1. ADEKUNLE AJASIN UNIVERSITY AKUNGBA AKOKO
A PROJECT WORK ON
MAP PROJECTIONS AND THEIR
PROPERTIES
BY
FALADE JOHN TEMIDAYO
MATRIC NO: 110416008
SUBMITTED TO THE DEPARTMENT OF MATHEMATICS AND INDUSTRIAL
MATHEMATICS, FACULTY OF SCIENCE, ADEKUNLE AJASIN UNIVERSITY
AKUNGBA AKOKO, ONDO STATE.
IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE
AWARD OF BACHELOR OF SCIENCE (B.Sc. Hons) DEGREE IN
MATHEMATICS.
NOVEMBER 2015
2. 2
CERTIFICATION
This is to certify that this project work was carried out by FALADE JOHN
TEMIDAYO with matriculation number 110416008 in the Department of
Mathematical Science.
_________________________ _______________
Dr. O.S Olusa DATE
Supervisor
_________________________ _______________
Dr. E.P Akpan DATE
Head of Department
_________________________ _______________
External Examiner DATE
4. 4
ACKNOWLEDGEMENT
My sincere appreciation goes to my project supervisor Dr. O.S Olusa for his
fatherly love, advice, suggestions, his direction, and contribution towards the
completion of this project, may the Lord bless and keep you sir. Special thanks to
the Head of Department of Mathematical Sciences Dr. E.P Akpan and all the
Lecturers in the Department of Mathematical Sciences.
I am very grateful to the contributions and prayers of my parent Mr and Mrs Falade
and that of my guardian Mrs Yusuf Funke. I will not forget the love of my grandpa
Late Chief Adeyemi Falade, may his gentle soul rest in perfect peace. I cannot
forget the encouragement I always receive from my only brother and friend Falade
Olagoke. I also appreciate my only sister Falade Faith and all my siblings for their
word of encouragement. I will never forget the support of my little brother
Aremora Damilare Samuel for been there for me all the time.
I also want to say thank you to Mr. Oni Sunday, Mr. Oluwadare Seyi, Mr. Alafe
Taiwo, Miss. Biola, Sis Tito and all 400 Level 2014/2015 set of Mathematical
Sciences for their love and encouragement.
I will love to show my sincere appreciation to all the Pastors in Deeper Life Bible
Church Oka Region and the Regional Coordinator (Pastor Akinniyi Idowu),
Associate Coordinators (Pastor Olusola Ajayi, Pastor Saliu John) and their wives
5. 5
all the Executives of Deeper Life Campus Fellowship (Bro Sunday, Sis Ojumola
Joy, Bro Asade Rotimi, Bro Adewole Jeremiah, Bro Ayodeji Samuel, Sis
Ogedengbe Damilola, Sis Fatoba Kitan, Sis Adebayo Iyanuloluwa, Sis Adebayo
Adesola and others), to all the workers in DLCF AAUA and all the members of the
Fellowship.
Finally, I want to appreciate the effort of the following people: Sis Ayegbo Ruth,
Bro Ajobiewe Jude, Bro Towolawi Olorunwa, Bro Otaraki David, Bro Ayeyemi
Tominiyi, Bro Ogunbodede Deji, Bro Jacob Bola, Sis Olaoye Blessing, Sis
Ologunagba Grace, Bro Bagun Festus, Sis Olanipekun Esther, Bro Oludaisi
Gabriel, Bro Onalo Naphtali, Bro Babayemi Tosin, my Lenovo PC, Sis Adebayo
Sefunmi, Sis Faith Essien, Sis Omoniregun Doyin, Sis Oripelaye Shola, Bro
Olaniyan Kunle and Bro Akande Akinsanmi.
6. 6
ABSTRACT
This project explains the concept of Map Projections, their properties and its
application to various mathematical structures.
The relevant definitions, properties, theorems and proofs are given.
7. 7
TABLE OF CONTENT
Title Page i
Certification ii
Dedication iii
Acknowledgement iv
Abstract vi
Table of Content vii
CHAPTER ONE INTRODUCTION
1.0 Introduction 1
1.1 Literature Review 4
1.2 Definition of Terms 7
CHAPTER TWO PROPERTIES
2.0 Properties of Map Projections 11
2.1 Introduction 11
2.2 Properties of Metric Projection 11
2.3 Homomorphisms 13
2.4 Projection on Banach Spaces 14
8. 8
2.5 Orthogonal Projection 15
2.6 Continuity of the Projection Map 17
2.7 Projection on Topological Group 18
2.8 Generalized and Hyper-Generalized Projection 19
2.9 Projection in Vector Space 23
2.10 Productof Projection 24
2.11 Sum of Projection 25
CHAPTER THREE APPLICATION
3.1 Topological Obstructions and Polytope Projection 27
3.2 Stereographic Projection 30
3.3 Orthogonal Projection 32
3.4 Matrix Projection 33
3.5 Application of Map Projection to the World 37
CHAPTER FOUR CONCLUSION
Conclusion 40
References 41
9. 9
CHAPTER ONE
1.0 INTRODUCTION
In mathematics, a projection is a mapping of set (or any mathematical
structures) into a subset or (substructures), which is equal to its square for mapping
composition (or its Idempotent).
The restriction to a subspace of a projection is also called a projection, even if the
idempotent property is lost.
The concept of projection in mathematics is a very old one, most likely
having its roots in the phenomenon of shadows cast by the real world objects to the
ground. This rudimentary idea was refined and abstracted, first in a geometric
context and later in other branches of mathematics. Over time differing version of
the concept developed, but today, in a sufficiently abstract setting, we can unify
these variations.
An everyday example of projection is the casting of shadows (or mapping)
onto a plane. The projection of a point is its shadow on the plane. The shadow of a
point on the plane is this point itself (idempotence).
The shadow of a three dimensional sphere is a circle. Originally, the notion
of projection mapping was introduced in Euclidean geometry to denote the
projection mapping of Euclidean spaces of three dimensional onto a plane in it.
