THE MAXIMUM SIZE OF COMBINATORIAL GEOMETRIES EXCLUDING WHEELS AND WHIRLS AS MINORS DISSERTATION. Presented to the Graduate Council of the

Size: px
Start display at page:

Download "THE MAXIMUM SIZE OF COMBINATORIAL GEOMETRIES EXCLUDING WHEELS AND WHIRLS AS MINORS DISSERTATION. Presented to the Graduate Council of the"

Transcription

1 ziq /VQfd THE MAXIMUM SIZE OF COMBINATORIAL GEOMETRIES EXCLUDING WHEELS AND WHIRLS AS MINORS DISSERTATION Presented to the Graduate Council of the University of North Texas in Partial Fulfillment of the Requirements For the Degree of DOCTOR OF PHILOSOPHY By James V. Hipp, Jr., B.S., M.S, Denton, Texas August, 1989

2 OK Hipp, James, Jr., The Maximum Size of Combinatorial Geometries Excluding Vheels and Whirls as Minors. Doctor of Philosophy (Mathematics), August, 1989, 68 pp., 15 illustrations, bibliography, 16 titles. Ve show that the maximum size of a geometry of rank n excluding the (q + 2) point line, the 3 wheel fg, and the 3 whirl 3F 3 as minors is (n - l)q + 1, and geometries of maximum size are parallel connections of (q + 1) point lines. Ve show that the maximum size of a geometry of rank n excluding the 5 point line, the 4 wheel and the 4 whirl as minors is 6n 5, for n > 3. Examples of geometries having rank n and size 6n 5 include parallel connections of the geometries V^g and PG(2,3). A long line is a line with more than two points. A chain of long lines ^,, ^ has the property that if i j, 1fl = 0 unless i j =1. A ring of long lines I» ^ has the property that i^, 1 2, a chain, and ^ meets only I^ and long lines; W not binary;. The geometry ^ is a binary ring of three is a ring of three 3 point lines that is is a binary rank 4 ring of four 3 point lines; is a rank 4 ring of four 3 point lines that is not binary. The geometry V^g is a rank 4 geometry consisting of two copies of PG(2,3), with three points removed from their line of intersection. The proofs make use of chains and rings of long lines.

3 TABLE OF CONTENTS Page LIST OF ILLUSTRATIONS iv Chapter I. INTRODUCTION 1 Matroids Matroid Minors Uniform, Binary, and Ternary Matroids Direct Sums and Parallel Connections Minor-Closed Classes of Matroids Size Functions and Growth Rates II. THE GROWTH RATE OF THE CLASS OF GEOMETRIES EXCLUDING THE (q+2)-p0int LINE, THE 3 WHEEL, AND THE 3-WHIRL AS MINORS 20 An Upper Bound on the Size Function Extremal Geometries III. THE GROWTH RATE OF THE CLASS OF TERNARY GEOMETRIES EXCLUDING THE 4-WHEEL AND THE 4 WHIRL AS MINORS 30 BIBLIOGRAPHY 67 in

4 LIST OF ILLUSTRATIONS Figure Page 1. PG(2,2) and PG(2,3) 2 2. The k Wheel Graph r 3, r 4, and r v i The Geometry N The Geometry N when k = r(c U 0 = 4 and \l\ = An 8 Point Plane with a Unique 4 Point Line N^, P^, and Vg and Pg A Rank (k+1) Chain of k Long Lines cl(i k _ 1 U = r P = N? P = P An Affine 8 Point Plane 61 IV

5 CHAPTER I INTRODUCTION Matroids Matroid theory is, in part, an extension of affine and projective geometry. Matroids were defined in several equivalent ways by Whitney [13]. Much of our development will follow Crapo and Rota [2]. We begin by defining a matroid using a closure operator, which is used to define closed sets. Geometric lattices of closed sets are next defined. Using geometric lattices, matroid rank is defined. Independent sets in a matroid are defined using rank. Finally, we define dependent sets and circuits in a matroid. Most properties of matroids have meaning on infinite sets, but we will only be concerned with finite matroids. The notation for the set difference of two sets X and Y is X\Y. For a finite set X, denote the number of elements of X by X. For a lattice L, and elements x and y of L, x V y is the join (least upper bound) of x and y, and x A y is the meet (greatest lower bound) of x and y. A matroid is a set M together with a closure operator cl satisfying the following four axioms (White [12, p. 38]):

6 (ell) For every ACM, AC cl(a); (cl2) For every A, B C M, if A C B then cl(a) C cl(b); (cl3) For every A C M, cl(cl(a)) = cl(a); (cl4) For every A C M and a, b G M, if a cl(a U {b})\cl(a) then b cl(a U {a})\cl(a). These four axioms state that a closure operator is increasing, monotone, idempotent, and satisfies an exchange axiom. The collection F of all subsets A of M such that cl(a) = A is called the collection of closed sets, or fiats, of M. If we add the additional requirements (cl5) cl(0) = 0, and (cl6) cl({x}) = {x} for each element x of M, the resulting matroid is called a combinatorial geometry. Important motivating examples for combinatorial geometries are projective geometries of dimension n 1, PG(n l,q), representable over GF(q), where q is a prime power. Figure 1 shows the geometries PG(2,2) and PG(2,3). The flats of these geometries are the empty set, individual points, lines, and the entire geometries. Figure 1. PG(2,2) and PG(2,3)

7 Given a matroid on M, we define the lattice of flats L(M), with elements the closed sets of M, ordered by set containment. For any two flats A and B of M, define A A B to be cl(a n B) = A n B, and A V B to be cl(a U B). A hyperplane H of a matroid on M is a proper closed subset of M, such that for all x e M\H, cl(h U {x}) = M. Define the points (or atoms) of L to be those elements of L which cover cl(0). Let L be a lattice with greatest element 1 and least element 0. L is said to be a point lattice if every nonzero element x in L can be written as the join of points of L. The lattice L is said to be semimodular if, whenever x and y are in L, x and y cover x A y implies x V y covers x and y. If a point lattice L is semimodular and contains no infinite chains, then L is said to be a geometric lattice. It is proved in Crapo and Rota [2, p. 2.11] that the lattice of flats of a combinatorial geometry of finite rank is a geometric lattice, and that each geometric lattice L corresponds to a unique combinatorial geometry, whose points are the points of L, and whose flats are the elements of L. Thus any theorem about geometries of finite rank corresponds to an equivalent theorem for geometric lattices, and conversely. The following proposition gives the necessary property of a geometric lattice to define rank. For proof, see Crapo and Rota [2, p. 2.13].

