52994 D. M. T. B. Dissanayake and A. A. I. Perera / Elixir Dis. Math.

129 (2019) 52994-52995

Available online at www.elixirpublishers.com (Elixir International Journal)

Discrete Mathematics
Elixir Dis. Math. 129 (2019) 52994-52995

Applications of Matroid Theory in Combinatorial Optimization and

Projective Geometry
D. M. T. B. Dissanayake and A. A. I. Perera
Department of Mathematics, University of Peradeniya, Sri Lanka.

ARTICLE INFO AB S T RA C T
Art i cl e h i sto ry : A Matroid is a set with an independent structure defined on it. A Matroid abstracts and
Received: 27 February 2019; generalizes the notion of linear independence in vector spaces and independence in
Received in revised form: graphs. Matroids unite the concepts of graph theory, linear algebra, projective geometry,
25 March 2019; transversal theory, and combinatorial optimization. Applications of Matroids involve
Accepted: 5 April 2019; different areas such as combinatorial optimization, network theory, coding theory and
many other areas. Matroids can be found in projective geometry; the fano plane of order
K ey w o rd s 2 gives rise to a Matroid. An important application of Matroids in optimization involves
Greedy Algorithm, the greedy algorithm. Kruskal’s algorithm for finding a minimal spanning tree which is
Kruskal’s Algorithm, an example of the greedy algorithm can be used to understand how Matroids can be
Linear Independence, involved in the greedy algorithm. Consider a network of vertices with weighted links
Matroid Theory. between the vertices. Our goal is to find a collection of links that connect all vertices
using the smallest weight. That is a spanning tree with minimal weights. Kruskal’s
algorithm can be generalized to a Matroid by taking a Matroid 𝑴 and a function 𝒘: 𝑴 →
ℝ which assigns weights to each element. The goal is to find the basis 𝑩 of 𝑴 such that
∑𝒘(𝒙) where 𝒙 ∈ 𝑩 is minimized. The greedy algorithm is a characterization of the
Matroid. Matroids are the structures in which the greedy algorithm works successfully.
© 2019 Elixir All rights reserved.

Introduction A Matroid can be defined in terms of bases as well. A

Matroids is a generalization of independence in graphs Matroid consists of a non-empty finite set and a non-
and linear independence in vector spaces. It was first empty collection of subsets of , called bases, satisfying
introduced by the American Mathematician Hassler Whitney the following properties:
in 1935. There are several equivalent ways to define a (i).no base properly contains another base;
Matroid. The two most important definitions of Matroids are (ii). if and are bases and if is any element of ,
in terms of independent sets and bases. Proof of each of the then there is an element of such that ( * +) * + is
properties of these definitions can be found in Oxley [2], also a base.
Whitney [4], and Wilson [5]. Examples of Matroids
Applications of Matroids involve different areas such as A -uniform on is the Matroid whose bases are those
combinatorial optimization, projective geometry, network subsets of with exactly elements. Independent sets are
theory, coding theory and many other areas. Matroids are those subsets of with not more than elements. A cyclic
connected to many branches of Mathematics such as linear Matroid of a graph is the Matroid defined by taking as
algebra, graph theory, combinatorial optimization, finite the set of edges of and bases as the edges of the spanning
geometry and abstract algebra. forest of . It is denoted by ( ). The independent sets of
Preliminaries
( ) are those sets of edges of that contain no cycle. If
Definitions (as in Wilson [5])
is a finite set of vectors in a vector space , then the vector
The definition of a Matroid in terms of independent sets
Matroid is defined by taking as bases all linearly independent
is as follows. A Matroid consists of a non-empty finite set
subsets of that span the same subspace as E.
and a non-empty collection of subsets of (called
Two Matroid and are said to be isomorphic if
independent sets) satisfying the following properties:
(i) Any subset of an independent set is independent; there is a one-one correspondence between their underlying
(ii) If and are independent sets with | | sets and . When a Matroid is isomorphic to the cycle
| |, then there Matroid of some graph, it is said to be a graphic Matroid.
is an element contained in but not in , such that * + Given a Matroid on a set , is said to be representable
is independent. over a field , if is isomorphic to ( ) for some graph .
With this definition a base is defined to be a maximal A transversal Matroid is obtained by taking a non-empty
independent set and a cycle is defined to be a minimal finite set and the partial transversals of , where
dependent set. ( ) is a family of non-empty subsets of , as the
independent sets.

