MAJOR POSTGRADUATE PROJECT ON

Topological Methods In Differential

Equations
INTERIM PROGRESS REPORT

to be submitted by

Roll No. : V22003

under the guidance of

Dr. Syed Abbas
Dr. Samir Shukla

for the award of the degree of

MASTER OF SCIENCE IN APPLIED MATHEMATICS

School of Mathematical and Statistical Science

Indian Institute of Technology Mandi

March 2024
Contents
1 Introduction 2

2 Some Other Applications of the Brouwer Degree 2

2.1 Borsuk’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Borsuk-Ulam theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.3 Jordan separation theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.4 Open mapping theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.5 Argument principal as a special case of degree of holomorphic map . . . . . 7

3 Degree Theory for Condensing Operators 8

3.1 Leray-Schauder degree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.2 Uniqueness of leray-schauder degree . . . . . . . . . . . . . . . . . . . . . . 9
3.3 Condensing and countably condensing map . . . . . . . . . . . . . . . . . . 11
3.4 Housedorff metric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

4 Degree Theory for Countably Condensing Mapping 13

4.1 Defining degree for a special class of mappings . . . . . . . . . . . . . . . . 14
4.2 Properties of the degree theory defined for countably condensing map . . . 15
4.3 Some fixed point theorems involving countably condensing map . . . . . . . 15

5 Applications to ODEs 17
5.1 Solvability result for a type of differential equations in Banach spaces . . . 17
5.2 Solvability result for a type of differential equation in Hilbert spaces . . . . 19

1
1 Introduction
Degree theory, a fundamental concept in mathematics, plays a crucial role in various
branches of the field. This report delves into the extensive applications of the Brouwer
degree, exploring its significance in different mathematical contexts. The Brouwer degree, a
key tool in algebraic topology, is instrumental in understanding and analyzing a wide range
of theorems and concepts.
The report begins by examining several important theorems that utilize the Brouwer
degree, including Borsuk’s theorem, the Borsuk-Ulam theorem, the Jordan separation the-
orem, the Open mapping theorem, and the connection between the winding number and
degree. These theorems highlight the versatility and applicability of the Brouwer degree in
diverse mathematical scenarios.
Moving forward, the report delves into Degree Theory for Condensing Operators, focus-
ing on essential topics such as the Leray-Schauder degree, uniqueness of the Leray-Schauder
degree, condensing and countably condensing map, and the Housdorff metric. These dis-
cussions provide a deeper insight into the properties and applications of degree theory in
operator theory and functional analysis.
Furthermore, an exploration of Degree Theory for Countably Condensing Mapping is
undertaken, elucidating the construction of the degree for countably condensing mappings,
properties associated with this theory, and fixed point theorems involving countably con-
densing mappings. This section aims to showcase the broad scope and implications of degree
theory in the context of countably condensing mappings.
In addition to theoretical discussions, this report investigates the practical applications
of degree theory in Ordinary Differential Equations (ODEs). By presenting solvability
results for specific types of differential equations in normed space which is also complete
spaces and Hilbert spaces.

2 Some Other Applications of the Brouwer Degree

Proposition 2.0.1. [1] Let Yj , j = 1, 2, . . ., be acollection of open sets which are disjoint
/ ψ Ȳ \ ∞
S
and contained in the interior of Y . Let x0 ∈ j=1 Yj ; then deg (ψ, Yj , x0 ) is zero
P
except for finitely many i, and deg (ψ, Y, x0 ) = j deg (ψ, Yj , x0 ).

Proof. SSince ψ
Ȳ \ ∞
S

i=1 Yj is closed, there exists a neighborhood N of x0 disjoint from
ψ Ȳ \ ∞ i=1 Yj , and let y be a regular value in N ; then

deg (ψ, Y, x0 ) = deg(ψ, Y, y), deg (ψ, Yj , x0 ) = deg (ψ, Yj , y) .

Since y has a finite number of preimages, ψ −1 (y) is contained in a finite number of the
Yj ’s, we can conclude the result.

Proposition 2.0.2 (Excision). Let C is a closed set contained in X̄ and y0 ∈

/ ψ(C)∪ψ(∂X),
then

2
 deg (ψ, X, y0 ) = deg (ψ, X\C, y0 ) .

Proof. Proof: Let X1 = X\C and apply the proposition 2.0.2.

2.1 Borsuk’s theorem

Theorem 2.1.1 (Borsuk’s Theorem). [2] If M ⊂ Rn is a bounded and symmetric set with
0 ∈ M , which is also open, and ψ ∈ C(M ) is an odd function such that 0 ∈
/ ψ(∂M ), then
the degree of ψ at 0 in M is odd.

Proof. WLOG, suppose that ψ ∈ C 1 (M̄ ) with Jψ (0) ̸= 0 because we know C 1 (M̄ ) is dense
C(M̄ ). Next, we define a mapping ζ ∈ C 1 (M̄ ) sufficiently close to ψ by induction as follows:
Let γ ∈ C 1 (R) be an odd mapping with γ ′ (0) = 0 and γ(t) = 0 if and only if t = 0. Put
ψ(z)
Mk = {z ∈ M : zk ̸= 0} and φ(z) = γ(z 1)
for all z ∈ M1 . Choose |p1 | in such a way that
p1 becomes regular value for φ on M1 . Put ζ1 (z) = ψ(z) − γ (z1 ) p1 , then 0 will be regular
value for ζ1 on M1 .
Assume that we already have an odd map ζk ∈ C 1 (M̄ ) close to ψ such that it admits
0 as a regular value for ζk on Mk . Then we set ζk+1 (z) = ζk (z) − γ (zk+1 ) pk+1 with |pk+1 |
suitably small such that 0 is regular value for ζk+1 on Mk+1 .
If z ∈ Mk+1 and zk+1 = 0, then

′
z ∈ Mk , ζk+1 (z) = ζk (z), ζk+1 (z) = ζk′ (z),
and hence Jζk+1 (z) ̸= 0. By induction, we also have ζn′ (0) = ζ1′ (0) = ψ ′ (0) and so 0 is a
regular value for ζn . By Definition of degree and gn is sufficiently close to ψ, we know that
X
deg(ψ, M, 0) = deg (ζn , M, 0) = sgn Jζn (0) + sgn Jζn (z),
z∈ζ −1 (0),z̸=0

and thus deg(ψ, M, 0) is odd.