8 PROPOSITION. If x and y are elements in a semimodular lattice L having no infinite chains, and x < y, then all maximal chains from x to y have the same length. In particular, we define the rank p{y) of y to be the length of a maximal chain from 0 to y. The integer p(y) equals the geometric rank (defined below) of a geometry whose lattice of flats is isomorphic to L. The rank of a point in L is one, the rank of a line is two, and the rank of a plane is three. For the geometry G with lattice of flats L, we define the rank r of each subset X of G by r(x) = In an n dimensional vector space over a field F, a set of vectors v^, Vg,..., v^ is linearly independent if a l v l + a 2 v = 0 implies that each a^ is zero. The rank of a set of vectors V in F n is defined to be the size of a maximal linearly independent subset of V. This notion of rank is used to define a matroid. The rank function r : 2 {0,1,...} is defined by r(a) = p(cl(a)), and satisfies the following three properties: For all X C M, x, y e M, (rl) r(0) = 0; (r2) r(x) < r(x U {x}) < r(x) + 1; (r3) If r(x) = r(x U {x}) = r(x U {y}), then r (X U {x,y}) = r(x).

9 Note that points of a geometry of rank n have rank 1, while hyperplanes have rank n 1. An important consequence of the exchange property (cl4) is that, if M is a matroid and K is a closed subset of M having rank n, then the collection of closed sets in M having rank n + 1 and containing K partition the points of M not in K. Ve will make use of this fact throughout Chapters 2 and 3. A combinatorial geometry on G has the property that the rank of any 2 element subset of G is 2. Thus in a geometry any single element has rank one. The independent sets of a geometry on G are those subsets I of G such that r(i) = 111. An equivalent definition of independent set can be stated in terms of the lattice of flats of G. It can be proved using the proposition above that for each closed set K there is a minimum size subset A such that cl(a) = K. This set A is independent, and A = r(k). The independence axioms for a matroid are: (11) 0 is an independent set; (12) If I is independent, then J is independent for all J C I; (13) If I and J are independent sets with I + 1 = J, then there is an element x of J such that I U {x} is independent- Subsets of M which are not independent are dependent. Define a circuit C of M to be a minimal dependent subset of M.

10 That is, C is dependent, and every proper subset of C is independent. (cl) No circuit is a proper subset of another; (c2) For every two circuits and Cg, if x e flcg, then there is a circuit C (C^ U C2)\{x}. The properties of circuits are satisfied by cycles in a graph, where the edges of a cycle are considered as elements of a circuit, and the edge set of a forest of a graph is regarded as an independent set. In a geometry, three collinear points form a circuit of rank two. Four coplanar points, no three collinear, form a circuit of rank three. A basis of a matroid on M is a maximal independent set of M. All bases of a matroid are the same size. Since the rank of a subset of M is the size of a maximal independent subset, the rank of M is the size of any basis of M. The two defining properties of the collection of bases of a matroid are: (bl) No basis is a proper subset of another basis. (b2) If and B 2 are bases of M and x e then there is a y e 1*2^1 suc h ( ^\{ x }) u {y} is a basis in M. These two axioms are satisfied by bases of a vector space, and by edge sets of spanning forests of graphs.

11 Matroid Minors For a matroid M and T C II, define the submatroid on M\T by cl M^T(X) = [cl(x)] n (M\T) for all X C M\T. It can be checked that cl^rj,satisfies the closure axioms for a matroid. The matroid on M\T is sometimes called the restriction of M to M\T. The set T is the set of elements deletedfrom M. For a geometry G, G\T is called a subgeometry of G, and inherits its geometric structure from that of G, via the closure operator. See Vhite [12, page 131] for further characterizations. A submatroid of a submatroid of M is a submatroid of M. The deletion of the set T from M may be done one element at a time, in any order, always giving the same collection of flats for M\T. For a matroid M and a subset X of M, we define the contraction M/X on the set M\X using the rank function. For each A c M\X, r j[/x(a) = r(a U X) r(x). In terms of closure, contraction of M by X can be defined as cl M^x(A) = [cl(a U X)] \X. The lattice of flats of M/X is lattice-isomorphic to the interval [cl(x),l] in the lattice of flats of M (see Velsh, [11, p. 66]). Contraction of a matroid of rank n by X will produce a matroid of rank n r(x). If x and y are elements of M, then M/{x.y} = (H/{x})/{y} and (M/{x})/{y} = (M/{y})/{x}. Thus contraction of a matroid by a subset X of M can be done one element at a time, in any order.

12 The concept of contraction in a matroid is analogous to that of contraction of an edge e in a graph T. The edges E(T) determine a matroid M(E(r)), whose independent sets are edges of forests in T. Now M(E(r))/{e} = M(E(r/e)), so that contraction in graphs is the same as contraction in matroids (White [12, Ch. 6]). The cycles in T/e are the minimal nonempty edge sets of the form C\{e}, for all cycles C in E(T). This characterization of contraction in terms of cycles in graphs carries over to circuits in a matroid. The circuits of M/X are the minimal sets of the form { C\X : C is a circuit of M }. The contraction of a matroid is a matroid, but the contraction of a geometry by a point is in general not a geometry. Contraction in a geometry is analogous to projection in a projective geometry. For example, contraction of the geometry G = PG(2,3) of Figure 1 by the point x is like the projection of G\{x} from x. There are four lines in G containing x. These lines, being coplanar, represent a 4 point line in the projection. The matroid on G/{x} will have 12 points, have rank 2, and will be partitioned into four sets of three elements, each set having rank 1. One may identify G/{x} with a 4-point line, by deleting eight of the points in G/{x}, leaving a 4-point line {a,b,c,d}. The process of identifying dependent pairs of points and removing any rank-0 elements is called simplification. The