The two main projection mapping of this kind are:
(a) The projection from a point onto a plane or central projection: If C is the
point, called the centre of projection; the projection of a point P different
from C is the intersection with the plane of the line CP. The point C and the
10. 10
point P such that line CP is parallel to the plane do not have any image by
the projection.
(b)The projection parallel to a direction D, onto a plane: The image of a point P
is the intersection with the plane of the line parallel to D passing through P.
In an abstract setting, we can generally say that a projection maps is a mapping
of a set or mathematical structures which is idempotent, which means that a
projection is equal to its composition with itself. A projection may also refer to a
mapping that has a left inverse.
Both notions are strongly related as follows: Let P be an idempotent map from a
set E into itself (i.e. π2
= π) and πΉ = π(πΈ) be the image of P. If we denote Ο the
map P viewed as a map from E onto F and by Ξ― the injection of F into E, then we
haveπΞ― = ππ πΉ.
Conversely, πΞ― = ππ πΉ means that πΞ― is idempotent. Also, if X1 and X2 are non-
void (i.e. non empty) sets, we define projection map
Ο1: X1 x X2 β X1 and Ο2: X1 x X2 β X2 by ΟΞ―(x1, x2) = xΞ―.
The aim of this work is to study map projections in Mathematics, the objectives
consist in studying the properties of this map projection as well as their
application. In this work we will be looking at various properties of map projection
has it has been extended or generalized to various mathematical situations,
frequently but not always related to geometry. Some of the areas in which map
projection has been used with its properties are in Set theory, Canonical
projections, Category theory, Orthogonal projection (Linear algebra), in topology
and other area with properties which will be discussed explicitly in the second
chapter of this work.
11. 11
In the third chapter of this work, we will be discussing the application of
projection mapping I different aspect, for example
(a) In the world: projection mapping has enable to map part of the surface of
the Earth onto a plane, i.e. the three dimensional projection
(Cartography)
(b)In set theory: An operation typified by the Ξ―th projection map written projj,
that takes an element x=(x1, x2, x3, . . . , xj, . . . , xk) of a Cartesian product
X1 x X2 x . . . x Xj x . . . x Xk to the value projj(x)=xj.
A mapping that takes an element to its equivalence class under a given
equivalence relation is known as the canonical projection. Also, evaluation map
sends a function f to the value f(x) for a fixed x. the space of functions Yx can be
identified with the Cartesian product ππΞ― and the evaluation map is the projection
map from the Cartesian product.
There are also many applications of map projections in Linear algebra that will
be discussed in the third chapter of this work, and other area of application.
12. 12
1.1 LITERATURE REVIEW
1.1.1 Definition:
SupposeV = U βW. Deο¬ne P: V β V as follows. For each v β V, write v =
u + w with u β U and w β W. Then u and w are uniquely deο¬ned for each v β V.
Put P(v) = u. It is straightforward to verify the following properties of P.
(i) P β L(V ).
(ii) P2 = P.
(iii) U = Im(P).
(iv) W = null(P).
(v) U and W are both P-invariant.
This linear map P is called the projection onto U along W (or parallel to W) and is
often denoted by P = PU,W.
1.1.2 Definition:
Let V be a vector space. A map π΄: π β π is a projection operator if it is linear
and satisο¬es
π΄2
= π΄
We shall assume that π΄ βΆ π β π is a projection operator.
Observe that I βA is also then a projection operator: (πΌ β π΄)2
= πΌ β π΄ (Since
π΄2
= π΄)
If a point y lies in the image of A then it is of the form π΄π₯, for some π₯ β π, and so
then π΄π¦ = π΄ (π΄π₯) = π΄2
π₯
13. 13
π΄2
π₯ = π΄π₯ = π¦. Thus,
π΄π¦ = π¦ If and only if y is in the image of A,
Put another way,
Im(π΄) = ker(I β π΄)
Applying this result to the projection operator I β π΄ gives
ker(π΄) = Im(I β π΄)
Any vector π₯ β π can be expressed as
π₯ = π΄π₯ + (I β π΄)π₯
Where the ο¬rst term π΄π₯ is clearly in the image of π΄ while the second term is
in ker(π΄). Furthermore, this decomposition is unique since any element y which is
in both ker(A) and Im(π΄) must be 0 because π¦ β Im(π΄) implies π¦ = π΄π¦ while
π¦ β ker(π΄) means Ay = 0. [15]
Thus V splits into a direct sum of the subspace Im(π΄) πππ ker(π΄):
π = Im(π΄) β ker(π΄)
14. 14
1.1.3 Definition
Rendering a picture of a three-dimensional object on a flat computer screen
requires projecting points in 3-Space to points in 2-Space. We discuss only one of
many points in β3
to points in β2
that preserves the natural appearance of an
object.
Parallel projection stimulates the shadow that is cast onto a flat surface by a
far away light source, such as the sun.
Also in Introduction to Linear Algebra with application [8], Jim Defranza
and Daniel Gagliardi in 2009, discussed map projection properties such as
homomorphisms, isomorphisms, epimorphisms and other properties.
Colin Adams and Robert Franzosa in their book [1]discussed about the map
projections on topology.
Looking at some properties of map projections, Erwin Kreyszig in
Introduction to Functional Analysis with Application, made mention of it [10].
A Treatise on projection map written by Thomas Craig [6], 1882, from
University of Michigan, also explained the application of map projections to the
world.