2.2 Borsuk-Ulam theorem

Theorem 2.2.1. The Borsuk-Ulam Theorem states that for any continuous function γ :
S n → Rn , there exists a point x such that γ(x) = γ(−x), where S n is the n-dimensional
sphere.

Proof. Given a function γ(x) = ζ(x) − ζ(−x), where γ is an odd function, we seek to
prove the existence of an x such that γ(x) = 0. If this is not true, we can normalize γ(x) by
dividing it by its absolute value to obtain a new function γ : S n → S n−1 , which also remains
odd. The restriction of this new γ to the equator S n−1 then has an odd degree according to
Borsuk’s Theorem 2.1.1. However, this restriction is homotopically trivial when restricted

3
to one of the hemispheres bounded by S n−1 , leading to a contradiction, thereby concluding
the theorem.

Let θ ⊂ Rn be a bounded open set and let ψ ∈ C(θ̄). Due to the homotopy invariance
property of the degree of mapping deg(ψ, θ, y), we can conclude that this degree is an integer
that remains constant as y varies within the connected component O of Rn \ ψ(∂θ). Hence,
it is appropriate to represent this integer as deg(ψ, θ, O). The connected component that is
unbounded is symbolized as O∞ . Consequently, we obtain the following formula:

Theorem 2.2.2. [2] Let θ ⊂ Rn be a bounded subsets which are open as well, ψ ∈ C(θ),
ϕ ∈ C(Rn ), and let Oi be the connected bounded subsets of Rn \ψ(∂θ). If q ∈
/ (gf )(∂θ), then
X
deg(ϕψ, θ, q) = deg (ψ, θ, Oj ) deg (ϕ, Oj , q) , (2.2.1)
j

where we have only finite terms which are not zero.

Proof. We initially demonstrate that 2.2.1 contains only a finite number of non-zero terms..
Let r > 0 be chosen such that the image of θ̄ under ψ is contained in the ball Br (0).
Consequently, we can deduce that the set Γ = Br (0) ∩ ϕ−1 (q) is compact. Moreover, Γ is a
subset of Rn \ψ(∂θ), which can be written as as the union of sets Oi for i ≥ 1. There exist a
finite number of indices i = 1, 2, · · · , t such that the union of Oi for i = 1 to t + 1 contains
Γ, where Ot+1 = O∞ ∩ Br+1 .. We have

deg (ψ, θ, Ot+1 ) = 0, deg (ϕ, Oi , q) = 0,

for i ≥ t + 2 since Oj ⊂ Br (0) and ϕ−1 (y) ∩ Oj = ∅ for j ≥ t + 2. Therefore, the right
side of (2.2.1) has only finitely many terms which are not zero.
We start by assuming that ψ belongs to the class C 1 (θ̄), ϕ belongs to the class C 1 (Rn ),
and q is a regular value of the function ϕψ.
X X
deg(ϕψ, θ, q) = sgnϕψ (r) = sgnϕ (ψ(r)) sgnψ (r),
r∈(ϕψ)−1 (q) r∈(ϕψ)−1 (q)

and note

4
 X
sgnϕ (z) sgnψ (r)
r∈ψ −1 (z),z∈ϕ−1 (q)
 
X X
= sgnϕ (z)  sgn Jψ (r)
z∈ϕ−1 (q),z∈ψ(θ) r∈ψ −1 (z)
X
= sgnϕ (z) deg(ψ, θ, z)
z∈ψ(θ),ϕ(z)=q
t X
X
= sgn Jϕ (z) deg(ψ, θ, z)
i=1 z∈Oi
X
= deg (ψ, θ, Oi ) deg (ϕ, Oi , q) .
i

In the more general situation ψ ∈ C(θ̄) and ϕ ∈ C (Rn ), Set

Um = {z ∈ Br+1 (0)\ψ(∂θ) : deg(ψ, θ, z) = m}

ωm = {i ∈ N : deg (ψ, θ, Oi ) = m} .
Clearly, Um = ∪i∈ωm Oi and therefore we get

" #
X X X X
deg (ψ, θ, Oi ) deg (ϕ, Oi , q) = deg (ϕ, Oi , q) = deg (ϕ, Um , q) .
i m i∈ωm m

Therefore, we need to show

X
deg(ϕψ, θ, q) = deg (ϕ, Um , q) . (2.2.2)
m

Since ∂Um ⊂ ψ(∂θ), we take ϕ0 ∈ C 1 (Rn ) such that

deg (ϕ0 ψ, θ, q) = deg(ϕψ, θ, q), deg (ϕ0 , Um , q) = deg (ϕ, Um , q) .

We suppose that Γ0 = Br+1 (0) ∩ ϕ−1
0 (q) is non-empty and then we have

κ (Γ0 , ψ(∂θ)) = inf {|r − z| : r ∈ Γ0 , z ∈ ψ(∂θ)} > 0.

Now we take ψ0 ∈ C 1 (θ̄) such that

max |ψ(r) − ψ0 (r)| < κ (Γ0 , ψ(∂θ)) , ψ0 (θ̄) ⊂ Br+1 (0)

r∈θ̄

and put

′
Um = {z ∈ Br+1 (0)\ψ0 (∂θ) : deg (ψ0 , θ, z) = m}

5
 ′ ∩ Γ and
Then we have Um ∩ Γ0 = Um 0

′ ′



deg (ϕ0 , Um , q) = deg ϕ0 , Um ∩ Um , q = deg ϕ0 , Um ,q .
Therefore, we have
X
X
′
deg (ϕ0 ψ0 , θ, q) = m deg ϕ0 , Um ,q = m deg (ϕ, Um , q) .
m m

Using a simple homotopy argument, we can conclude that deg (ϕ0 ψ0 , θ, q) = deg (ϕ0 ψ, θ, q).
This confirms the validity of Theorem 2.2.2.