13 simplification of M will be a geometry s(m) whose lattice of flats is the same as that of M. The rank-1 elements of L(M) correspond to the points of s(m). In this work, the contraction of a geometry by a point or line in G will be a geometric contraction. That is, the contraction and ensuing simplification are combined as one operation. When two points x and y, with cl({x}) = cl({y}), are identified, we will customarily delete whichever point is convenient, because the choice of such deletions does not affect the lattice of flats of M. The elements of G/{x} deleted in the process of simplification are said to be destroyed. Ve will write G/x instead of s(g/{x}) for geometric contraction. Throughout this work, The number of points destroyed in the contraction of a geometry G by a point p of G is given by G G/p = 1 + X( / 2), where the sum is taken over the 1 collection of all lines in G containing p. When contracting a geometry by a set X, the contraction can either be done one point at a time, with a simplification after each contraction, or with the simplification taking place after all contractions have been completed. Note that contraction by a rank--0 set is simply the deletion of that set. A matroid on N is said to be a minor of the matroid on M if N C M and the matroid on N can result from a sequence of deletions or contractions of elements of M\N. It can be proved (Velsh [11, p. 65]) that, if x and y are elements of M, then

14 10 (M\{ x })/{y} = (M/{y})\{x}. Thus, if X and Y are disjoint subsets of M, an arbitrary sequence of single element deletions of the elements of X or contractions by the elements of Y can either be represented by (M\X)/Y, or by (M/Y)\X. Uniform Matroids, Binary Matroids, and Ternary Matroids Define the uniform matroid U (m < n) on a set of n m,n v - / elements, by defining its bases to be all subsets of size m. The geometries U2 ^ and U2 g are the 4 point and 5 point lines, respectively. Many, but not all, matroids can be represented as sets of vectors with coordinates from a field F. For example, the Fano matroid F^, which is PG(2,2), can be represented as 7 distinct nonzero vectors in [GF(2)] 3. Matroids that can be represented over GF(2) are called binary matroids. A matroid whose elements may be represented as the edges of a graph (a graphic matroid) can be represented over any field. Matroids which are representable over GF(3) are called ternary matroids. The Fano matroid is not ternary, hence is not graphic. Two matroids of interest to us are the k-wheel and the k-whirl, where k > 2. The k-wheel is the graphic matroid corresponding to the k-wheel graph (See Figure 2). This graph has k edges that are "spokes" and k edges which form a

15 11 "rim", a circuit in Note that has rank k. The k" k complete graph is also the three wheel graph, which is why fg is also denoted as M(K^). The k whirl constructed from 55^ be Jetting the circuits of can be be (1) any circuit of besides the rim circuit; (2) any set of elements formed by adding a spoke of 59^ to the rim circuit of 55^. Figure 2. The k Wheel Graph. The k whirl is not binary, hence is not a graphic matroid. Figure 3 illustrates 5^, 5T^, 35^, and as geometries. The cycles of length 3 in the k-wheel graph evidently correspond to the 3-point lines in The rim cycle in the k-wheel graph corresponds to a circuit of rank k 1 in 5^., while those same points form an independent set in the construction of f 1. Ve will use the convention of only drawing in lines which have more than two points. Since any two points determine a line in a geometry, it is clear what the 2-point lines are in these figures.

16 12 r 3 r- r 4 r Figure 3 Direct Sums and Parallel Connections A proper subset A of a matroid on M is a separator of M if r(a) + r(m\a) = r(m). of M is a separator of M. M is connected if no proper subset Equivalently, M is connected if, for every pair x and y in M, there is a circuit in M containing them both. For further characterizations see Brylawsky [1] and White [12, p. 176]. The importance of wheels and whirls was shown by Tutte [10, p. 1320] (see also Velsh [11, p. 80]). Define a connected matroid on M to be S-connected if the removal of any two elements from M results in a submatroid that is connected. Define an element x of a 3 connected matroid on

17 13 M to be essential if neither M\{x} nor M/x is 3-connected. Tutte proved the following. THEOREM. A 3-connected matroid has every element essential if and only if it is a wheel or a whirl. For disjoint sets N and M and matroids on M and N, define the direct sum M N using the rank function. For all X C M U N, r jj e jj(x) = Tjj(X n M) + r^(x fln). The closed sets of M N are of the form K U L, where K is closed in M and L is closed in N. A circuit of M N is either a circuit of M or a circuit of N. Thus a direct sum is not connected. A basis in M N is the union of a basis in M with a basis in N. See White [12, p. 173] for further characterizations of direct sum. Let M be a matroid which is not connected, with separator A C M. Then the matroid on M is the direct sum of its submatroids on M\A and on A, respectively. Let M and N be two matroids with M fl N = {p}. The parallel connection P(M,N) will have closed sets being those subsets K of M U N such that both K fl M and K n N are closed in M and N, respectively. The main fact about parallel connections relevant to this work relates to contraction. In a matroid on R, if there is a point p of R such that R/p is not connected, then the matroid on R is a,parallel

18 14 connection. That is, there are sets E 1 and E 2 such that U Eg = R., E^ n Eg = {p}, and R/p = [(R\(E 2 \{p}))/p] [(IXCE^p}))/?]. Thus R = P(R\(E 2 \{p}),r\(e 1 \{p})). See White [12, p. 178]. Minor-Closed Classes of Matroids A class of geometries is a collection of geometries closed under isomorphism. A minor closed class of geometries is a collection # of geometries with the property that if G is in tf, then every geometric minor H of G is in The intersection of two minor closed classes of geometries is a minor-closed class. If {G^Gg,...} is a collection of geometries, define the excluded-minor class $a(g 1,G 2,...) of geometries to be the collection of geometries which do not contain any element of p as a minor. All excluded-minor classes are minor-closed. Every minor-closed class can be viewed as an excluded-minor class, simply by excluding the geometries which are not in it. An example of a minor-closed class is %C (q), the class of geometries not containing U 2 q+2 as a minor, where q > 1. Thus U (q) = ^(U 2^+2 ). When q is a prime power, let J?(q) be the class of geometries representable over GF(q). If H is a geometry which is representable over GF(q), then every geometric minor of H is also representable over GF(q). Every geometry H of rank n representable over GF(q) is a subgeometry of PG(n l,q). A geometry representable over