Tele:
E-mail address: [email protected]
© 2019 Elixir All rights reserved.
52995 D. M. T. B. Dissanayake and A. A. I. Perera / Elixir Dis. Math. 129 (2019) 52994-52995
Results problem is an example of a system of distinct representatives
One of the most important application of the Matroids is or SDR [1]).
in the field of combinatorial optimization. It is the association Matroids are also applied in projective geometry. To
between Matroids and greedy algorithms. One important understand the connection between Matroid theory and
property of Matroids is that they are the structures in which projective geometry the Fano plane of order 2 can be
the greedy algorithm works successfully. considered. Here, it gives rise to a Matroid named the Fano
Consider a network of vertices with links between the Matroid.
vertices, and each link has a weight. Then goal is to find a
collection of links that connect all vertices, using the smallest
weight. That is a spanning tree with minimal weights. For this
problem, Kruskal’s algorithm for finding a minimal spanning
tree which is an example of the greedy algorithm, guarantees
to find a minimal spanning tree. This algorithm can be
generalized to any Matroid. Consider a Matroid and a
function : → ℝ which assigns weights to each element.
The goal is to find the basis of such that ∑ ( ) where Figure 1. Fano Matroid.
∈ is minimized. The optimal solution will necessarily be Conclusion(s)
a basis. The algorithm is as follows: Matroids unite the concepts of graph theory, linear
1. Initialize a set of elements to be the empty set . algebra, projective geometry, transversal theory, and
2. Sort the elements of according to weight. combinatorial optimization. This research work considered
3. Run through the elements, starting with the smallest only certain applications of Matroid theory which shows the
weighted elements. For each element , add to unless close connection between Matroids and greedy algorithm in
* + is dependent. combinatorial optimization, and projective geometry.
Any time the greedy algorithm guarantees an optimal Kruskal’s algorithm which is an example of the greedy
solution for all weight functions, the underlying structure algorithm can be used to understand how the greedy
must be a Matroid. Moreover, only when the structure is a algorithm can be generalized using Matroids. Fano plane of
Matroid, the greedy algorithm guarantees to return an optimal order 2 can be considered to understand the connection
solution. between Matroid theory and projective geometry. There are
Other examples for which the greedy algorithm and many applications of Matroids in different areas. To
Matroids can be involved are the task scheduling problem and understand the concepts of these applications further study in
assignment problem. The latter can be solved by applying the Matroids is needed. Future work includes the study of the
greedy algorithm to the transversal Matroid. definition of Matroid for finite geometries of dimension
Let * + be a set of works ordered by their importance beyond 2.
(priority) and let * + be a set of employees capable to do one References
or various of these works. We suppose that the works will be [1] Lawler, E. (2001). Combinatorial Optimization:
done at the same time (and thus each employee can do just Networks and Matroids, Dover Publications,
one work each time). The problem is to assign the works to Mineola, NY.
the employees in an optimal way, that is to maximize the [2] Oxley, J. (1992). Matroid Theory, Oxford Science
priorities. Publications, Oxford University Press, Oxford.
The problem can be solved by applying the greedy [3] White, N. (1992.) Matroid Applications, Cambridge
algorithm to the transversal Matroid where * + is the
University Press.
non-empty finite set and * + with the set of
[4] Whitney, H. (1935). “On The Abstract Properties of
works for which employee is qualified. It can be noticed that
the maximal number of works that can be done at the same Linear Independence”, American Journal of
time is equal to the biggest partial transversal of with the Mathematics, Vol 57. John Hopkins University
function : → ℝ corresponding to the importance of the Press.
work. [5] Wilson, R. (1996). Introduction to Graph Theory,
Using a similar attempt, determining an optimal schedule Harlow, Pearson Education Limited.
can be done by identifying the transversal Matroid and by
applying the greedy algorithm (the solution to the scheduling