By using the formula we have just prooven, we can proove the following form of Jordan’s
separation theorem:

2.3 Jordan separation theorem

Theorem 2.3.1 (Jordan Separation Theorem). If M1 and M2 are two compact sets in Rn
that are topologically equivalent to each other, then the complements Rn \M1 and Rn \M2
have the same number of connected components.

Proof. Let τ : M1 → M2 is a homeomorphism onto M2 , where τ ′ is a continuous extension

of τ to Rn . By the Tietze extension theorem, we can ensure that τ −1 serves as a continuous
extension of τ −1 : M2 → M1 to Rn . Assuming that Un are the components of Rn \M1 and
these components are bounded.
Vj are the bounded components of Rn \M2 . It is important to note that ∂Ui ⊂ M1 and
Vj ⊂ M2 . For any fixed n, let Wk denote the components of Rn \τ (∂Ui ).
[ [
. Uj = Rn \M2 ⊂ Rn \τ (∂Vn ) = Wk ,
j k

For every j, there is a k such that Uj ⊂ Wk ; hence, specifically, U∞ ⊂ W∞ . For any q ∈ Vn ,

it is noted that τ −1 τ ′ x = x for x ∈ ∂Vn ; therefore, by Theorem 2.2.2, we obtain:
  X  
1 = deg τ −1 τ ′ , Vn , q = deg τ ′ , Vn , Wk deg τ −1 , Wk , q .

Put Nk = {j : Uj ⊂ Wk }. Then we get

  X  
deg τ −1 , Wk , q = deg τ −1 , Uj , q ,
j∈Nk

deg τ , Vn , Wk = deg τ ′ , Vn , Uj ,
′

for all j ∈ Nk . Therefore, we obtained

6
 XX  
deg τ ′ , Vn , Uj deg τ −1 , Uj , q


1=
k j∈Nk
X   (2.3.1)
deg τ ′ , Vn , Uj deg τ −1 , Uj , Vn ,


=
j

Rn \ τ −1 (M) ⊂ Rn \τ −1 (∂Uj ). For any fixed j, the same reasoning

since q ∈ Vn ⊂
implies that
X  
deg τ ′ , Vn , Uj deg τ ′ , Vn , Uj deg τ −1 , Uj , Vn .



1= (2.3.2)
n

By combining equations (2.3.1) and (2.3.2), we can conclude that Rn \M1 and Rn \M2
possess an equal number of connected components.

2.4 Open mapping theorem

Theorem 2.4.1 (Open Mapping Theorem). Suppose χ ⊂ Rn is an open subset and G : χ →
Rn is a continuous function that is locally injective. Then, G is a mapping that preserves
openness.
Proof. We aim to demonstrate that for every x0 ∈ χ, there exists a radius r > 0 such that
the image of the ball Br (x0 ) under G encloses a ball which is centered at G (x0 ). WLOG, we
can assume that x0 = 0. If not, we can define χ1 = χ−{x0 } and G1 (x) = G (x + x0 )−G (x0 ).
 G is injective on the ball Br (0). Define the
Choose a positive value r > 0 such that
1 t
function ℵ(t, x) = G 1+t x − G − 1+t x for all (t, x) ∈ [0, 1] × Br (0). The function ℵ is
continuous, and it never equals zero for any (t, x) ∈ [0, 1] × ∂Br (0). If there exists a point
1 t
(t, x) in [0, 1] × ∂Br (0) where ℵ(t, x) = 0, then it would imply 1+t x = − 1+t x due to the
injectivity of G and x = 0, which leads to a contradiction. Thus, we obtain:

deg (G, Br (0), 0) = deg (ℵ(1, ·), Br (0), 0) ̸= 0,

As ℵ(1, ·) is an odd function, select a positive value t such that t < inf {|G(x)| : x ∈ ∂Br (0)}.
Then,

deg (G, Br (0), y) = deg (G, Br (0), 0) .

At this point, we have Bt ⊂ G (Br (0)), demonstrating that G is an open mapping.

2.5 Argument principal as a special case of degree of holomorphic map

Theorem 2.5.1. Consider B(0, 1) ⊂ C as the ball with unit radius, where Ω = ∂B(0, 1),
and let φ : B(0, 1) → C be a C 1 function. Suppose p ∈
/ φ(Ω). Then,
Z
1 1
deg(φ, B(0, 1), p) = dw. (2.5.1)
2πi φ(Ω) w − p

7
Proof. To establish (2.5.1 ), it is adequate to consider the scenario where p ∈
/ φ (Qφ ).
Suppose φ−1 (p) = {z1 , z2 , · · · , zk }. Our task is to demonstrate:
Z k
1 1 X
dw = sgn Jφ (zi ) (2.5.2)
2πi φ(Ω) w−p
i=1

Select a small δ > 0 such that the closures of Vi are disjoint, where Vi = B (zi , δ). For
w ∈ Vi , define sgn Jφ (w) = sgnφ (zi ), with Vi ⊂ B(0, 1) and φ restricted to Vi being a
homeomorphism for i = 1, 2, . . . , k. Let Qi = ∂Vi . Thus, φ (Qi ) forms a Jordan curve with
a inside its interior. The orientation of φ (Qi ) matches that of Qi if Jφ (zi ) > 0, and is
opposite if Jφ (zi ) < 0.
Now, let U = B(0, 1)\ ∪ki=1 Vi . Then |φ(w) − p| > ρ in U for some ρ > 0. Divide U into small
rectangles Mj such that |φ(w) − φ(s)| < ρ on each Mj . Because the image φ (∂ (Mj ∩ V))
doesn’t wind around p, we have s (φ (∂ (Mj ∩ V)) , p) = 0, and summing over all Mj yields
Z k Z
1 X 1
dw = dw.
φ(Ω) w−p φ(Qi ) w−p
i=1

Given that the orientation of φ (Qi ) is dictated by Jφ (zi ), the curve φ (Qi ) completes
precisely one revolution around a. Therefore, we obtain:
Z
1
dw = sgn Jφ (zi ) .
φ(Qi ) w − p

Hence the theorem 2.5.1 is prooved.