19 15 GF(q) cannot have Ug ^+2 as a minor. Therefore, (q) C U (q). Many characterizations of classes of geometries in terms of excluded minors are known. See White [12, p. 146] for a list. Size Functions and Growth Rates For a class of geometries define the (partial) size function h(n) = max{ G : G #, r(g) = n } if the maximum exists, and is undefined otherwise. Define the growth rate g of by g(n) = h(n) h(n 1) whenever h(n) is defined. There are several results concerning the size function and growth rate of classes of combinatorial geometries. Dirac [3] proved that the number of edges of a graph on n vertices having no topological K^ as a subgraph is at most 2n 3. In Kung [5], growth rate results for various excluded minor classes of binary geometries were obtained. Kung proved that the growth rate of Jf (2) fl &a(m(k g ),F 7 ) is given by g(n) =3, for n > 3, and that the growth rate of Jf (2) fl &(C^Q,M(Kg)) is 4. See [5, p. 839] for a definition of C^Q. Related to the above results are two theorems proved in Kung [6, p. 209]. The graphs Kg and K^ ^ are significant because they are the minimal nonplanar graphs.

20 16 THEOREM. The maximum growth rate of (2) fl &z(m(kg)) is at least 5 and at most 8. THEOREM. The maximum growth rate of =^(2) n &z(m(kg g)) is at least 8 and at most 10. Another line of investigation is to find maximum size estimates for classes of geometries not contained in (2). The main theorem of Chapter 2 is a refinement of a theorem of Kung. Note that is M(K 4 ). THEOREM (Kung [7, p. 233]). When q is odd, the maximum growth rate of 8a(, 3T fl % (q) is at least q and at most q + 1. Vhen q is even the maximum growth of fl It (q) is at least q and at most q + 2. The growth rate of the class &z(s^,5r 3 ) fl U (q) is proved to be precisely q in Chapter 2. Furthermore, the rank n geometries in this class having size equal to the size function h(n) are parallel connections of (q + 1) point lines. In Chapter 3, the maximum size of geometries excluding ^2,5' ^4' anc^ is found. Oxley [8, p. 70] characterized the extremal geometries of Jg (2) fl «a(, and proved that it has size function h(n) given by h(n) equals 3n - 2 if n is odd, and 3n 3 if n is even. Oxley also characterized

21 17 the extremal geometries of =^(3) fl &z( [9, p. 241], and proved that the size function h(n) equals 4n 3 if n is odd, and 4n 4 if n is even. In Chapter 3 of this work, we show that the size function of &a( n U (3) is is given by h(n) = 6n - 5, whenever n is at least 3. This is a partial result to the general problem of estimating the growth rate of H» k ) n v (q).

22 CHAPTER BIBLIOGRAPHY 1. T. H. Brylawski, A Combinatorial Model for Series- Parallel Networks, Trans. Amer. Math. Soc. 154 (1971), H. H. Crapo and G. C. Rota, On the Foundations of Combinatorial Theory: Combinatorial Geometries (Preliminary Ed.), Cambridge, Massachusetts and London, England, The M.I.T. Press, G. A. Dirac, In abstrakten Graphen voustandige 4 Graphen und ihre Unterteilungen, Math. Nachr. 22 (I960), C. Greene, Lectures on Combinatorial Geometries, Notes taken by D. Kennedy from the National Science Seminar on Combinatorial Theory, Bowdoin College, 1971, not published. 5. J. P. S. Kung, Growth Rates and Critical Exponents of Classes of Binary Combinatorial Geometries, Trans. Amer. Math. Soc. (2), 293 (1986), J. P. S. Kung, Excluding the Cycle Geometries of the Kuratowski Graphs from Binary Geometries, Proc. London Math. Soc. (3), 55 (1987), J. P. S. Kung, The Long-Line Graph of a Combinatorial Geometry. I. Excluding M(K^) and the (q+2)-point Line as Minors, Quart. J. Math. Oxford Ser. (2), 39 (1988), J. G. Oxley, The Binary Matroids with no 4 Wheel Minor. Trans. Amer. Math. Soc. (1), 301 (1987), J. G. Oxley, A Characterization of the Ternary Geometries with No M(K^)-Minor, J. Combin. Theory Ser. B, 42 (1987), V. T. Tutte, Connectivity in Matroids, Canad. J. Math. 18 (1966),

23 D. J. A. Welsh, Matroid Theory, London and New York, Academic Press, N. L. Vhite (ed.), Theory of Matroids, Cambridge University Press, Cambridge, H. Whitney, On the Abstract Properties of Linear Independence, Amer. J. Math. 57 (1935), ,

24 CHAPTER II THE GROWTH RATE OF THE CLASS OF GEOMETRIES EXCLUDING THE (q+2)-p0int LINE, THE 3-WHEEL, AND THE 3-VHIRL AS MINORS In this chapter the growth rate of the class of geometries excluding the 3-whirl r 3, 3-wheel 5^, and the (q + 2) point line U2 q + 2 i s found. The main theorem was conjectured by Kung [2, p. 234]. Denote by <U (q) the class of geometries excluding U 2 q+2 as a minor. THEOREM. For q > 2, the maximum growth rate of = U (q) n a( F" 3 ) is q. Moreover, a maximum size rank-n geometry in if has size (n - l)q + 1 and is a parallel connection of n - 1 (q + l)-point lines. A long line is a line that contains more than two points. A triangle of long lines consists of three distinct long lines and such that each pair of these three lines meets (intersects) at a point, and l ± fl l Q f l^ n Zg. It is evident that a triangle of long lines contains a 3 or a W as a subgeometry. 20

