cech homology functor with compact carriers and coefficients in the field of rational numbers K from the category of Hausdorfftopological spaces and continuous maps to the category of graded vector spaces and linear maps of degree zero. Thus HX {HqX} is a graded vector space, with HqXbeing the q-dimensional ˘Cech homology group with compact carriers of X. For a continuous map f :X → X, Hf is the induced linear map f{fq} where fq :HqX → HqX.
The ˘Cech homology functor can be extended to a category of morphismssee11, page 364and also note that the homology functor H extends over this category, that is, for a morphism
φ
X⇐p Γ−→q Y
:X −→Y 1.3
we define the induced map
H φ
φ:HX−→HY 1.4
by putting φq◦p−1 .
Let φ : X → Y be a multivalued mapnote for eachx ∈ X we assumeφx is a nonempty subset ofY. A pair p, qof single valued continuous maps of the form X ←p Γ →q Yis called a selected pair ofφwrittenp, q⊂φif the following two conditions hold:
ipis a Vietoris map,
iiqp−1x⊂φxfor anyx∈X.
Definition 1.11. An upper semicontinuous compact mapφ :X → Y is said to be admissible and we writeφ∈AdX, Yprovided that there exists a selected pairp, qofφ.
Definition 1.12. An upper semicontinuous mapφ : X → Y is said to be admissible in the sense of Gorniewiczand we writeφ ∈ADX, Yprovided that there exists a selected pair p, qofφ.
Definition 1.13. A mapφ ∈ AdX, Xis said to be a Lefschetz map if for each selected pair p, q ⊂ φ the linear mapq p−1 : HX → HX the existence of p−1 follows from the Vietoris theoremis a Leray endomorphism.
Ifφ:X → Xis a Lefschetz map, we define the Lefschetz setΛφ orΛXφby Λ
φ
Λ
qp−1 : p, q
⊂φ
. 1.5
Definition 1.14. A Hausdorfftopological spaceXis said to be a Lefschetz space provided that everyφ∈AdX, Xis a Lefschetz map andΛφ/{0}that impliesφhas a fixed point.
Also we present Krasnoselskii compression and expansion theorems inSection 2in the Fr´echet space setting. Let E E,| · | be a normed linear space and C⊆E a closed cone.
For r >0 let BC0, r {x∈C: |x| ≤r} and it is well known that BC0, r B0, R∩C where B0, r {x∈E: |x| ≤r}. Our next result,Theorem 1.8, was established in12and Theorem 1.10can be found in13.
Theorem 1.15. LetE E,| · |be a normed linear space,C⊆Ea closed cone,r,Rconstants, and 0< r < R. Suppose thatF∈ UκcB0, R∩C, Cis compact with
⎧⎨
⎩
y≥ |x| ∀y∈Fx, x∈∂BC0, r,
y≤ |x| ∀y∈Fx, x∈∂BC0, R. 1.6
ThenFhas a fixed point in BCr,R{x∈C:r≤ x ≤R}.
Theorem 1.16. LetE E,| · |be a normed linear space,C⊆Ea closed cone,r,Rconstants, and 0< r < R. Suppose thatF∈ADC, Cis completely continuous with
⎧⎨
⎩
y≤ |x| ∀y∈Fx, x∈∂BC0, r,
y≥ |x| ∀y∈Fx, x∈∂BC0, R. 1.7
ThenFhas a fixed point in BCr,R.
Now letI be a directed set with order≤and let {Eα}α∈Ibe a family of locally convex spaces. For eachα∈I, β∈Ifor which α≤βlet πα,β:Eβ → Eα be a continuous map. Then the set
x xα∈
α∈I
Eα:xαπα,β
xβ
∀α, β∈I, α≤β
1.8 is a closed subset of
α∈IEα and is called the projective limit of {Eα}α∈I and is denoted by lim←Eα or lim←{Eα, πα,β} or the generalized intersection14, page 439
α∈IEα.
2. Fixed Point Theory in Fr ´echet Spaces
Let E E,{| · |n}n∈N be a Fr´echet space with the topology generated by a family of seminorms {| · |n : n ∈ N}; here N {1,2, . . .}. We assume that the family of seminorms satisfies
|x|1≤ |x|2≤ |x|3 ≤ · · · for everyx∈E. 2.1
A subset X of E is bounded if for every n∈N there exists rn >0 such that|x|n ≤ rn
for all x ∈X. For r >0 and x ∈E we denote Bx, r {y ∈E : |x−y|n ≤ r ∀n ∈N}.