3 Degree Theory for Condensing Operators

3.1 Leray-Schauder degree
Lemma 3.1.1. Think of X as a real normed space that’s complete, Λ as a limited open
part of it, and Z as a continuous map from the closure of Λ to E. Then, for any positive
number ρ, there exists a finite space H and a continuous map Zρ from the closure of Λ to
H.

∥Zρ ω − Zω∥ < ρ for all ω ∈ Λ̄.

Proof. Since Z Λ̄ is compact within E, for any positive value of ρ, there is a limited set
{w1 , w2 , . . . , wn } from the closure of Λ.

Z(Λ̄) ⊂ ∪ni=1 B (Zwi , ρ)

Now, we define a mapping Zρ : Λ̄ → H = span {Zw1 , Zw2 , · · · , Zwn } as follows:

8
 n
X ψi (ω)
Zρ ω = Zwi for all ω ∈ Λ̄
Γ(ω)
i=1

Here, ψi (ω) = max{0, ρ − ∥Zω − Zwi ∥} and Γ(ω) = ni=1 ψi (ω). It can be verified that
P
Zρ meets the requirements of Lemma 3.1.1, thus concluding the theorem.

Definition 3.1.1 (Leray-Schauder Degree). Consider a continuous map ω in the form of

ω = id − γ, where γ is compact on E and γ ∈ / ω(∂E). Let γ ρ be a finite-rank perturbation
ρ
satisfying ∥γ − γ ∥C 0 < ρ and ρ < δ/2 (δ = inf x∈∂E ∥γ − ω(x)∥ = inf y∈ω(∂E) ∥γ − y∥ > 0),
b
and with γ ρ (Ē) ⊂ Y ρ ⊂ X. Then, for any finite-dimensional subspace X ρ containing both
Y ρ and γ, the Leray-Schauder degree is defined as

deg(ω, E, γ) := deg (ω ρ , E ∩ X ρ , γ) ,

where ω ρ = id − ω ρ .

Theorem 3.1.2. The Leray-Schauder degree exhibits the following properties:

1. If q ∈ E, then deg(id, E, q) = 1.
2. For E 1 , E 2 ⊂ E, disjoint open subsets of E, and q ∈


/ ζ Ē\(Ω1 ∪ Ω2 ) , it follows that
deg(ζ, E, q) = deg(ζ, E 1 , q) + deg(ζ, E 2 , q).
3. For any continuous paths t 7→ ωt = id − Lt , where Lt ∈ K(Ē) and q ∈ / ωt (∂E), the
degree deg (ωt , E, pt ) is constant for all t ∈ [0, 1].
4. deg(ζ, E, q) = deg(ζ − q, E, 0).
The function deg is known as a degree theory.

Lemma 3.1.3. Consider a real complete normed linear space X, where E ⊂ X is a bounded
and closed subset. Let L : E → X be a compact and continuous map such that L(z) ̸= z
for every z ∈ E. Then, there exists δ0 > 0 such that for all t ∈ [0, 1] and z ∈ E, we have
z ̸= tLδ1 (z) + (1 − t)Lδ2 (z), where δi ∈ (0, δ0 ) and Lδi : E → Fδi for i = 1, 2 as in Lemma
3.1.1.

Proof. Suppose the this is false. Then there exist some δ1j → 0, δ2j → 0, tj → t0 , wj ∈ E
such that tj Lδj wj + (1 − tj ) Lδj wj = wj for j = 1, 2, · · · .
1 2
As L is a compact map, the sequence (Lwj )∞ j=1 has a converging subsequence, denoted
as (Lwjk ), which converges to y ∈ X. According to Lemma 3.1.1, Lδi wjk → y for i =
1, 2. Therefore, we have wjk → y ∈ E. Consequently, we arrive at Ly = y, leading to a
contradiction.

3.2 Uniqueness of leray-schauder degree

Consider a real normed space which is also complete space X, where Γ ⊂ X is an open
bounded set, and L : Γ̄ → X is a continuous compact mapping. Assuming that 0 ∈ /
(I − L)(∂Γ), according to Lemma 3.1.3, there exists ρ0 > 0 such that

9
 w ̸= tLρ1 w + (1 − t)Lρ2 w for all t ∈ [0, 1], w ∈ ∂Γ,
Here, with ρi ∈ (0, ρ0 ) and Lρi : Γ̄ → Rρi for i = 1, 2 as in Lemma 3.1.1, Brouwer’s
degree deg (I − Lρ , Γ ∩ Rρ , 0) is well-defined. This allows us to define

deg(I − L, Γ, 0) = deg (I − Lρ , Γ ∩ Rρ , 0)
where ρ ∈ (0, ρ0 ).
Brouwer degree doesn’t change under homotopy. Thus we have

deg (I − Lρ1 , Γ ∩ span {Rρ1 ∪ Rρ2 } , 0) = deg (I − Lρ2 , Γ ∩ span {Rρ1 ∪ Rρ2 } , 0)

But Lρi : Γ̄ ∩ span {Rρ1 ∪ Rρ2 } :→ Ri for i = 1, 2, so by Intermediate Value Theorem

we have

deg (I − Lρ1 , Γ ∩ span {Rρ1 ∪ Rρ2 } , 0) = deg (I − Lρ1 , Γ ∩ Rρ1 , 0)

and

deg (I − Lρ2 , Γ ∩ span {Rρ1 ∪ Rρ2 } , 0) = deg (I − Lρ2 , Γ ∩ Rρ2 , 0) .

Thus we have

deg (I − Lρ1 , Γ ∩ Rρ1 , 0) = deg (I − Lρ2 , Γ ∩ Rρ2 , 0)

and the degree for the map that we defined is well defined. For the more general
situation, if p ∈
/ (I − L)(∂Γ), we define deg(I − L, Γ, p) = deg(I − L − p, Γ, 0). Excision
property of Leray-Schauder degree is followed by excision property that we defined in 2.0.2

Theorem 3.2.1. Suppose you have a complete normed space X, where X0 is a closed
subspace of X, and Λ is a bounded open subset of X. If L : Λ̄ → X0 is a continuous and
compact map, and p ∈ X0 , then the degree of (I − L) over Λ with respect to p is equal to
the degree of (I − L) over Λ ∩ X0 with respect to p.