25 21 A geometry L is a chain if L is the union of m long lines l ±, l 2,... I m, n <-1 = 1 if i - j =1, and Uj n 1.1 = 0 if i j >2, where 1 < i < m, 1 < j < m. Note that L has rank at most m + 1. Throughout this chapter and the next, we will use the idea of a ring of long lines. If m = 3, a ring is a triangle of long lines. If m > 4, define L to be a ring of long lines if L is the union of m long lines..., ^ (indices modulo m), with the condition that 11, fl /. = 1 if i - j = l or if i - j = m - 1, modulo m, and (. fl ^ =0 otherwise. For a ring, the subscripts {l,2,...,m} are considered to be integers modulo m. Note that L may have rank as high as m. If L is either a chain or a ring of long lines, a point p in L such that p is the intersection of two lines I. and I. is i J called a joint of L. Note that which points of L are joints may depend on how the long lines are labelled in L, because there may be other long lines in L besides the l ±. Define the length of a chain or a ring of m long lines to be m. LEMMA 1. Let m > 4 and let R be a rank m ring of m long lines ip J 2,..., of the l^, 1 < i < m. If iis a long line in R, then I is one PROOF. Suppose that Iis a long line of R, and that I is not one of the Let J = { X[L,x 2,...,x m } be the joints of R, where ^ = l ± fl I m, and = J. n l ± _ ± if 2 < i < m. Let N be

26 22 the points of R which are not joints. Then R is the disjoint union of J and N. The subgeometry J has size m, and is an independent set, since cl(j) = R. Since Ihas at least three points and is a subset of R, /meets at least three of the long lines I L e t L be a minimum length chain in R, such that /fll = 3. That is, L is a chain, and k = 'j ^ 'j+1 U * * U By relabelling the ^ if necessary assume that L = ^ U l 2 U... U / n. Now the rank of L is n + 1, since there n + 1 joints of R in L. But cl(^ U ^ U U = cl(i U l ± U l 2 U... U contradicting the rank of L. 0 LEMMA 2. Let L be a ring of m long lines, with m > 4, of rank m. If y is on ^ and not on l m l or then L/y is a ring of m 1 long lines of rank m 1. PROOF. Write the long lines of L as { i. : 1 < i < m }. Suppose r(l) = m, and let y be a point on which is not on l l or 'm 1* Then the geometry L/y has rank m - 1. Since no other long line in L contains y besides I, only points of I m m are destroyed in constructing L/y. Hence L/y contains at least m 1 long lines. To prove that L/y is a ring, we show that any three noncollinear points a, b, and c in L U L U... U I with i z m 1 7 no more than one of {a,b,c} being on I will have rank 3 in L/y. Denote by r^y the rank function in L/y. From the m,

27 23 definition of contraction we have r L/y({ a > b ' c }) = r ({ a >b»c} U {y}) - r({y}). But for any such a, b, and c, no two of which are on I, m 7 r({a,b,c,y}) = 4. Hence r^y({a,b,c}) = 3 so that a, b, and c are noncollinear in L/y. Therefore L/y is a ring of m 1 long lines, of rank m 1. Now L/y is a ring, since ^ meets in L/y and by Lemma 1, there is no m th long line in L/y. Lemma 2 is related to the fact that if E is a k-wheel (k-whirl), the geometric contraction E/x by a point x of E which is not a joint of R will be a (k 1) wheel ((k - 1) whirl). LEMMA 3. If L is a ring of m long lines of rank m, then L contains a or a as a minor. PROOF. Suppose that for all i < m any rank i ring of i long lines has a or a W as a minor, and suppose L is a ring of m long lines rank m. Since ^ is long, let y e it' y * l n-± n l n> 7 n l ±. Now the geometry L/y has rank m - 1, and consists of long lines L, L, I i ^ in 1, where now l ± n ^ and ^ n ^ are identified as a single point fll n ± in L/y. By Lemma 2 above, L/y is a ring of m a * l n 6 lines of rank m 1, and by induction, H contains ^3 or a as a n

28 24 LEMMA 4. Suppose that 3Z U (q) is a minor-closed class of geometries, n > 3, and for each i, 2 <i<n, He 3 and r (H) = i imply H < (i l)q + 1. If G is a geometry in 3, r(g) = n, and G > (n - l)q + 2, then every point x in G is on at least two long lines in G, and the set of long lines containing x has at least q + 3 points. PROOF. Let x be a point of G. Now G/x is in 3, and by hypothesis G/x < (n - 2)q + 1. Assume G = (n - l)q + 2, since a larger value for G will not decrease the number of long lines through a point. The (n 2)q + 1 lines containing x partition the (n - l)q + 1 points of G\{x}. Let G/x = t< (n - 2)q + 1. Then t points in G\{x} determine the t lines containing x, and there are still (n - l)q t > q points to be distributed among the t lines. These (n - l)q t points cannot be on a single line, because G is in U (q). If the total number of long lines containing x is s, then the number of points on long lines containing x is at least 1 + [(( n ~~ 1)<1 + 1) t] + s, which is greater than or equal to q + 3. D Note that if the geometry G in Lemma 4 satisfies G = (n - l)q + l and every point in G is on at least two long lines, we obtain that each point of G is on long lines containing at least q + 2 points.

29 25 It now follows that for any long line Iin a geometry G satisfying the previous lemma, and a point x on I, the number of points not on Ion long lines meeting I \{x} is at least 2q. This is proved by observing that there must be at least (\l\ l)((q + 2) ( / 1)) points not on Ion long lines which meet I \{x}, which has a minimum value of 2q if \l\ is 3 or q + 1. Also, from the above remarks, it follows that if G has size (n - l)q + 1, r(g) = n, and each point in G is on at least two long lines, this lower estimate becomes 2q - 4 for q > 2. Geometries Having Maximum Size The maximal rank 2 geometry in # = (q) fl F" 3 ) is the (q + 1) point line. Vhen the rank n is greater than 2, there exist geometries of size (n - l)q + 1 in which are parallel connections consisting of n - 1 long lines each of length q + 1. For rank 3, this consists of two (q + 1) point lines which meet. The following proposition shows that these geometries are maximal. PROPOSITION 5. Suppose G is a rank-n geometry in U (q) of size at least (n - l)q + 2, where n > 3, and q > 2. Then G contains a or a as a minor. PROOF. The proof is obvious if n = 2. If n = 3, let a and b be two points of G. Suppose that G does not contain a O