To E we associate a sequence of Banach spaces {En,| · |n} described as follows. For every n∈Nwe consider the equivalence relation∼ndefined by
x∼ny iffx−y
n0. 2.2
We denote by En E/∼n,| · |n the quotient space, and by En,| · |n the completion of En with respect to| · |nthe norm on En induced by| · |nand its extension toEnis still denoted by
| · |n. This construction defines a continuous map μn :E → En. Now since2.1is satisfied the seminorm| · |n induces a seminorm on Em for every m ≥ n again this seminorm is denoted by| · |n. Also2.2defines an equivalence relation on Em from which we obtain a continuous map μn,m : Em → En since Em/∼n can be regarded as a subset of En. Now μn,mμm,k μn,k if n≤ m ≤ k and μn μn,mμm if n ≤ m. We now assume the following condition holds:
⎧⎨
⎩
for eachn∈N, there exists a Banach spaceEn,|·|n and an isomorphism
between normed spaces
jn:En−→En. 2.3 Remark 2.1. iFor convenience the norm on En is denoted by| · |n.
iiIn our applications EnEn for each n∈N.
iiiNote if x∈En orEnthen x∈E. However if x∈En then x is not necessaily in E and in fact En is easier to use in applicationseven though En is isomorphic to En. For example if EC0,∞, then En consists of the class of functions in E which coincide on the interval0, nand EnC0, n.
Finally we assume
⎧⎨
⎩
E1⊇E2 ⊇ · · ·and for eachn∈N, jnμn,n1jn1−1 x
n≤ |x|n1 ∀x∈En1. 2.4
here we use the notation from14, i.e., decreasing in the generalized senseLet lim←En
or ∞
1En where ∞
1 is the generalized intersection 14denote the projective limit of {En}n∈N note πn,m jnμn,mjm−1 : Em → En for m ≥ n and note lim←En ∼ E, so for convenience we write Elim←En.
For each X ⊆E and each n∈N we set Xn jnμnX, and we let Xn, intXn and
∂Xn denote, respectively, the closure, the interior, and the boundary of Xn with respect to
| · |nin En. Also the pseudointerior of X is defined by
pseudo−intX
x∈X:jnμnx∈Xn\∂Xn for everyn∈N
. 2.5
The setXis pseudoopen ifX pseudo−intX. Forr > 0 andx∈En we denoteBnx, r {y∈En:|x−y|n≤r}.
We now show how easily one can extend fixed point theory in Banach spaces to applicable fixed point theory in Fr´echet spaces. In this case the mapFn will be related to F by the closure property2.11.
Theorem 2.2. LetEand En be as described above, X a subset of E and F : Y → 2E where intXn ⊆Yn for each n∈N. Also for each n∈N assume that there exists Fn : intXn → 2En and suppose the following conditions are satisfied:
0∈pseudo−intX, 2.6
⎧⎨
⎩
for eachn∈ {2,3, . . .}ify∈intXnsolvesy∈FnyinEn thenjkμk,njn−1
y
∈intXk fork∈ {1, . . . , n−1}, 2.7
⎧⎨
⎩
for eachn∈N, Fn: intXn −→CKEn
is an upper semicontinuous countably condensing map, 2.8
for eachn∈N, y /∈λFny inEn ∀λ∈0,1, y∈∂intXn, 2.9
⎧⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎨
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎩
for any sequence yn
n∈N with yn∈intXn and yn∈Fnyn in En forn∈N and for everyk∈N there exists a subsequence Nk⊆ {k1, k2, . . .}, Nk⊆Nk−1 for
k∈ {1,2, . . .}, N0N, and a zk∈intXk with jkμk,njn−1
yn
−→zk in Ek as n−→ ∞inNk,
2.10
⎧⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎨
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎩
if there exists aw∈Y and a sequence yn
n∈N
with yn ∈intXn and yn ∈ Fnyn in En such that for everyk∈N there exists a subsequence S⊆ {k1, k2, . . .} ofN withjkμk,njn−1
yn
−→w in Ek as n−→ ∞in S, thenw∈Fw inE.