Proof. Since L(Λ̄) ⊂ X0 , we choose a finite dimensional space F ⊂ X0 with p ∈ F and

L1 : Λ̄ → F such that ∥Lx − L1 x∥ < δ in Definition 3.2 for small δ > 0. Then we have by
excision property Leray-Schauder degree

deg(I − L, Λ, p) = deg (I − L1 , Λ ∩ F, p) = deg (I − L, Λ ∩ X0 , p) .

10
3.3 Condensing and countably condensing map
Definition 3.3.1. In a real normed space X, let S : E → X is a mapping, where K
represents the measure of noncompactness.
1. If, for all subsets B ⊂ E, which are bounded and the inequality K(SB) ≤ kK(B)
holds, then S is called a k-set contraction, for some k > 0.
2. S is considered to be condensing if, for all subsets B ⊂ E, which are bounded and
K(B) > 0, it satisfies K(SB) < K(B).

3.4 Housedorff metric

In a metric space (E, d), where F ⊂ E is a subset, the diameter of F , denoted by diam(F ),
is defined as the supremum of the distances between points in F , i.e., supp,q∈F d(p, q). If
diam(F ) < +∞, we say that F is bounded. The Hausdorff metric H is defined for two
bounded sets F and C as follows:
( )
H(F, C) = max sup d(p, C), sup d(q, F )
p∈F q∈C

Proposition 3.4.1. Let (Y, d) be a metric space. Then (NY , H) is a metric space. Where
NY is the collection of all bounded subsets of Y .

Proof. Obviously, H(M, N ) ≥ 0 for any M, N ∈ NY , H(M, N ) = 0 if only if M = N and

H(M, N ) = H(N, M ).
For M, N, O ∈ NY , we have

d(p, N ) ≤ d(p, r) + d(r, N ), d(q, M ) ≤ d(q, r) + d(r, M )

for all r ∈ O, p ∈ M and q ∈ N . Thus

d(p, N ) ≤ inf d(p, r) + sup d(r, N )

r∈O r∈O
d(q, M ) ≤ inf d(q, r) + sup d(r, M )
r∈O r∈O

and so

H(M, N )
( )
≤ max sup d(p, O) + sup d(r, N ), sup d(q, O) + sup d(r, M )
p∈M r∈O q∈N r∈O
( ) ( )
≤ max sup d(p, O), sup d(r, M ) + max sup d(r, N ), sup d(q, O)
p∈M r∈O r∈O q∈N

=H(M, O) + H(O, N ).
Thus (NY , H) is a metric space.

11
Proposition 3.4.2. |K(M ) − K(N )| ≤ 2H(M, N ).

Definition 3.4.1. Consider E as a real normed space which is also complete space, where
L : Γ → E is a mapping, and K represents the measure of noncompactness.
1. L is termed a countably k-set contraction if, for all countably bounded subsets φ ⊂ Γ,
it satisfies K(Lφ) ≤ kK(φ), for some k > 0.
2. L is considered a countably k-set contraction if, for all countably bounded subsets
φ ⊂ Γ, it fulfills K(Lφ) ≤ kK(φ), where k > 0 is a constant.
3. H(t, x) : [0, 1] × Γ → E is termed a homotopy where for each t ∈ [0, 1] we have a
countably condensing map if, for all bounded and countable subsets φ ⊂ Γ with K(φ) > 0,
it satisfies K(H([0, 1] × φ)) < K(φ).

Theorem 3.4.1. Consider X as a Normed space which is complete and A ⊂ C([a, b], X)
as a bounded equicontinuous subset. Then K(A(x)) is continuous on [a, b], where A(x) =
{f (x) : f (·) ∈ A}, and
Z b  Z b
K f (x)dx : f (·) ∈ A ≤ K(A(x))dx.
a a

Proof. First, we will see why K({f (x) : f (·) ∈ A}) is a function with continuity property on
the this interval [a, b]. For any ϵ > 0, since A satisfies the conditions to be equicontinuous,
we can find δ > 0 such that ∥f (x) − f (x′ )∥ < ϵ for all x, x′ ∈ [a, b] satisfying |x − x′ | < δ.
Thus we have

S {f (x) : f (·) ∈ A}, f x′ : f (·) ∈ A ≤ ϵ



for x, x′ ∈ [a, b] satisfying |x − x′ | < δ. From this and (6) of Proposition 3.4.2, we infer that

f x′ : f (·) ∈ A



K({f (x) : f (·) ∈ A}) − K ≤ 2ϵ
for x, x′ ∈ [a, b] satisfying |x − x′ | < δ. Thus K({f (x) : f (·) ∈ A}) is a continuous
function on [a, b].
Given any partition of [a, b] : a = x0 < x1 < · · · < xn = b, where xi = a + i b−a n for
i = 0, 1, · · · , n. For any ϵ > 0, due to the property of equicontinuity of A, there will be
N > 0 such that if n > N , then...

∥f (xi ) − f (x)∥ < ϵ for all f (·) ∈ A, x ∈ [xi−1 , xi ]

for i = 1, 2, · · · , n. So we have

b xi
b−a
Z Z
Σni=1 f (xi ) − f (x)dx = Σni=1 (f (xi ) − f (x)) dx < ϵ(b − a)
n a xi−1

for all n > N . Therefore, we have

12
 ( n )! Z b 
X b−a
K f (xi ) : f (·) ∈ A −K f (x)dx : f (·) ∈ A ≤ 2ϵ(b − a),
n a
i=1

i.e.,
( n )! Z b 
X b−a
lim K f (xi ) : f (·) ∈ A =K f (x)dx : f (·) ∈ A .
n→∞ n a
i=1
 
n b−a b−a
K Σi=1 f (xi ) : f (·) ∈ A ≤ Σni=1 K ({f (xi ) : f (·) ∈ A}) .
n n
Therefore, we get the following
Z b  Z b
K f (x)dx : f (·) ∈ A ≤ K(A(x))dx.
a a
This prooves the theorem.