15. 15
1.2 DEFINITION OF TERMS
1.2.1 Definition:
Let A and B be a non-empty sets. A relation f from A and B is called a
mapping (or a map or a function) from A to B if for each element x in A there is
exactly one element y in B (called the image of x under f) such that x is a relation f
to y. If f is a mapping from A to B, we write
f: AβB or
π΄
π
β π΅
1.2.2 Definition
i. Let π βΆ π΄ β π΅. The sets A and B are called, respectively, the domain and co
domain of the mapping f.
ii. A mapping f: XβX such that f(x) =x for all x Ι X is called the Identity
mapping on X and is denoted by Ξ―x.
iii. Suppose y0 β Y. Deο¬ne a Constant mapping π: π β π by π(π₯) = π¦0 for
all π₯ β π.
iv. Let A be a non-empty subset of a set X. then the mapping π: π΄ β π such
that π(π) = π for each π Ι π΄ is called an Inclusion map of A into X.
v. Let f: A β B. The subset of B containing of every element that is the image
of some element in A is called the image (or range) of the mapping f and is
denoted by Imf. That is; π°ππ = {π¦ Ι π΅ | π¦ = π(π₯) πππ π πππ π₯ Ι π΄}
16. 16
1.2.3 Definition:
Let X and S be sets. The set of all possible mappings from X to S is denoted
by π π
. That is; π π
= {π | π: π β π}
Suppose X and S are finite sets having m and n elements, respectively. Consider
any mapping π: π β π. The image of any given element in X can be any one
of the n elements in S. Therefore, the m elements in X can be assigned images
in n π₯ n π₯ n π₯ . . . π₯ n = nm
ways. This just means that there are exactly nm
distinct mapping from X to S. Thus, for finite sets X and S, we have the result
that |Sx|=|S||x|
1.2.4 Example
Let X = {0, 1}. Then there are 22=4 distinct mapping from X to X
f1: X β X with 0 0, 1 0
f2: X β X with 0 0, 1 1
f3: X β X with 0 1, 1 0
f4: X β X with 0 1, 1 1
1.2.5 Definitions
A mapping π: π΄ β π΅ is
a) Injective (or one-to-one) if, for all π₯1, π₯2Ι π΄, π₯1 β π₯2 π(π₯1) β π(π₯2) i.e. if
x1 and x2 are distinct elements of A, then f(x1) and f(x2) are distinct elements
of B.
b) Surjective (or onto) if, for every y Ι B, y = f(x) for some x Ι A. That is the
image is the range i.e. if y Ι B, f -1(y) is a non-empty subset of X
17. 17
c) Bijective (or one-to-one-correspondence) if f is injective and surjective. In
this case, there is a function f-1:B β A with f-1 Β° f = IA: A β A and f Β° f -1 =
IB: B β B. Note that f -1: B β A is also bijective and (f -1)-1 = f.
d) Let f : A β B. A mapping π: π΅ β π΄ is called an inverse of f if f Β° g =IB and
g Β° f = IA.
e) A mapping g: B β A is
i. A left inverse of f if g Β° f = IA.
ii. A right inverse of f if f Β° g =IB
f) If π: π΄ β π΅ is a mapping and π β π΄, the mapping from S to B given by a β f
(a) for a Ι S is called the restriction of f to S and is denoted by f |S:S βB
1.2.6 Examples of Induced Mappings
a) Let f: A β B. Let S ΟΉ A and T ΟΉ B such that f(x) Ι T for all x Ι S. Then f
induces the mapping g: S β T given by g(x) = f(x) for all x Ι S. The mapping
g is also called the restriction of f.
b) There is an important mapping determined by a subset of a set. Let S ΟΉ X
and let A = {0, 1}. Then S determines the mapping fS: X β A given by
π(π₯) = 1 ππ π₯ Ι π
π(π₯) = 1 ππ π₯ Ι π
The mapping ππ is called the characteristics function of S.
Conversely, given a mapping π: π β π΄, let S = {π₯ππ | π(π₯) = 1}
Then it is clear that the characteristic function of S is the given mapping f.
Moreover, S is the unique subset of X with f as its characteristic function.
This proves that the mapping πΉ: Ζ€(π₯) β π΄ π₯
by π β ππ , is bijective.
18. 18
It is clear that we could have the set 2 = {1, 2} in a place of A. Thus, there is
a one-to-one-correspondence between the set Ζ€(x) and 2 π
. Because of this,
the set Ζ€(x) is also written as 2 π
.
c) Let E be an equivalence relation on the set X. Then E induces the surjective
mapping Ζ€: X β X|E, with x β E(x) is the equivalence class of x under E.
The mapping is called the canonical mapping from X to the quotient set
X|E.
d) Given sets S and T, there are two canonical mapping with domain S x T β
namely, p: S x T β S with (x, y) β x for all (x, y) Ι S x T and
q: S x T β with (x, y) β y for all (x, y) Ι S x T.
The mapping p and q are called the projections from S x T onto S and T
respectively.
19. 19
CHAPTER TWO
PROPERTIES OF MAP PROJECTIONS
2.1 INTRODUCTION
Here in this chapter, we shall be looking at the properties of map projections in
many areas. It follows that projection mapping play a key role in many areas. In
particular we shall be discussing the following; Properties of metric projection,
projection in Banach space, properties of projections on topological spaces,
generalized and hyper-generalized projection mapping on Hilbert space,
Orthogonal projections, Isomorphism, Homomorphism, Continuity of projection
mappings, and the Product and Sum of projection.
2.2 PROPERTIES OF METRIC PROJECTION
2.2.1 Definition: Let D ΟΉ H be a nonempty subset and let x Ι H. The point
y Ι D is called the metric projection of a point x onto a subset D, if for any
z Ι D there holds the inequality.
||π¦ β π₯|| β€ ||π₯ β π¦||
The metric projection of a point x onto D is denoted by ππ· π₯.
2.2.2Theorem: [2] Let π₯ππ», π·βπ» be a nonempty, con vex and closed
subset and let π¦ππ·. The following conditions are equivalent:
(i) y = PD x,
(ii) < π₯ β π¦, π§ β π¦ > β€ 0 for all π§ Ι π·.
20. 20
Proof:
(i) (ii). Let y = PD (x), z Ι D and let π§ = π¦ + π(π§ β π¦) for π Ι (0; 1).
Obviously, π§π ππ· since D is convex. We have by the properties of the scalar
product
|| π₯ β π¦||2
β€ || π₯ β π§π||2
= || π₯ β π¦ β π(π§ β π¦) ||2
= || π₯ β π¦||2
β 2πβΉπ₯ β π¦, π§ β π¦βΊ+ π2
|| π§ β π¦ ||2
Since Ζ> 0, we have
< π₯ β π¦, π§ β π¦ > β€ π
2β || π§ β π¦||2
If we let π β 0 in the last inequality, we obtain (ii) the limit.