30 26 or a F 3 as a minor. Partition the 2q points of G\{a,b} into those which are on the line cl({a,b}) and those which are not. The former has size at most q 1, and the latter has size at most q. Thus G has at most 2q + 1 points, contradicting the assumption that G has at least 2q + 2 points. Suppose n > 4 and q > 2. Let I^ be a long line in G, with two points x^, x 2 on l^. Now every point in G is on at least two long lines, by Lemma 1, so let ^ be another long line containing x 2. By the remarks preceding this theorem, there are at least 2q points not on 1 2 on long lines which meet ^\{x 2 }. If any such long line Ialso meets l ±, then l ±, ^2> and Iform a triangle of long lines. Otherwise, assume that none of these lines meet l^. Call the collection of long lines which meet ^\{x 2 } K 2 the union of the long lines in Jfy, and T 2 the set l ± U ^ U K 2. Now T 2 > 2q + 5, so that if the rank of T 2 is 3 it follows that G contains a or a T as a subgeometry. If the rank of T 2 is at least 4, then there is a long line meeting / 2 \{x 2 } such that ^ U Z 2 U / 3 has rank 4. Let x^ = fl. Suppose that for i < n the chain L. = L U L U... U Z has been constructed in G, with rank i + 1, where e x 2 = *1 n *2' x 3 = *2 n *3» * > x i = n ^» with K. constructed for each j < i, T. = L U L U... U /. U K.. L z J J 3 Further suppose that for j < i, rank(t.) = j + 1 implies G J o contains a or a W as a minor. There are at least 2q

31 27 points not on I^ which are on long lines meeting Call this collection of long lines and let K i = U{ I: I }. Suppose that a line in JS. meets a line J, 1 < j < i. Let m = max{ j : 1 < j < i and a line in meets I }. J Then a line Imeeting (. and l m, together with, form a ring of i - m + 2 long lines of rank i m + 2. If no long line meeting /^\{x^} meets any of the long lines in L.^, then the subgeometry """i = *1 U ^2 U ' ' * U IIKj forms a rank (i + 1) subgeometry of G. The contraction T^/x^ is a rank i geometry with points Z 2 U J 3 U... U I. U I.. Now U jg U... U is a chain of long lines in T^/x^, the rank of T i /x 1 is i, and by the induction hypothesis G contains a ^ or a W ^ as a minor. Since i cannot exceed n, this proves the proposition. D Ve now prove that any rank n geometry G of size (n - l)q + 1 in ^ is a parallel connection of (q + l)-point lines. For q = 2, see [1]. Assume q > 3. It is obviously true for n = 3 that G is the parallel connection of two (q + 1) point lines. Suppose that for n = 3, 4,..., k that Kg K = (n l)q + 1, and r(k) = n imply K is a parallel connection of n - 1 (q + l)-point lines, and let G be a rank (k + 1) element of with G = kq + 1. If each point of G is on at least two long lines, then from the

32 28 remarks preceding Proposition 5 we have that for any long line Iand a point x on Ithere are at least 2q 4 points not on Iwhich are on long lines meeting I \{x}, and we may use the proof of Proposition 5 to show that G contains a or a T as a minor, a contradiction of the definition of Since each point of G is on at least one long line, there is a point p of G such that p is on only one long line. Call this line I. Now Ihas length q + 1. It follows from the induction hypothesis that there is a second point c on I, such that c is on a long line besides I. Now G/p is a parallel connection of k - 1 (q + l)-point lines by hypothesis. Thus every plane P containing Iin G such that Py > 2 is a parallel connection of Iwith another long line of size q + 1. The geometry consisting of Iand the long lines meeting Iis a geometrically closed subset of G. If it is all of G we are done. Otherwise, consider a plane P containing p such that P meets a long line V which meets I, and P/p = q + 1. If P is not a (q + 1) point line together with p, then I U I' U P will have a ^ or a r 3 as a minor, a contradiction. Similar arguments can be used to show that G is k 1 parallel connections of k long lines, each of size q + 1. This completes the proof of the theorem.

33 CHAPTEft BIBLIOGRAPHY 1. T. H. Brylawski, A Combinatorial Model for Series- Parallel Networks, Trans. Amer. Math. Soc. 154 (1971), J. P. S. Kung, The Long Line Graph of a Combinatorial Geometry. I. Excluding M(K^) and the (q+2)-point Line as Minors, Quart. J. Math. Oxford Ser. (2) 39 (1988),

34 CHAPTER III THE GROWTH RATE OF THE CLASS OF GEOMETRIES EXCLUDING THE 5-POINT LINE, THE 4 VHEEL AND THE 4 WHIRL AS MINORS The main result of this chapter is THEOREM. Let h(n) be the size function of the class # of geometries excluding ^ or as minors. Then h(n) =6n 5, for n > 3. From this theorem it follows that the growth rate of is 6 when n is greater than 3. Examples of rank n geometries whose size is h(n) include the ternary geometries which are parallel connections of PG(2,3) and V lg (see below). Since 2^(3) includes the ternary geometries, these geometries are also extremal rank-n geometries in Jf (3) n T^). The main result is a special case of the general unsolved problem of estimating the maximum growth rate of the class of geometries excluding the (q + 2)-point line, the k-wheel and k whirl as minors. Define V ig to be the rank-4 ternary geometry consisting of two 13-point planes, with three points removed from their 30

35 31 line of intersection. Note that V^g is 3 connected. The ternary geometries without or 2T 4 as a minor, having rank n, whose size is h(n), include PG(2,3), V ig, and geometries which are parallel connections of PG(2,3) or V ig. For rank 4, V^g has size h(4). For rank 5, the parallel connection of two copies of PG(2,3) at a point p has size 25 and does not contain or 5T 4 as a minor. For n > 6, there are rank n parallel connections of copies of PG(2,3) and V 19 which do not have ^ or 5T 4 as a minor, and which have size h(n). Figure 4. V 1Q. It is known that the class ^(3) of ternary geometries has size function (3 n - 1). This is also true of U (3) [1, p. 86]. The extremal geometries of It (3) are PG(n 1,3). Define two points x and y in a plane P to be tied in P if there is a point z in P, not collinear with x and y, such that cl({x,z}) and cl({y,z}) are both long lines. All geometries in this chapter are assumed to be in U (3).