Theorem 3.4.2. Let X be a normed space which is also complete space and P ⊂ C([c, d], X)
be a bounded subset having equicontintinuity property, where c, d ∈ R. Then

K(P) = max K({f (t) : f (·) ∈ P})

t∈[c,d]

Proposition 3.4.3. Let X be a normed space which is also complete space, Λ ⊂ X be a

subset which is bounded and S : Λ̄ → X satisfying the property of countably condensing
map. Put W = {x ∈ Λ̄ : Sx = x}. Then we can find a convex subset K having compactness
property such that

1. W ⊆ K;

2. if x0 ∈ conv (K ∪ {Sx0 }), then x0 ∈ K;

3. K = conv(S(K ∩ Λ̄)).

Proposition 3.4.4. Consider X as a normed space which is also complete space, where
Λ ⊂ X is a bounded subset and L : Λ̄ → X is a countably condensing map. Define
K1 = conv(LΛ̄), Kn+1 = conv L Kn ∩ Λ̄ for n ≥ 1, and K = ∩∞



n=1 Kn . As a result, K
is both convex and compact.

4 Degree Theory for Countably Condensing Mapping

Here we will define degree theory for the maps which are condensing mappings.

13
4.1 Defining degree for a special class of mappings
Let X be a normed space which is also complete space and Θ ⊂ X be an open subset having
boundedness. Suppose J : Θ̄ → X is a continuous mapping having property of countably
condensing map, with 0 ∈
/ (I − J )(∂Θ). If 0 ∈
/ (I − J )(Θ), we define deg(I − J , Θ, 0) = 0.
Otherwise, let Ω = {x ∈ Θ̄ : Lx = x} and consider K as the subset having convexity
and compactness property described in Proposition 3.4.3. The set K is nonempty as Ω is
contained in K. The mapping J : K ∩ Θ̄ → K is well-defined, following property 3 of
Proposition 3.4.3.

If W : X → K is a retraction, then J W is a compact mapping because it maps

bounded subsets of X to bounded subsets of K, which are relatively compact. The set
W −1 (Θ) is open in X. Given

the assumption that 0 ∈ / (I − J )(∂Θ) and the fact that

0∈ / (I −J W) ∂ W −1 (Θ) ∩ Θ , the Leray-Schauder degree deg I − J W, W −1 (Θ) ∩ Θ, 0
is defined in a nice way for each retraction W. Now, we proceed to define

deg(I − J , Θ, 0) = deg I − J W, W −1 (Θ) ∩ Θ, 0 ,

(4.1.1)

where the deg I − J W, W −1 (Θ) ∩ Θ, 0 is so called Leray Schauder degree.

To see this definition is doesn’t depend upon the choosen retraction, we have to proove,
if W1 , W2 : X → K are two retractions, then

deg I − J W 1 , W1−1 (Θ) ∩ Θ, 0 = deg I − J W 2 , W2−1 (Θ) ∩ Θ, 0 .

Define W(λ, x) = λW 1 (x) + (1 − λ)κ2 (x) for all (λ, x) ∈ [0, 1] × X. Then, W(λ, ·) :
X → K acts as a retraction for each
λ ∈ [0, 1]. It is evident that x ̸= J W(λ, x) for
(λ, x) ∈ [0, 1]∂ r1−1 (Θ) ∩ κ−1
2 (Θ) ∩ Θ . Therefore, utilizing the homotopy property, we
obtain

deg I − J W 1 , W1−1 (Θ) ∩ W2−1 (Θ) ∩ Θ, 0

= deg I − J W 2 , W1−1 (Θ) ∩ W2−1 (Θ) ∩ Θ, 0 .

It is quite easy to see that

/ W1−1 (Θ) ∩ Θ\W1−1 (Θ) ∩ W2−1 (Θ) ∩ Θ

0∈
and

/ W2−1 (Θ) ∩ Θ\W1−1 (Θ) ∩ W2−1 (Θ) ∩ Θ

0∈
Thus by excision theorem we can implies that

deg I − J W 1 , W1−1 (Θ) ∩ Θ, 0 = deg I − J W 1 , W1−1 (Θ) ∩ W2−1 (Θ) ∩ Θ, 0

and

14
 deg I − J W2 , W2−1 (Θ) ∩ Θ, 0 = deg I − J W2 , W1−1 (Θ) ∩ W2−1 (Θ) ∩ Θ, 0 .

Therefore, we have

deg I − J W 1 , W1−1 (Θ) ∩ Θ, 0 = deg I − J W 2 , W1−1 (Θ) ∩ W2−1 (Θ) ∩ Θ, 0 .

The degree for the compact set defined in 3.4.4 can be established in a similar manner,
aligning with the previously defined degree.

4.2 Properties of the degree theory defined for countably condensing map

Theorem 4.2.1. The degree for the maps that are defined by 4.1.1 has the property
of homotopy equivalence:
Consider S(λ, ω) : [0, 1] × Ē → E as continuous map which is also countably condens-
ing map, meaning K([0, 1]×A) < K(A) for all countable subsets A of Ē with K(A) > 0,
and S(λ, ω) ̸= ω for all (λ, ω) ∈ [0, 1] × ∂E. In this case, deg(I − S(λ, ·), E, 0) remains
constant regardless of the value of λ ∈ [0, 1].

Proof. Let

U0 = conv(S([0, 1] × J¯)), Un = conv S [0, 1] × Un−1 ∩ J¯

For n ≥ 1, consider U = ∩∞ n=0 Un , which is compact according to Proposition 3.4.4,

and S([0, 1] × U) → U. Let κ
: E → U be a retraction. It holds that w ̸= S(λ, κw)
for all w ∈ ∂ κ−1 (J ∩ U) ∩ J . Consequently, deg I − S(λ, ·)κ, κ−1 (U ∩ J ) ∩ J , 0 is

independent of λ. Thus, the homotopy proprty can be drawn from the d excision property
of the Leray-Schauder degree.