2.11
ThenFhas a fixed point inE.
Remark 2.3. Notice that to check2.10we need to show that for each k ∈ Nthe sequence {jkμk,njn−1yn}n∈Nk−1⊆intXk is sequentially compact.
Proof. From Theorem 1.1for each n ∈ N there exists yn ∈ intXn with yn ∈ Fnyn we applyTheorem 1.1with U intXn and note jnμn0 ∈ Xn\∂Xn intXn. Let us look at {yn}n∈N. Notice y1 ∈ intX1 and j1μ1,kjk−1yk ∈ int X1 for k ∈ {2,3, . . .} from2.7.
Now 2.10 with k 1 guarantees that there exists a subsequence N1 ⊆ {2,3, . . .} and a z1 ∈ int X1 with j1μ1,njn−1yn → z1 in E1 as n → ∞ in N1. Look at {yn}n∈N1. Now j2μ2,njn−1yn ∈ int X2 for k ∈ N1. Now 2.10 withk 2 guarantees that there exists a subsequence N2 ⊆ {3,4, . . .} of N1 and a z2 ∈int X2 with j2μ2,njn−1yn → z2 in E2
as n → ∞ in N2. Note from2.4and the uniqueness of limits that j1μ1,2j2−1z2 z1 inE1 since N2⊆N1 note j1μ1,njn−1yn j1μ1,2j2−1j2μ2,njn−1yn for n∈N2. Proceed inductively to obtain subsequences of integers
N1⊇N2⊇ · · ·, Nk⊆ {k1, k2, . . .}, 2.12
andzk∈int Xkwithjkμk,njn−1yn → zkinEkasn → ∞inNk.Notejkμk,k1jk1−1 zk1 zkin Ekfork∈ {1,2, . . .}.
Fix k∈N. Note
zkjkμk,k1jk1−1 zk1jkμk,k1jk1−1 jk1μk1,k2jk2−1 zk2 jkμk,k2jk2−1 zk2· · ·jkμk,mjm−1zmπk,mzm
2.13
for every m≥k. We can do this for eachk ∈N. As a resulty zk∈lim←EnEand also notey∈Y sincezk∈int Xk⊆Ykfor eachk∈N. Also since yn ∈Fn yn inEnfor n∈Nk
andjkμk,njn−1yn → zkyinEkasn → ∞inNkone has from2.11thaty∈FyinE.
Remark 2.4. From the proof we see that condition2.7can be removed from the statement of Theorem 2.2. We include it only to explain condition2.10 seeRemark 2.3.
Remark 2.5. Note that we could replace int Xn ⊆ Yn above with int Xn a subset of the closure of Yn in En if Y is a closed subset of E so in this case we can take Y X if X is a closed subset of E. To see this note zk ∈ int Xk,y zk ∈ lim←En E and πk,mym → zk in Ek as m → ∞ and we can conclude that y∈Y Y note thatq∈Y if and only if for every k ∈ N there exists xk,m ∈ Y,xk,m πk,nxn,m for n ≥ k with xk,m → jkμkq in Ek as m → ∞.
Remark 2.6. Suppose inTheorem 2.2we replace2.10with
⎧⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎨
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎩
for any sequence yn
n∈N with yn∈int Xn andyn∈Fnyn inEn forn∈Nand
for everyk∈Nthere exists a subsequence Nk⊆ {k1, k2, . . .}, Nk⊆Nk−1 for
k∈ {1,2, . . .}, N0N, and azk∈int Xk with jkμk,njn−1
yn
−→zk inEk asn−→ ∞inNk.
2.14
In addition we assume F : Y → 2E with int Xn ⊆ Yn for each n ∈ N is replaced by F :X → 2E and suppose2.11is true with w ∈Y replaced by w∈X. Then the result in Theorem 2.2is again true.
The proof follows the reasoning inTheorem 2.2except in this case zk ∈int Xk and y∈X.
Remark 2.7. In fact we could replacein fact we can remove it as mentioned inRemark 2.4 2.7inTheorem 2.2with
⎧⎨
⎩
for eachn∈ {2,3, . . .}if y∈int Xn solvesy∈FnyinEn
thenjkμk,njn−1 y
∈int Xk fork∈ {1, . . . , n−1}, 2.15
and the result above is again true.