36 32 LEMMA 1. Let P be a rank-3 geometry in U (3), and x and y be points in P. If there are at least five points in P not on the line containing x and y, then x and y are tied in P. PROOF- Ve may as well assume that there are exactly five points in P\cl({x,y}). First, suppose that there is a line Icontaining either x or y, but not both, such that Ihas four points. Assume without loss of generality that I contains x. Since there are at least two points of P not in cl({x,y}) U I, and P/y = 4, y is collinear with one of these points and a point z of I \{x}, and we have the required z. If there is no such 4 point line l f then x and y are each in distinct 3-point lines and l y, respectively, with x not in I and y not in I. If I and I meet, we are y x x y done, otherwise we use that there is a fifth point z which is collinear with y and a point of I. This concludes the A proof. g If P is a plane of size 9 or larger, then for any two points x and y of P, x and y are tied in P, by Lemma 1. LEMMA 2. Let!be a long line in a plane P, such that P\i = 4. Then at most one pair of points x and y in /is not tied in P.

37 33 PROOF. Suppose that Icontains two points x and y that are not tied in P. Since each of x and y is on a long line in P besides I, x is on a line k^ = {x,x^,x 2 } in P, y is on a line k 2 = {y,y 1,y 2 } in p > such that the tw0 lines do not meet. Let z 1 be a third point of I. Now z^ is on a line containing either or x 2, and either y^^ or y 2, since P/z^ < 4. Therefore and x are tied in P, as are z^ and y. If z 2 is a fourth point of I, then z 2 and x are tied in P, and z 2 and y are tied in P. Ve use the fact that x^^ is collinear with each point of & 2 \{y} and either z 1 or z 2 to conclude that z^ and z 2 are tied in P. A rank 4 ring of length 4 contains a rank 4 ring R such that each long line of R has only three points. If the four points of R which are not joints of R are independent, then R is 5T 4. Otherwise R is Thus a rank-^1 ring of length 4 contains either a or a f 4 as a subgeometry. The following lemma generalizes this. LEMMA 3. Let R be a ring of m long lines of rank m with m > 4. Then R contains a rank 4 ring of length 4 as a minor. PROOF. The lemma is obvious for for m = 4. Suppose that the theorem is true for m > 4, and R is a ring of m + 1 long lines, having rank m + 1. Choose an arbitrary point y of R

38 34 which is not a joint. By the definition of a ring, y is only on one long line in R, so R/y is a ring of m long lines, and has rank m, by Lemma 2 of Chapter 2. Let D be a chain of long lines I ^» ^ - 3, r(d) = k + 1, having joints x^ = ^ (2 < i < k)> with x^, points on ^ (respectively) which are not joints of D. Let N be a geometry containing D, such that there are points a, b, c, d in N\D, with the property that { x k+l' a ' b }' {x^^jcjd}, and {x^,b,d} are distinct lines (Figure 5). N Figure 5 LEMMA 4. The geometry N contains a ^ or a as a minor. PROOF. By restricting to a subgeometry of N if necessary, we may assume that N contains no 4 point lines. Ve first prove the lemma for k = 3. Label the points of N as in Figure 6. There are possibly other long lines in N besides those represented in Figure 6, according to the cases which follow. First, if N has rank 5, then we are done by Lemma 1. So suppose that r(n) = 4. The points x^, x^, and

39 35 b determine a plane T. If T contains Xg and yg, then T is an 8 point plane. Now, by Lemma 1, x^ and x^ are tied in T, so that x^, Xg, Xg, and x^ are joints of a rank 4 ring of long lines in N. If T contains x^ and y^, the argument is the same. So assume that T is simply the set {x^,a,b,c,d,x^}. Consider the planes in N containing cl({xg,x^}). There are at most four. Let p! = c1 (( x 2» x 3» x 4})» P 2 = cl ({ x i' x 3' x 4})» Pg = cl({a,xg,x^}), and = cl({c,xg,x^}). From the construction of N and our assumption about T, these four planes are distinct. Therefore, y^ is in one of these. From the construction of D, y^ is not in P^ or Pg. Suppose that y^ is in P^. Then y^ is on a long line in P^, since P^ is a ^ or a N l i yi Figure 6. The geometry N when k = 4 has a ^ or a as a minor. The plane T = cl({x^,x^,b}). There are now four cases to consider. If y^ is collinear with Xg and d, then Xg, x^, b, and d form the joints of a rank 4 ring of four long lines, and we have the lemma. If y^ is collinear with yg and c, then Xg, Xg, y^,

40 36 and y 3 form the four joints of a rank 4 ring of four long lines. If is collinear with y 3 and d, then y 3, x^, b,and d are the joints of a rank 4 ring of four long lines. Finally, if y^ is collinear with x 3 and c, then x^, y^, c, and d are the joints of a rank 4 ring of four long lines. If y^ is in Pg, the proof is similar. This concludes the proof when k = 3. Suppose k > 4, and for j < k, the lemma is true. If the rank of N is k + 2, we are done by Lemma 3, so assume that r(n) = k + 1. Ve divide the argument into three cases. First, suppose that x-^.and y^ are in the plane T determined by a, b, c, and d. Then, in the geometry N/y k, the subgeometry D/y^ is a chain of k 1 long lines of rank k, while P/y^ is a long line containing x^ and x^. Thus we have a ring of k long lines, of rank k, which we know has a f 4 or a F 4 as a minor by Lemma 3. The argument is similar if T contains y^ and Xg. Otherwise, T just consists of x^, a, b, c, d, and In this case, we may contract N by y^. The resulting geometry satisfies the induction hypothesis for k 1. o LEMMA 5. Let C be a chain of k > 5 long lines of rank k + 1, and q a point not in C U cl({xj c _^JX^JX^^}), such that x^ is in cl({xj c _ 1, x k>xk +1,q}). Then C U {q} contains a or a as a minor.