(ii) (i). By the properties of the scalar product and by (ii) we obtain
for any z Ι D
||π§ β π₯||2
= ||π§ β π¦ + π¦ β π₯||2
= ||π§ β π¦||2
+ ||π¦ β π₯||2
+ 2βΉπ§ β π¦, π¦ β π₯βΊ
β₯ ||π¦ β π₯||2
,
which, by the definition of the metric projection, gives (ii)
2.2.3 Corollary:Let π· ΟΉ π» be nonempty, convex and closed. Then Fix PD= D.
Consequently, the metric projection PD is an idempotent operator.
Proof:
If x Ι D, then it follows from the definition of the metric projection that
x = PD x. If x Ι D, then x = PD x since PD x.
21. 21
2.3 HOMOMORPHISMS
2.3.1 Deο¬nition: [7] If G and Gβ are multiplicative groups, a function π βΆ πΊ β πΊβ
is a homomorphism if, for all π, π β πΊ, π(π Β· π) = π(π) Β· π(π).
On the left side, the group operation is in G, while on the right side it is in Gβ. The
kernel of f is deο¬ned by
πππ(π) = πβ1
(πβ) = {π β πΊ βΆ π(π) = πβ}.
In other words, the kernel is the set of solutions to the equation f(x) =eβ. (If Gβ is
an additive group, πππ(π) = πβ1
(0β). Examples The constant map f: G β Gβ
deο¬ned by π(π) = πβ is a homomorphism. If H is a subgroup of G, the inclusion
i: H βͺ G is a homomorphism. The function π: π β π deο¬ned by π(π‘) = 2π‘ is a
homomorphism of additive groups, while the function deο¬ned by π(π‘) = π‘ + 2 is
not a homomorphism. The function β: π β β β {0} deο¬ned by β(π‘) = π‘2
is a
homomorphism from an additive group to a multiplicative group.
2.3.2 Theorem: Suppose G1 and G2 are additive groups. Deο¬ne an addition on
G1ΓG2 by (a1,a2)+(b1,b2)=( a1 +b1,a2 +b2). This operation makes G1ΓG2 into a
group. Its βzeroβ is (01,02) and β(a1,a2)=(βa1,βa2). Theprojections Ο1: G1 ΓG2 β
G2 and Ο2: G1 ΓG2 β G2 are group homomorphisms. Suppose G is an additive
group. We know there is a bijection from {functions f: G β G1 ΓG2} to {ordered
pairs of functions (f1,f2) where f1 : G β G1 and f2 : G β G2}. Underthis bijection, f
is a group homomorphism iο¬ each of f1 and f2 is a group homomorphism.
22. 22
Proof:
It is transparent that the product of groups is a group, so letβs prove the last part.
Suppose G, G1, and G2 are groups and f = (f1,f2) is a function from G to G1 Γ G2.
Now f(a + b)=( f1(a + b),f2(a + b))
And f(a)+f(b)= (f1(a),f2(a))+(f1(b),f2(b)) = (f1(a)+f1(b),f2(a)+f2(b)).
An examination of these two equations shows that f is a group homomorphism if
and only if each of f1 and f2 is a group homomorphism.
2.3.3 Definition: If f: R β Rβ is a bijection which is a ring homomorphism, then
f-1: Rβ β R is a ring homomorphism. Such an f is called a ring isomorphism. In the
case R = Rβ, f is also called a ring automorphism.
2.4 PROJECTION ON BANACH SPACES
A projection in a Banach space is a continuous linear mapping P of the space into
itself which is such that P2=P. Two closed linear manifolds M and N of a Banach
space B are said to be complementary if each z of B is uniquely representable as
x+y, where x is in M and y in N.
This is equivalent to the existence of a projection for which M and N are the range
and null space. It is therefore also true that closed linear subsets M and N of B are
complementary if and only if the linear span of M and N is dense in B and there is
a number π > 0 such that ||x + y|| β₯ Ξ΅||x|| if x is in M and y in N. It is known
that a Banach space M is complemented in each Banach space in which it can be
embedded if it is isomorphic with a complemented subspace of the space (m) of
bounded sequences. In particular, if M is a subspace of a Banach space Z and is
23. 23
isometric with a subspace M' of (m), then there is a projection of Z onto M of norm
less than or equal to X if there is a projection of (m) onto M' of norm equal to Ζ.
Thus the existence of a complement in (m) for a subspace M of (m) is independent
of the method by which M is embedded in (m). Any separable Banach space is
isometric with a subspace of (m). Hence a separable Banach space is
complemented in each space in which it can be embedded if and only if it is
complemented in (m). It has been conjectured that no separable subspace of (m) is
complemented in (in), or perhaps that separable complemented subspaces of (m)
are reflexive. There are two classes of separable Banach spaces which are known
to be not complemented in (m). These are the separable Banach spaces which have
a subspace isomorphic with (c0) and the separable Banach spaces whose first
conjugate space is not weakly complete, of which h is an example. It is interesting
to note that (c0) is complemented in any separable Banach space B in which it can
be embedded. In particular, the projection of B onto (c0) can be of norm 2 but there
are separable Banach spaces containing (c0) for which the norm of a projection
onto (c0) must have norm as large as 2.
2.5 ORTHOGONAL PROJECTION
We begin by describing some algebraic properties of projections. If M and N are
subspaces of a linear space X such that every x β X can be written uniquely as
x = y+z with y β M and z β N,
then we say that X = MβN is the direct sum of M and N, and we call N a
complementary subspace of M in X.
The decomposition x = y + z with y β M and z β N is unique if and only if
26. 26
Let t: = 2q + d(0, C), so that q + t< r . Given w,wβ Ι qBX, let us set
C:= C + w β wβ, u: = pC (wβ)
Then, observing that, for any y β C,
we have
|| (π’ + π€ β π€β) β π€|| = π’ β π€ β€ ||π¦ β π€β|| = ||(π¦ + π€ β π€β)β π€β||
ππ( π€) = ππ ( π€β) + π€ β π€β
.