4.3 Some fixed point theorems involving countably condensing map

Theorem 4.3.1. Consider X as a normed space which is also complete space, where Λ ⊂ X
is an bounded subset which is also open and containing 0, and P : Λ̄ → X is a map with
continuity property this mapping is also countably condensing map. If z ̸= λP z for each
λ ∈ [0, 1) and z ∈ ∂Λ, then P possesses a point within Λ̄ such that P z = z.

Proof. Let’s assume that P z ̸= z for all z ∈ ∂Λ. Define S(λ, z) = tP z for all (λ, z) ∈
[0, 1]× Λ̄. It is evident that {S(λ, ·)}λ∈[0,1] forms a homotopy wich are countably condensing
maps for each λ. Given this, we have S(λ, z) ̸= z for each z ∈ ∂Λ. Consequently, deg(I −
P, Λ, 0) = deg(I, Λ, 0) = 1. Hence, the equation P z = z has a solution in Λ, which prooves
the theorem.

15
Corollary 4.3.1. Consider X as a normed space which is also complete space, where B ⊂ X
is an open subset having boundedness property containing 0, and P : B̄ → X is a continuous
map which is also countably condensing map. If ∥P ω∥ ≤ ∥ω∥ for all ω ∈ ∂B, then P
possesses a within B̄, such that P z = z.

Proof. Let us suppose that P ω ̸= ω for each ω ∈ ∂B. Or else, the result is true.
we have λP ω ̸= ω for each ω ∈ ∂B and λ ∈ [0, 1) because

|ω − λP (ω)| ≥ ||ω| − λ|P (ω)|| > λ||ω| − |P (ω)|| > 0.

And λ is less than equal 1. By Theorem 4.3.1, P has a point in B̄ satisfying P z = z.

Theorem 4.3.2. Consider a nonempty subset D ⊂ E such that it is bounded,closed and

convex, and let P : D → D is a map such that it is continuous and countably condensing
map. Then, P possesses a fixed point in D.

Proof. Let’s assume that P is a map which is λ-set contraction for some λ ∈ [0, 1). Define
P1 = conv P (D) and Pi+1 = conv P Pi for i = 1, 2, . . .. According to Proposition 3.4.4, the
intersection J = ∩∞
i=1 Pi is satisfies convexity and compactness properties, and the mapping
P : J → J is established.
of J . Choose i
Now, we demonstrate the non-emptiness
i

n i

z0 ∈ D, then P z0 ∈ Pi for i ≥ 1.
It follows that J {P z0 , i ≥ n} ≤ λ J {P z0 , i ≥ 0} for n ≥ 1.

P i z0 , i ≥ n P i z0 , i ≥ 0



J =J ,

∞
Therefore, J {P i z0 , i ≥ 0} = 0. Consequently, the sequence P i z0 i=1 has a subsequence

that converges to a point in J , indicating that J is not empty. Hence, P possesses a fixed
point.
Now, assuming P is countably condensing, select z0 ∈ D, and for any λ ∈ (0, 1), define
Sz = λP z + (1 − λ)z0 for all z ∈ D. Since S is λ-set contraction, it has a point in D, such
that Sz = z. Let λn → 1. Then there exist zn ∈ D such that

λn P zn + (1 − λn ) z0 = zn ,
This means that the closure of the set of images of P zn as n ≥ 1 is equal to the closure
of the set of zn as n ≥ 1. Given that P is countably condensing, the sequence (zn )∞ n=1
contains a subsequence znj converging to a point y ∈ D. Due to the continuity of P , we
have P y = y, so we got a fixed point.

Theorem 4.3.3. Consider X as a normed space which is also complete space with P :
X → X as a countably condensing map also satisfies the continuity. This ensures one of
the following situation:

1. P has a point in X such that P z = z;

2. {z : P z = κz for some κ > 1} is an unbounded set.

16
Proof. Consider that {z : P z = κz for some κ > 1} is a bounded subset of X. Take d > 0
such that {z : P z = κz for some κ > 1} ⊂ B(0, d). If P z = z for some z ∈ ∂B(0, d),
then we have 1st situation holds. So we may suppose that P z ̸= z for all z ∈ ∂B(0, d).
Thus by the same reasoning as in 4.3.1 z ̸= tP z for all t ∈ [0, 1] and z ∈ ∂B(0, d), so
deg(I − P, B(0, d), 0) = 1 and P has a point in B(0, d) such that P z = z.

5 Applications to ODEs
5.1 Solvability result for a type of differential equations in Banach spaces
We see applications of degree for set contractive mappings to ordinary differential equations
in normed space which is also complete spaces.

Theorem 5.1.1. Consider X as a normed space which is also complete space, where
g(x, z) : [0, 1] × B(z0 , r0 ) → X is continuous map that satisfies

K (g ([0, 1] × B (z0 , c)) ≤ κK (B (z0 , c)) for all c ∈ (0, c0 ) ,

where κ ∈ (0, 1) and c0 > 0 are constants. Then there exists x0 ∈ (0, 1] such that the
initial value problem (
z ′ (x) = g(x, z(x)), x ∈ (0, x0 ) ,
(5.1.1)
z(0) = z0 ,
has a solution.