Remark 2.8. Usually in our applications one has∂Xn ∂intXn soXn int Xn. If X is a pseudoopen subset of E then for each n∈N one hassee15that Xn is a open subset of En so int XnXn.
Essentially the same reasoning as inTheorem 2.2now usingTheorem 1.7establishes the following result. We will need the following definitions.
Let E and En be as described inSection 2. For the definitions below X ⊆E and F: Y → 2E with int Xn⊆Yn for each n∈Nor int Xn a subset of the closure of Yn in En
if Y is a closed subset of E. In addition assume for each n∈N that Fn: int Xn → 2En. Definition 2.9. We say F ∈KY, E if for each n∈None hasFn ∈ Kint Xn, En i.e., for each n ∈ N, Fn : int Xn → CKEn is an upper semicontinuous countably condensing map.
Definition 2.10. F ∈ K∂Y, EifF ∈ KY, Eand for eachn ∈ N one hasx /∈Fnxforx ∈
∂ int Xn.
Definition 2.11. F ∈K∂Y, E is essential in K∂Y, E if for each n∈ N one has thatFn ∈ K∂intXnint Xn, Enis essential in K∂intXnint Xn, En i.e., for eachn∈N, every mapGn∈ K∂intXnint Xn, EnwithGn|∂intXn Fn|∂intXnhas a fixed point in int Xn.
Remark 2.12. Note that ifjnμn0 ∈ Un for each n ∈ N then 0 ∈ K∂Y, E is essential in K∂Y, E see7.
Definition 2.13. We assume jnμn0 ∈int Xn for n∈N. F,0 ∈ K∂Y, E are homotopic in K∂Y, E, written F ∼ G in K∂Y, E, if for each n ∈ N one has Fn ∼ jnμn0 in K∂intXnint Xn, En.
Theorem 2.14. LetEandEnbe as described above,Xa subset ofEandF:Y → 2Ewhere intXn⊆ Ynfor eachn∈Nor int Xna subset of the closure ofYninEn(ifY is a closed subset ofE). Also for
eachn∈Nassume that there existsFn: intXn → 2En and supposeF ∈K∂Y, E,2.6,2.7, and the following condition holds:
F ∼0 inK∂Y, E. 2.16
Also assume2.10and2.11hold. ThenFhas a fixed point inE.
Proof. Fixn ∈ N. NowRemark 2.12guarantees that the zero mapi.e.,Gx jnμn0is essential inK∂UnUn, Enfor eachn ∈ N. NowTheorem 1.7guarantees thatFnis essential in K∂UnUn, En so in particular there exists yn∈Unwith yn ∈Fnyn. Essentially the same reasoning as inTheorem 2.2withRemark 2.5establishes the result.
Remark 2.15. Notice that2.6and2.17could be replaced by F ∼G in K∂Y, E of course we assume G∈K∂Y, E and we must specify Gn for n∈N here.
Remark 2.16. Condition2.7can be removed from the statement ofTheorem 2.14.
Remark 2.17. Note thatRemark 2.6holds in this situation also.
As an application ofTheorem 2.2we discuss the integral equation
yt ∞
0
Kt, sf s, ys
ds fort∈0,∞. 2.17
Theorem 2.18. Let 1≤p <∞be a constant and 1< q≤ ∞the conjugate top. Suppose the following conditions are satisfied:
for eacht∈0,∞, the maps−→Kt, sis measurable, 2.18
sup
t∈0,∞
∞
0
|Kt, s|qds 1/q
< ∞, 2.19
∞
0
K t, s
−Kt, sqds−→0 ast−→t, for eacht∈0,∞, 2.20
⎧⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎨
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎩
f:0,∞×R → R is aLp-Carath´eodory function:
by this one means athe mapt−→f
t, y
is measurable for ally∈R;
bthe mapy−→f t, y
is continuous for a.e. t∈0,∞;
cfor eachr >0 there existsμr ∈Lp0,∞such that y≤r impliesf
t, y≤μrtfor a.e. t∈0,∞,
2.21
⎧⎪
⎪⎪
⎪⎪
⎪⎪
⎨
⎪⎪
⎪⎪
⎪⎪
⎪⎩
there exists a continuous nondecreasing function ψ:0,∞−→0,∞and aφ∈Lp0,∞with f
s, y≤φsψy∀y∈R and a.e. s∈0,∞,
2.22
∃r >0 withr > K1ψr whereK1 sup
t∈0,∞
∞
0
φs|Kt, s|ds. 2.23
Then2.17has at least one solution in C0,∞.