Since by definition of C we have π(πΆ, πΆβ ) β€ ||π€ β π€β|| , we note that
π(0, πΆβ ) β€ π(0, πΆ) + π(πΆ, πΆβ ) β€ π‘
and πΆ, πΆβ β π‘(π ). Thus, we have
Since
||Pc(w)β pC (wβ)|| β€ ||pC (w) β pC (wβ) + ||pC (w) β Pc(wβ )
β€ Ξ³-1 2h(r )||w β wβ||) + || w β wβ||(w) β pCc(w),
we get the second inequality.
2.7 PROJECTION ON TOPOLIGICAL GROUP
2.7.1 Theorem:Let G be a topological group and H be a subgroup of G. Then the
canonical projection π : πΊ β πΊ = π» is an open map.
Proof:
Let V be an open subset of G. Then it implies that the image Ζ₯ (V) is open in G/H if
and only if π -1(π (V)) is open in G. This set however is easily seen to equal VH,
27. 27
which is the union of the translate V h as h ranges through H and is therefore an
open set.
2.7.2 Definition: Let H be a subgroup of a topological group G. The quotient
topology on G/H is defined such that a set U is a subset of G/H and U is open if
and only if π β1
(U ) is open in the topology of G, where π βΆ πΊ β πΊ/π» is the
canonical projection
2.8 GENERALIZED AND HYPER-GENERALIZED PROJECTION
Let H be a separable Hilbert space and L(H ) be a space of all bounded linear
operators on H . The symbols R(A), N (A) and Aβ denote range, null space and
adjoint operator of operator A β L(H ). Operator A β L(H ) is a projection
(idempotent) if π΄2
= π΄, while it is an orthogonal projection if π΄
β
= π΄ = π΄2
.
Operator is hermitian (self adjoint) if π΄ = π΄β
, normal if π΄π΄β
= π΄β
π΄ and unitary if
π΄π΄β
= π΄β
π΄ = πΌ . All these operators have been extensively studied and there are
plenty of characterizations both of these operators and their linear combinations.
The Moore-Penrose inverse of π΄ β πΏ(π» ), denoted by π΄β
, is the unique solution
to the equations
π΄π΄β
π΄ = π΄, π΄β
π΄π΄β
= π΄β ,
(π΄π΄β
)β
π΄π΄β
,(π΄β
π΄)β
= π΄β
π΄.
Notice that A exists if and only if R(A) is closed. Then π΄π΄β
is the orthogonal
projection onto R(A) parallel to π (π΄β
), and π΄β
π΄ is the orthogonal projection onto
R(Aβ) parallel to N (A).
Consequently, πΌ β π΄π΄β
is the orthogonal projection onto N (Aβ) and πΌ β π΄β
π΄ is the
28. 28
orthogonal projection onto N (A).
For π΄ β πΏ(π» ), an element π΅ β πΏ(π» ) is the Drazin inverse of A, if the
following hold:
π΅π΄π΅ = π΅, π΅π΄ = π΄π΅, π΄ π+1
π΅ = π΄ π
for some non-negative integer n. The smallest such n is called the Drazin index of
A. By AD we denote Drazin inverse of A and by ind(A) we denote Drazin index of
A.
If such n does not exist, πππ(π΄) = β and operator A is generalized Drazin
invertible. Its inverse is denoted by Ad. Operator A is invertible if and only if
πππ(π΄) = 0.
If πππ(π΄) β€ 1, operator A is group invertible and AD is its group inverse, usually
denoted by π΄β
.
Notice that if the Drazin inverse exists, it is unique. Drazin inverse exists if R(An)
is closed for some non-negative integer n.
Operator A β L(H ) is a partial isometry if AAβ A = A or, equivalently, if π΄β
=
π΄β
. Operator A β L(H ) is EP if π΄π΄β
= π΄β
π΄, or, in the other words, if π΄β
=
π΄ π·
= π΄β
. Set of all EP operators on H will be denoted by Β£Ζ€(H ). Self-adjoint
and normal operators with closed range are important subset of set of all EP
operators. However, converse is not true even in a finite dimensional case.
These operators extend the idea of orthogonal projections by deleting the
idempotency requirement. Namely, we have the following definition:
30. 30
β π ) = πΈ(ββ π).
Moreover, A has the following spectral representation
π΄ = β« πππΈπ , π(π΄) π€βπππ πΈπ = πΈ(π) is the spectral projection associated
with the point π β π(π΄).
From π΄4
= π΄, we conclude π΄3
π (π΄) = πΌπ (π΄) and π3
= 1, or, equivalently
π(π΄) β {0,1, π
2ππ
3β
, π
β2ππ
3β
}. Now,
π΄ = 0πΈ(0) β 1πΈ(1) β π
2ππ
3β
πΈ (π
2ππ
3β
) β π
β2ππ
3β
πΈ (π
β2ππ
3β
), here E(Ξ») is
the spectral projection of A associated with the point π β π( π΄) such that πΈ( π) =
0 if π β π( π΄), πΈ( π) = 0 if π β {0, 1, π
2ππ
3β
, π
β2ππ
3β }
Ο(A)β and πΈ(0) β
πΈ(1) β πΈ (π
2ππ
3β
) β πΈ(π
β2ππ
3β
) = πΌ . From the fact that π(π΄2
) = π(π΄β
) and
from uniqueness of spectral representation, we get π΄2
= π΄β
.
(a β c) If π΄β
= π΄2
, then π΄4
= π΄π΄2
π΄ = π΄π΄β
π΄ = π΄. Multiplying from the left
(from the right) by π΄β
, we get π΄β
π΄π΄β
π΄ = π΄β
π΄ (π΄π΄β
π΄π΄β
= π΄π΄β
), which proves
that π΄β
π΄ (π΄π΄β
) is the orthogonal projection onto π (π΄β
π΄)= π (π΄β
) = π (π΄)
β₯
(π (π΄π΄β
) = π (π΄) = π(π΄β
)
β₯
), i.e., π΄(π΄β
) is a partial isometry.