Proof. n o n c o
0
J = sup ∥g(x, z)∥ : (x, z) ∈ [0, 1] × B (z0 , c0 ) , x0 = min 1, .
J
Clearly, the first theorem of ordinary differential equations (ODE) in complete normed space
is equivalent to the following integral equation:
Z x
z(x) = z0 + g(s, z(s))ds, (5.1.2)
0
Put E = C ([0, x0 ] , X) having norm ∥z(·)∥ = max {∥z(x)∥ : x ∈ [0, x0 ]}. Then E is a
normed space which is also complete space. We also define K = {z(·) ∈ X : z (x0 ) = z0 , ∥z(x) − z (x0 )∥ ≤
c0 }. Then K is a subset which is bounded and closed and satisfies convexity.
Now, LLet a mapping S : K → K is given by
Z x
Sz(x) = z0 + g(s, z(s))ds for all z(·) ∈ K. (5.1.3)
0
It is simple to verify that S satisfies continuity. Now, we have to show that S is con-
densing. Which means, for any subset which satisfies, B of K with K(B) > 0, we see by
3.4.2 and 3.4.1,

17
 K(SB) = sup K({Sz(x) : z(·) ∈ B})
x∈[0,x0 ]
 Z x 
= sup K z0 + g(s, z(s)) : z(·) ∈ B
x∈[0,x0 ] 0
)!
n o
≤ sup K z0 + x conv (g(s, z(s)) : s ∈ [0, x0 ]) : z(·) ∈ B
x∈[0,x0 ]
!
n o
≤ sup K z0 + x conv (g ([0, x0 ] × B)
x∈[0,x0 ]

≤ x0 sup K (g ([0, x0 ] × B))

x∈[0,x0 ]

≤ x0 κK(B).
Thus S is prooven to be a condensing mapping, so S has a point in K such that Sz = z,
i.e., 5.1.2 has a solution. Which implies, 5.1.1 has a solution. This completes the proof.

Lemma 5.1.2. If U, U ′ ∈ L2 (0, K; Y ), where Y is a Hilbert space and U(x + K) = −U(x)

for all x ∈ R, then
√ Z K  12
K ′ 2
∥U∥∞ ≤ U (s) ds
2 0

Rx RK
Proof. Since U(x) = U(0) + 0 U ′ (s)ds and U(x) = U(K) − x U ′ (s)ds, and x ∈ (0, K) we
have
Z x Z K 
1 ′ ′
U(x) = U (s)ds − U (s)ds ,
2 0 x
Z x Z K
1 ′
|U(x)| = U (s)ds − U ′ (s)ds ,
2 0 x
Z x Z K
1 ′
|U(x)| ≤ U (s)ds + U ′ (s)ds ,
2 0 x
Z x
1
|U(x)| ≤ U ′ (s)ds .
2 0

Since it is true for every x ∈ (0, K). Thus the conclusion follows by Holder’s inequality
√ Z K  21
K ′ 2
∥U∥∞ ≤ U (s) ds .
2 0

18
5.2 Solvability result for a type of differential equation in Hilbert spaces
Theorem 5.2.1. Let Y be a normed space additionally it is complete inner product spcace,
and C > 0, Y(x, ζ) : R × Y → Y be a continuous mapping satisfying

K(Y([0, C] × B) ≤ λK(B),
for every subsets B of Y such that B is bounded, λ ∈ (0, 1) , and we have λC < 1.
If Y(x + C, −ζ) = −Y(x, ζ) and ∥Y(x, ζ)∥ ≤ N ∥ζ∥ + J (x) for all (x, ζ) ∈ R × E, where
0 ≤ N C < 2, and J (·) ∈ C 2 (0, C), then the following problem
(
ζ ′ (x) = Y(x, ζ(x)), x ∈ R
(5.2.1)
ζ(x + C) = −ζ(x),

has a solution.

Proof. Consider La = {ζ(·) : R → Y is continuous, ζ(x + C) = −ζ(x), x ∈ R}. Define

∥ζ(·)∥a = maxx∈[0,C] for ζ(·) ∈ La , and it can be verified that La forms a complete normed
space under this norm. It is straightforward to show that 5.2.1 can be written as the
following equation:
Z C Z x
1 1
ζ(x) = − Y(s, ζ(s))dx + Y(s, ζ(s))ds
2 x 2 0
We establish a mapping S : La → La as follows:
Z C Z x
1 1
Sζ(x) = − Y(s, ζ(s))dx + Y(s, ζ(s))ds for all ζ(·) ∈ La .
2 x 2 0
For any bounded subset U of La , we have, by Theorem 3.4.2, that

K(SU ) = max {K({Sζ(x) : ζ(·) ∈ U })}

x∈[0,C]

By the same reasoning as in Theorem 5.1.1, we get

K(SU ) ≤ λCK(U )
Thus S is a λC-set contraction.
Now, we show that Sζ(·) ̸= λζ(·) for all. λ > 1 and ζ(·) ∈ La with
 −1 √ Z C  12
NC C 2
∥ζ(·)∥a > 1− J (x)dx .
2 2 0
Suppose this is not true, then there exists ζ(·) ∈ La with
 √ Z C  12
N C −1 C

2
∥ζ(·)∥ > 1 − J (x)dx ,
2 2 0

19
 such that Sζ(·) = κζ(·) for some κ > 1. Then we have

κζ ′ (x) = Y(x, ζ(x)). (5.2.2)

Multiplying 5.2.2 by ζ ′ (x) and integrate over [0, C], we have
Z C Z C
′ 2
κ Y(x, ζ(x))ζ ′ (x)dx
ζ (x) dx =
0 0
Z C Z C (5.2.3)
≤N ∥ζ(x)∥ ζ ′ (x) dx + J (x) ζ ′ (x) dx.
0 0
In view of Lemma 5.1.2, we have
√ Z C  12
C ′ 2
|ζ|∞ ≤ ζ (s) ds . (5.2.4)
2 0

Putting 5.2.4 in 5.2.3,

C C Z C  21 Z C  12
NC
Z Z
′ 2 ′ 2 2 ′ 2
κ ζ (x) dx ≤ ζ (x) dx + J (x)dx ζ (x) dx ,
0 2 0 0 0

i.e.,
 12  12
C
N C −1
Z   Z C
′ 2 2
ζ (x) dx ≤ 1− J (x)dx .
0 2 0
Again, by Lemma 5.1.2, we get
 −1 √ Z C  12
NC C 2
∥ζ(·)∥a ≤ 1− J (x)dx ,
2 2 0
This leads to a contradiction. Therefore, according to Theorem 4.3.3, S possesses a fixed
point in La , implying that the problem 5.2.1 has a solution. This concludes the proof.

20
References
[1] Louis Nirenberg. Topics in nonlinear functional analysis, volume 6. American Mathe-
matical Soc., 1974.

[2] Yeol Je Cho and Yu-Qing Chen. Topological degree theory and applications. Chapman
and Hall/CRC, 2006.

21