Remark 2.19. One could also obtain a multivalued version ofTheorem 2.18by using the ideas in the proof below with the ideas in16.
Proof. Here E C0,∞, Ek consists of the class of functions in E which coincide on the interval0, k, Ek C0, kwith of course πn,m jnμn,mjm−1 : Em → En defined by πn,mx x|0,n. We will applyTheorem 2.2with
X{u∈C0,∞:|u|n≤r for eachn∈N}, 2.24 here |u|nsupt∈0,n|ut|. Fix n∈N and note
XnXn{u∈C0, n:|u|n≤r} 2.25
with
int Xn{u∈C0, n:|u|n< r}. 2.26 Let Fn: int Xn → En be given by
Fn yt n
0
Kt, sf s, ys
ds. 2.27
Also let Y X we will useRemark 2.5and let F :Y → E be given by Fyt
∞
0
Kt, sf s, ys
ds. 2.28
Clearly2.6and2.7hold, and a standard argument in the literature guarantees that F : int Xn → En is continuous and compact so 2.8 holds. To show2.9 fix n ∈ N and suppose that there exists x∈∂ int Xn so|x|nrand λ∈0,1 with xλFnx. Then for t∈0, none has
|xt| ≤ψ|x|n n
0
|Kt, s|φsds≤ψ|x|nK1, 2.29
so |x|n≤ψ|x|nK1, that is,r ≤ψrK1. This contradicts2.23, so2.9holds. To show2.10 consider a sequence {yn}n∈N with yn ∈C0, n,ynFnyn on0, nand |yn|n< r. Now to show2.10we will show for a fixed k∈N that {jkμk,njn−1yn}n∈S⊆int Xkis sequentially compact for any subsequence S of {k, k1, . . .}. Note for n∈S that jkμk,njn−1yn yn|0,k so {jkμk,njn−1yn}n∈S is uniformly bounded since |yn|n≤r for n∈S implies |yn|k≤r for n∈S. Also {jkμk,njn−1yn}n∈Sis equicontinuous on0, ksince for n∈ S andt, x∈0, k note there existshr ∈Lp0,∞ with|fs, yns| ≤hrs for a.e. s∈0, none has
jkμk,njn−1 ynt
−jkμk,njn−1 ynx
≤ n
0
|Kt, s−Kx, s|f
s, ynsds
≤ ∞
0
hrspds
1/p∞
0
|Kt, s−Kx, s|qds 1/q
.
2.30
The Arzela-Ascoli theorem guarantees that {jkμk,njn−1yn}n∈S ⊆ int Xk is sequentially compact. Finally we show 2.11. Suppose there exists w ∈ C0,∞ and a sequence {yn}n∈N with yn ∈ int Xn and yn Fnyn in C0, n such that for every k ∈ N there exists a subsequence S⊆ {k1, k2, . . .} of N with yn → w in C0, k as n → ∞ in S. If we show
wt ∞
0
Kt, sfs, wsds fort∈0,∞, 2.31
then 2.11holds. To see 2.31fix t ∈ 0,∞. Consider k ≥ t and n ∈ S as described above. Then ynFnyn for n∈S and so
ynt− k
0
Kt, sf
s, yns ds
n
k
Kt, sf
s, yns
ds, 2.32
so
jkμk,njn−1 ynt
− k
0
Kt, sf
s, jkμk,njn−1
yns ds ≤
n
k
|Kt, s|hrsds. 2.33
here2.21 guarantees that there exists hr ∈ Lp0,∞ with |fs, yns| ≤ hrs for a.e.
s∈0, nLet n → ∞through S and use the Lebesgue Dominated Convergence theorem to obtain
wt−
k
0
Kt, sfs, wsds ≤
∞
k
|Kt, s|hrsds 2.34
since jkμk,njn−1yn → winC0, k. Finally let k → ∞ note2.19to obtain
wt− ∞
0
Kt, sfs, wsds0. 2.35
Thus2.11holds. Our result now follows fromTheorem 2.2withRemark 2.5.