(c β a) If A is a partial isometry, we know that AA* is orthogonal projection onto
π (π΄π΄) = π (π΄). Thus, π΄π΄β
π΄ = π π ( π΄)
π΄ = π΄ and π΄ π΄2
= π΄β
= π΄π΄2
π΄ = π΄
implies π΄2
= π΄β
31. 31
2.9 Projection in Vector Space
Let V be a vector space over a field K. Let B={x1, x2, x3,β¦, xn} be a base of V over
K. Then for all xβV there exist unique scalars Ξ»1, Ξ»2,β¦,Ξ»n in K such that
βππ = 1ππ π₯π = π₯.
We call the ππ the i-th component of x with respect to base B. Let pi(x) denote the
π β π‘β component of x with respect to base B, where ππ is a projection map.
2.9.1 Theorem:
1. ππ: π β πΎ is a linear map and therefore ππ β π»πππ(π, πΎ).
2. {p1,p2,β¦,pn} is a base of π»πππΎ(π, πΎ).
3. There is an isomorphism πΉ: π β π»πππΎ(π, πΎ) such that πΉ(π₯π) = ππ for all
1 β€ π β€ π.
Proof:
Let π£, π€ β π, where π£ = π1 π₯1 + β― π π π₯ π and π€ = π1 π₯ π + β― π π π₯ π. Since π£ + π€ =
(π1 + π1)π₯1 + β―+ (π π + π π)π₯ π, we have ππ(π£ + π€) = ππ + ππ = ππ(π£) + ππ(π€).
Now suppose π β πΎ. Clearly,ππ£ = (ππ1)π₯1 + β―+ (ππ π)π₯ π, and thus ππ(ππ£) =
πππ = πππ(π£).
Assume ππ β πΉ such that
π1 π1 + β―+ π π π π = 0 β π»πππΎ(π, πΎ)
If ππ β 0 for some i, we have
(c1 p1+β―+cn pn)(xi)=c1 p1(xi)+β―+cn pn(xi)=ci pi(xi)=ciβK
34. 34
Multiplying this by π2 from the right, we have 2π2 π1 π2 = 0, so that π2 π1 = 0 and
π1 β₯ π2.
Conversely, if π1 β₯ π2, then π1 π2 = π2 π1 = 0. This implies π2
= π. Since π1
and π2 are self-adjoint, so is π = π1 + π2 . Hence π is a projection.
(b) We determine the closed subspace π β π» onto which π projects. Since π =
π1 + π2 , for every π₯ β π» we have
π¦ = ππ₯ = π1 π₯ + π2 π₯
Here π1 π₯ β π1 and π2 π₯ β π2. Hence π¦ β π1 β π2 , so that π β π1 β π2. We
show that π β π1 β π2 Let π£ β π1 β π2be arbitrary. Then π£ = π¦1 + π¦2 Β·
Here, π¦1 β π1 and π¦2 β π2. Applying P and using π1 β₯ π2, we thus obtain
ππ£ = π1 ( π¦1 + π¦2) + π2( π¦1 + π¦2) = π1 π¦1 + π2 π¦2 = π¦1 + π¦2 = π£.
Hence, π£ β π πππ π β π1 β π2. Together, π = π1 β π2.
35. 35
CHAPTER THREE
APPLICATION OF MAP PROJECTION
3.1 TOPOLOGICALOBSTRUCTIONS AND POLYTOPE PROJECTIONS
We devise a criterion for projections of polytopes that allows us to state when a
certain subcomplex may be strictly preserved by a projection. We associate an
embedding problem to the projection problem. Then we describe methods from
combinatorial topology, which may yield obstructions to the associated
embeddability problem. Finally, we specialize the obstructions to the problem of
preserving certain skeleta of polytopes by projections.
3.1.1 Associated Polytope And Subcomplex
We build a bridge between projection problems and embeddability problems as
follows: We associate a polytope with certain simplex faces to a projection of a
polytope with certain strictly preserved faces via Gale duality. The simplex faces
of this associated polytope form a simplicial complex. If we can show that this
simplicial complex cannot be embedded into the boundary of the associated
polytope, then there is no realization of the polytope that allows for a projection
preserving the given subcomplex.
Sanyal uses the same approach to analyze the number of vertices of Minkowski
sums of polytopes, since Minkowski sums are projections of products of polytopes.
The vertices of a (simple) polytope P give rise to a simplicial complex Ξ£0. If π βΆ
π β π(π) is a projection preserving the vertices, then Ξ£0 is realized in a
(simplicial) sphere whose dimension depends on dimΟ(P). So if the simplicial
complex Ξ£0 cannot be embedded into that sphere then there exists no realization of
the polytope such that all vertices survive the projection.
36. 36
Proposition 3.1.2: [14] Let Ο :β π
ββ π
be the projection to the ο¬rst e coordinates
of a d-polytope π given by its facet inequalities (A(e),A(dβe))( π₯
π₯β²) β€ 1 with A(e) β
βmΓe, A(dβe) β βmΓ(dβe), x β βe, and xβ² β βdβe. If for each facet F of P at least one
vertex vβ F survives the projection then the rows of A (dβe) are the Gale transform
of a polytope.
Proof: The rows of the matrix A (dβe) are the Gale transform of a polytope if
for every row ai
(dβe) (i β [m]) the remaining rows of A(dβe) ai
(dβe) are
positively spanning. But for every facet F there exists a vertex v β F that
survives the projection. Hence, the truncated normals corresponding to the
facets containing this vertex positively span βdβe. Thus A (dβe) is the Gale
transform of a polytope. So if we project a d-polytope to βe such that some
of the vertices survive the projection as described in the above proposition
we obtain a polytope by Gale duality.
Deο¬nition 3.1.3 (Associated polytope): Let Ο be a projection of a d-dimensional
polytope P on m facets to βe that preserves one vertex v β F for every facet F of P.
Then the (π β (π β π) β 1) βdimensional polytope on m vertices obtained via
Gale transformation as described in Proposition 3.1.2 is the associated polytope
π΄(P,Ο).
Further every face G that is preserved by the projection yields an associated face
AG = [m]HG of the associated polytope π΄(π, π) since Gale duality transforms
positively spanning vectors into faces of the polytope. All these associated faces AG
are simplices. This yields the following subcomplex of the associated polytope.
Deο¬nition 3.1.4 (Associated subcomplex): Let Ο be a projection of a d-
dimensional polytope P on m facets to βe that preserves one vertex vβF for every
37. 37
facet F of P, and let S be the subcomplex of P that is preserved under projection.
Then the associated subcomplex πΎ(π, π) is the simplicial complex:
πΎ(π, π) βΆ= {[π] π» πΊ | πΊ β π}.
The subcomplex consists of all the facets and their faces.
Now we obtain the following theorem which links the projection of a polytope
preserving certain faces with the embedding of the associated subcomplex into the
associated polytope.
Example 3.1.5:(Projectionof the product of triangles preserving all vertices).
We will use the technique developed in this section to show that there exists no
realization of the product (β2)2 ββ4 of two triangles β2 such that the projection
Ο :β4β β2 to the plane preserves all 9 vertices.
The product of two triangles is a 4-polytope on 6 facets. Since the projection is to
β2, the associated polytope A((β2)2,Ο) is a 3-dimensional polytope. Let us label the
facets of the two triangles by a0,a1,a2 and aβ²0,aβ²1,aβ²2. These are also the labels of the
vertices of the associated polytope A((β2)2,Ο). Each vertex of the product lies on
two facets corresponding to two edges of each of the factors. Thus the associated
complex K((β2)2,Ο) has an edge for every pair (ai,aβ²j) with i,j β [3]. So if there
exists a projection of the product of two triangles to the plane preserving all its
vertices, then this yields an embedding of the complete bipartite graph on 3 + 3
vertices K3,3 into the boundary of a 3-polytope. But since K3,3 is not planar there
exists no 3-polytope with K3,3 in its boundary. This implies that there exists no
realization of (β2)2 such that all vertices survive the projection to the plane. In the
above example we used the non-planarity of the graph K3,3 as a topological
38. 38
obstruction to show that the projection of a product of two triangles to the plane
cannot preserve all the vertices.
3.2 STEREOGRAPHIC PROJECTION
Definition 3.2.1:[9] Let S2 denote the unit sphere x2+y2+z2 =1 in R3 and let N = (0,
0, 1) denote the "north pole" of S2. Given a point M β S2, other than N, then the
line connecting N and M intersects the xy-plane at a point P. Then stereographic
projection is the map
π: π2
β {π} β πΆ: π βΌ π.
Definition 3.2.2: Consider the unit sphere x2 + y2 + z2 = 1 in three dimensions,
capped by the tangent plane z = 1 through the North Pole. We want to deο¬ned a
projection from the sphere onto this plane. If P is a point on the sphere, let P' be
the intersection of the ray from the south pole Ξ = (0, 0, β1) to P with the plane.
This deο¬nition fails if P is Ξ itself. Therefore stereographic projection maps all
points on the sphere except Ξ to a point on the polar plane, and its inverse wraps
the plane around the complement of Ξ .
Explicitly, if π = (π₯, π¦, π§) π€ππ‘β π§ β β1 then the parameterized line through Ξ
and π ππ π + π‘(π± β π) = (π‘π₯, π‘π¦, 1 β π‘ β π‘π§). This intersects π§ = 1 when
π‘ =
2
(1 + π§)
which makes
39. 39
π β² = (π, π, 0), π =
2π₯
(1 + π§)
, π =
2π¦
(1 + π§)
Example 3.2.3: Take A = C. Then β1
(β)β²
is the set of one-dimensional linear
subspaces of β2. We can choose a unique basis of x ββ1
(β)β²
of the form (1, z)
unless x = (0, z),z β β{0}, and βx = β(0,1). In this way we obtain a bijective map
from β1
(β)β²
to cβͺ {β}, the extended complex plane. Using the stereographic
projection, we can identify the latter set with a 2-dimensional sphere. The complex
coordinates make it into a compact complex manifold of dimension 1, the Riemann
sphere ββ1
.
Theorem 3.2.4: The image of a straight line in β under stereographic projection is
a circle through π, with π excluded. The image of a circle in β under
stereographic projection is a circle not containing π. The inverse image of any
circle on π2 is a straight line together with β if the circle passes through π,
otherwise a circle.
Proof: Since a straight line in the π₯1 π₯2-plane together with π determines a
unique plane, the intersection of which with π2 is the image of the straight
line we only need to consider the case of a circle in β. If it has center π and
radius r its equation is |π§ β π|2
= π2
or| π§|2
β 2π π( πΜΏπ§) + | π|2
= π2
.
Substituting π§ =
π₯1+ππ₯2
1βπ₯3
into this, using that π₯1
2
+ π₯2
2
+ π₯3
2
= 1 and π₯ β
1, we get 1 + π₯3 β 2π₯1 π π π β 2π₯2 πΌπ π + (1 β π₯3)(| π|2
β π2) = 0
which is the equation of a plane. [5]Conversely, a circle on the Riemann
sphere is determined by three distinct points. The inverse images of these
three points determine a circle in β. The image of this circle is clearly the
original circle.
40. 40
3.3 ORTHOGONAL PROJECTION
Rendering a picture of a 3-dimensional object on a flat computer screen requires
projecting points in 3-Space to a point in 2-Space. We will discuss only one of
many methods to project points in β3
to points in β2
that preserve the natural
appearance of an object.
Parallel projection simulates the shadow that is cast onto a flat surface by a far
away light source, such as sun. The Figure below shows rays intersecting an object
in 3-Space projection into 2-Space.
The orientation of the axes in the figure above is such that the π₯π¦- plane represents
the computer screen.
To show how to find the π₯π¦ coordinates of the projected point, let the vector
ππ = [
π₯ π
π¦ π
π§ π
]
represent the direction of the rays. If π₯0, π¦0, π§0 is a point β3
, then the parametric
equations of the line going through the point and in the direction of ππ are given by
{
π₯( π‘) = π₯0 + π‘π₯ π
π¦( π‘) = π¦0 + π‘π¦ π
π§( π‘) = π§0 + π‘π§ π
β π‘ β β
The coordinates of the projection of (π₯0, π¦0, π§0) onto the π₯π¦ plane are found by
letting π§( π‘) = 0. Solving for all π‘2 we obtain
π‘ =
βπ§0
π§ π
41. 41
Now, substituting this value of π‘ into the first two equations above, we find the
coordinates of the projected point, which are given by
π₯ π = π₯0 β
π§0
π§ π
π₯ π, π¦π = π¦0 β
π§0
π§ π
π¦ π πππ π§ π = 0
The components of ππ can also be used to find the angles that the rays make with
the π§-axis and π§-plane. In particular, we have
tan Ο =
π¦ π
π₯ π
πππ tan π =
β π₯ π
2 + π¦ π
2
π§ π
where π is the angle ππ makes with the π₯π§ plane and π is the angle made with the
π§-axis. On the other hand, if the angles π and π are given, then these equation can
be used to find the projection vector ππ.