Essentially the same reasoning as inTheorem 2.2now usingTheorem 1.8establishes the following result.
Theorem 2.20. LetEandEnbe as described in the beginning ofSection 2,Ca closed cone inE,U, andV are bounded pseudoopen subsets ofEwith 0 ∈U⊆ U⊆ V, andF :Y → 2E. Also assume eitherWnVn∩Cn⊆Ynfor eachn∈N(hereWnVn∩Cn) orVn∩Cna subset of the closure of YninEn(ifY is a closed subset ofE). Also for eachn∈NassumeFn:Wn → 2En and suppose that the following conditions hold (hereΩnCn∩Un):
⎧⎨
⎩
for eachn∈ {2,3, . . .} ify∈Wn\Ωn solvesy∈Fny in En thenjkμk,njn−1
y
∈Wkfork∈ {1, . . . , n−1}, 2.36
⎧⎪
⎪⎪
⎨
⎪⎪
⎪⎩
for eachn∈N, Fn:Wn−→CK
Cn is an upper semicontinuousk-setcountablycontractive map here0≤k <1.
2.37
Also for eachn∈Nassume either
⎧⎨
⎩ y
n≥ |x|n ∀y∈Fnx, ∀x∈∂Ωn, y
n≤ |x|n ∀y∈Fnx, ∀x∈∂Wn, 2.38 or
⎧⎨
⎩ y
n≤ |x|n ∀y∈Fnx, ∀x∈∂Ωn, y
n≥ |x|n ∀y∈Fnx, ∀x∈∂Wn
2.39
hold. Finally suppose that the following hold:
⎧⎪
⎪⎪
⎨
⎪⎪
⎪⎩
for everyk∈N and any subsequenceA⊆ {k, k1, . . .}
if x∈Cn is such thatx∈Wn\Ωn for somen∈A then there exists aγ >0 withjkμk,njn−1x
k≥γ,
2.40
⎧⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎨
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎩
for any sequence yn
n∈N withyn∈Wn\Ωn
and yn∈Fnyn in En forn∈N and for everyk∈N there exists a subsequence Nk⊆ {k1, k2, . . .}, Nk⊆Nk−1 for k∈ {1,2, . . .}, N0N, and azk∈Wk with jkμk,njn−1
yn
−→zk in Ekas n−→ ∞inNk,
2.41
⎧⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎨
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎩
if there exists aw∈Y and a sequence yn
n∈N
with yn ∈Wn\Ωn and yn∈Fnyn in En such that for everyk∈N there exists a subsequence
S⊆ {k1, k2, . . .} ofN withjkμk,njn−1 yn
−→w in Ek as n−→ ∞in S, thenw∈Fw inE.
2.42
ThenFhas a fixed point in E.
Proof. Fix n∈N. We would like to applyTheorem 1.8. Note that we know from15that Cn
is a cone and Un and Vn are open and bounded with jnμn0∈Un⊆Un⊆Vn.Theorem 1.8 guarantees that there exists yn ∈Wn\Ωn with yn ∈Fnyn in En. As inTheorem 2.2there exists a subsequence N1 ⊆ {2,3, . . .} and a z1 ∈ W1 with j1μ1,njn−1yn → z1in E1 as n → ∞ in N1. Also yn ∈ Wn\Ωn together with2.40yields |j1μ1,njn−1yn|1 ≥ γ for n∈N and so|z1|1 ≥γ. Proceed inductively to obtain subsequences of integers
N1 ⊇N2⊇ · · ·, Nk⊆ {k1, k2, . . .} 2.43
and zk∈Wk with jkμk,njn−1yn → zk in Ek as n → ∞ in Nk. Note jkμk,k1jk1−1 zk1 zk in Ek for k ∈ {1,2, . . .} and |zk|k ≥ γ. Now essentially the same reasoning as in Theorem 2.2withRemark 2.5guarantees the result.
Remark 2.21. Condition2.36can be removed from the statement ofTheorem 2.20.
Remark 2.22. Note 2.40 is only needed to guarantee that the fixed point y satisfies
|jkμky|k ≥ γ for k ∈ N. If we assume all the conditions inTheorem 2.20 except 2.40 then again F has a fixed point in E but the above property is not guaranteed.