3.4 MATRIX PROJECTION
3.4.1 Projection to a Line
[4] Projection matrix π projects vector b to a.
Let π b= πa= π, error π = π β π
βΉ π β₯ π β πT π = 0 = π π( π β π) = π π( π β ππ)
βΉ π π( π β ππ) where π is a scalar
βΉ ππ π
π = π π
π
βΉ π =
π π
π
π π π
=
π. π
π. π
ππ = ππ = ππ = π
π π
π
π π π
βΉ
ππ π
π π π
π βΉ π =
ππ π
π π π
45. 45
3.5.1 Orthographic
The orthographic projection is how the earth would appear if viewed from a distant
planet. Since the light source is at an infinite distance from the generating globe,
all rays are parallel. This projection appears to have been first used by astronomers
in ancient Egypt, but it came into widespread use during World War II with the
advent of the global perspective provided by the air age. It is even more popular in
today's space age, often used to show land-cover and topography data obtained
from remote sensing devices. The generating globe and half-globe illustrations in
this bookare orthographic projections, as is the map on the front cover of the book.
The main drawback of the orthographic projection is that only a single hemisphere
can be projected. Showing the entire earth requires two hemispherical maps.
Northern and southern hemisphere maps are commonly made, but you may also
see western and eastern hemisphere maps.
3.5.2 Stereographic
Projecting a light source from the antipodal point on the generating globe to the
point of tangency creates the stereographic projection. This is a conformal
projection, so shape is preserved in small areas. The Greek scholar Hipparchus is
credited with inventing this projection in the second century BC. It is now most
commonly used in its polar aspect and secant case for maps of polar areas. It is the
projection surface used for the Universal Polar Stereographic grid system for polar
areas, as we will see in the next chapter. A disadvantage of the stereographic:
conformal projection is that it is generally restricted to one hemisphere. If it is not
restricted to one hemisphere, then the distortion near the edges increases to such a
degree that the geographic features in these areas are basically unrecognizable. In
past centuries, it was used for atlas maps of the western or eastern hemisphere.
46. 46
3.5.3 Azimuthal equidistant
The azimuthal equidistant projection in its polar aspect has the distinctive
appearance of a dart boardβequally spaced parallels and straight-line meridians
radiating outward from the pole. This arrangement of parallels and meridians
results in all straight lines drawn from the point of tangency being great circle
routes. Equally spaced parallels mean that great circle distances are correct along
these straight lines. The ancient Egyptians apparently first used this projection for
star charts, but during the air age it also became popular for use by pilots planning
long-distance air routes. In the days before electronic navigation, the flight
planning room in major airports had a wall map of the world that used an oblique
aspect azimuthal equidistant projection centered on the airport. You will also find
them in the public areas of some airports. All straight lines drawn from the airport
are correctly scaled great circle routes. This is one of the few planar projections
that can show the entire surface of the earth.
3.5.4 Lambert Azimuthal Equal Area
In 1772 the mathematician and cartographer Johann Heinrich Lambert published
equations for the tangent case planar Lambert azimuthal equal area projection,
which, along with other projections he devised, carries his name. This planar equal
area projection is usually restricted to a hemisphere, with polar and equatorial
aspects used most often in commercial atlases. More recently, this projection has
been used for statistical maps of continents and countries that are basically circular
in overall extent, such as Australia, North America, and Africa. You will also see
the oceans shown on maps that use the equatorial or oblique aspects of this
projection. The Lambert azimuthal equal area projection is particularly well suited
for maps of the Pacific Ocean, which is almost hemispheric in extent.
48. 48
In this project work, we have been able to see the introduction, concept and
relation to real life of map projections. We have also seen some of the properties of
map projections and the some applications of map projection in relation to
different aspects of mathematics and some of its applications to the real world.
REFERENCES
[1]C. Adams and R. Franzosa, Introduction to Topology, Pearson (2006)
49. 49
[2]Andrzej Cegielski, Properties of Metric Projection, Institute of Mathematics
Technical University Ilmenau (2008)
[3] Apollonius of Perga, Treatise on conic sections, translated by T. L. Heath,
Cambridge University Press, 1896.
[4]C. Bennewitz Complex Analysis Fall 2006
[5]P.B Bhattacharys et al; Basic Abstract Algebra; Cambridge University Press;
1995
[6]Thomas Craig; Map Projections and Its Applications; 1882
[7]E.H Conell, Element of Abstract Algebra (1999)
[8]J. Defranza et al; Introduction to Linear Algebra with Applications; Mac Graw
Hill; 2009
[9]Igor V. Dolgachev, Introduction to Algebraic Geometry August 19, 2013
[10]Erwin Kreyszig Introductory Functional Analysis with Applications University
of Windsor John Wiley & Sons New York Santa Barbara London Sydney Toronto
1978
[11]Hunter, Bounded Linear Operator on Hilbert Space, University of Califonia,
Davic United States
[12]Ling-Hsiao Linear Algebra, Projection and Projection Matrix, Institute of
Space Science, National Central University Lyu (2012)
[13]S.Radosavljevi Pairs of projections on a Hilbert space: properties and
generalized invertibility, Department of Mathematics LinkΓΆping University,
LinkΓΆping, Sweden 2012.
50. 50
[14]R. Sanyal, Topological obstructions for vertex numbers of Minkowski sums,
Journal of Combinatorial Theory. Series A(2008).
[15]A. Sengupta; Funcional Analysuis; Fall (2002).