Essentially the same reasoning as inTheorem 2.2just applyTheorem 1.10in this case establishes the following result.
Theorem 2.23. LetEandEnbe as described above,Xa convex subset ofE,andF :Y → 2Ewhere intXn⊆Ynfor eachn∈Nor intXn a subset of the closure ofYninEn(ifYis a closed subset ofE).
Also for eachn∈Nassume that there existsFn: intXn → 2En and suppose that2.6,2.7,2.9, 2.10,2.11and the following condition hold:
⎧⎨
⎩
for eachn∈N, Fn∈ Uκc
intXn, En is an
upper semicontinuous countably condensing map. 2.44
ThenFhas a fixed point inE.
Proof. Fix n ∈ N. We would like to apply Theorem 1.10. Note that we know from 15 that int Xn is convex. FromTheorem 1.10 for each n ∈ N there exists yn ∈ int Xn with yn ∈Fnyn in En. Now essentially the same reasoning as inTheorem 2.2withRemark 2.5 guarantees the result.
Remark 2.24. Note Remarks2.4,2.6, and2.7hold in this situation also.
Now we present some Lefschetz type theorems in Fr´echet spaces. Let E and En be as described above.
Definition 2.25. A set A ⊆ E is said to be PRLS if for each n ∈ N, An ≡ jnμnA is a Lefschetz space.
Definition 2.26. A set A⊆E is said to be CPRLS if for each n∈N, An is a Lefschetz space.
Theorem 2.27. LetEandEnbe as described above,C ⊆Eis an PRLS, andF :C → 2E. Also for eachn∈ Nassume that there existsFn :Cn → 2En and suppose that the following conditions are satisfied:
⎧⎨
⎩
for eachn∈ {2,3, . . .} ify∈Cn solvesy∈Fnyin En thenjkμk,njn−1
y
∈Ck fork∈ {1, . . . , n−1}, 2.45
for eachn∈N, Fn∈AdCn, Cn, 2.46
for eachn∈N, ΛCnFn/{0}, 2.47
⎧⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎨
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎩
for any sequence yn
n∈N withyn∈Cn and yn∈Fnyn in En forn∈N and for every∈N there exists a subsequence Nk⊆ {k1, k2, . . .}, Nk⊆Nk−1 for k∈ {1,2, . . .}, N0N, and azk∈Ck with jkμk,njn−1
yn
−→zk in Ekas n−→ ∞inNk,
2.48
⎧⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎨
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎩
if there exists aw∈Cand a sequence yn
n∈N
with yn ∈Cn andyn∈Fnyn in En such that for everyk∈N there exists a subsequence S⊆ {k1, k2, . . .} ofN withjkμk,njn−1
yn
−→w in Ek as n−→ ∞in S, thenw∈Fw inE.
2.49
ThenFhas a fixed point inE.
Proof. For each n ∈ N there exists yn ∈ Cn.Now the same reasoning as in Theorem 2.2 guarantees the result.
Remark 2.28. Condition2.45can be removed from the statement ofTheorem 2.27.
Remark 2.29. Suppose inTheorem 2.27, one has
⎧⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎨
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎩
for any sequence yn
n∈N with yn∈Cn andyn∈Fnyn inEn forn∈N and
for everyk∈N there exists a subsequence Nk⊆ {k1, k2, . . .}, Nk⊆Nk−1 for k∈ {1,2, . . .}, N0 N, and azk∈Ck with jkμk,njn−1
yn
−→zk inEkasn−→ ∞inNk
2.50
instead of2.48and F:C → 2E is replaced by F:Y → 2E with C⊆Y andCn ⊆Yn for each n∈Nand suppose that2.49is true with w∈C replaced by w∈Y. Then the result inTheorem 2.27is again true.
In fact we could replace Cn ⊆ Yn above with Cn a subset of the closure of Yn in En if Y is a closed subset of E so in this case we can take Y C if C is a closed subset of E.
In fact in this remark we could replacein fact we can remove it as mentioned in Remark 2.4 2.45with
⎧⎨
⎩
for eachn∈ {2,3, . . .}if y∈Cn solvesy∈FnyinEn thenjkμk,njn−1
y
∈Ckfork∈ {1, . . . , n−1} 2.51
and the result above is again true.