41 37 PROOF. First, if x^, q and a point of either cl({xj c _^) or cl({xj c,xj c+^}) are collinear, we are done by Lemma 3. Otherwise, if x^ either lies in the plane cl({xj c _^jxj^q}) or in cl({xj c,xj c+^,q}), the geometry (C U {q})/q will either contain a ring of k long lines of rank k or a ring of k 1 long lines of rank k 1. If the first two conditions do not hold, then the geometry (C U {q})/q will have x^ in the plane determined by ^, x^, and x^^, but not in the lines *k_l = or = cl ({ x k' x k+l}) * If either of these two lines has four points, then there will be a rank k ring of long lines in (C U {q})/q having x^, x 2,, Xj^, and a point of either or ^ as joints. So assume that these two lines have only three points each. Call the third points of and ^ y^^ an d respectively. If x^ is collinear with Xj^ and a point p of ^ then after contracting (C U {q})/q by p, we obtain a ring of k 2 long lines of rank k 2. Otherwise, x^ is collinear with y-^ ^ and either y^ or x^^, in which case (C U {q})/q contains a ring of k long lines of rank k. If, in Lemma 5, x^ is not in cl({x k _ 1,x^,x^+1,q}), for a given point q not in C, then in (C U {q})/^ the plane cl({xk_^jxj^xj^}) will be the same as in C U {q}. This fact will be useful in the main proof.

42 38 LEMMA 6. Let C be a chain of k long lines of rank k + 1, k > 3, and Q a set of points not in C U c l ^ X j ^ > on long lines containing either x k + i or >v Ifx i or yi is collinear with two points of Q, then C U Q contains a 3^ or a as a minor. PROOF. Without loss of generality we may assume that x^ is collinear with points and qg of Q. Also, we may as well assume that there is no long line in C U Q containing x^ and either or y^, for then we would be done by Lemma 3. Hence, there are points p^ and pg in Q, such that = cl({q^,p^}) and = cl^qgjpg}) each contains either x^^ or y^. Again, without loss of generality assume that contains y^. Ve then have two cases, either that rag contains y^ or In the first case, we are done, by Lemma 4. In the second, first assume that m^ and do not intersect. If cl({q^,q2}) and cl({xj c,xj c+^}) are skew, we are done. Otherwise, we may contract by if k > 4 to obtain an appropriate ring of long lines. If k = 3 and q^, qg, x k, and are coplanar in a plane P, then by Lemma 1 there is a point z in P such that x^, Xg, x,^, and z form the joints of a ring of four long lines of rank four. If ra^ contains y^, m^ contains an d and do intersect, their point of intersection will not be q^ or qg, since by hypothesis there is no long line in C U Q containing either x 1 or y^, and either XJ c+ -J_ or y^. By relabelling if

43 39 necessary, assume that = p 2 - The points x^, q^, q 2, x k+i, p 1? y k, and x k l are coplanar, and form a P 7 (see Figure 9). Thus ^ and x^.are collinear with p^. Thus there is a ring of k long lines of rank k, with joints x^ x 2,..., x^, completing the proof. LEMMA 7. Let C be a chain of k > 4 long lines, of rank k + 1, and Q defined as in Lemma 6, such that either x^ or y^ is collinear with a point of C and a point of Q. Then C U Q contains a or a as a minor. PEOOF. Ve may assume that x^ is collinear with a point z of C and a point in Q, and that there is a second point q 2 in Q such that q 1, q 2, and x k+1 are collinear, by relabelling C if necessary. If we label the long lines of C by I^ = cl({x^,x^+1 }), 1 < i < k, then z is either x^, where 2 < i < k + 1, or z is a point y^ in C which is not a joint, where 1 < i < k. If z is x^ then x^, x^,..., and q^ form the k joints of a ring of k long lines of rank k in C U Q. If z is Xj for i > 3 then x^, x 2,..., x^ are the i joints of a ring M of i long lines of rank i, with the point q^ in cl({xj,xj}) not a joint of M. If z is y^, then similar arguments produce the desired rings.. Finally, if z is x 2 or y^, we contradict the assumption that Q n C is empty. Therefore, by Lemma 3, Lemma 7 is proved.

44 40 LEMMA 8. Let C be a chain of m long lines of rank m + 1 with the usual labelling, and Ia 4 point line meeting C at a point of that is not a joint of C, such that Idoes not depend on cl({x m _ 1 > x m,x m+1 }), but such that I C cl(c)- Then C U 1contains a or a as a minor. PROOF. Suppose that m = 3. Label the points of I by a^, a 2, a^, and a^, so that a^ is x^. If Iand are coplanar, then these two lines meet, and we are done. Otherwise, Iand are skew. Consider the four planes in C U I containing I. Label these planes P ]L = cl(j U {x 3 }), P 2 = cl(z U {y 2 }), P 3 = cl(z U {x 2 }), and P 4 = cl(z U {y^}) (see Figure 7). One of these four planes contains x^. Clearly P^, Pg, and P 4 do not contain x^ since C is of rank 4, and Iis skew with l ±. Therefore, x^ is in the plane P 2, and is collinear with y 2 and a point of I. This point is not a^, because C has rank 4. Since the other three points of I were labelled arbitrarily, assume that x^ is collinear with y 2 and a.^. Therefore the points a^, a^, x^, and y 2 are the four joints of a ring of long lines of rank 4. Suppose that the lemma is true for 3 < j < m, and C and Isatisfy the hypotheses of the lemma, with r(c) = m + 1. If I meets some point of C, then C U I has a ring of four long lines of rank 4 as a minor by Lemma 3. Otherwise, we have that Iis skew with each of the lines 1 2,..., Let P = cl(z m U I). If Xj P, then x^ is collinear with a

45 41 point of Iand x m, and we obtain a ring of i - 1 long lines of rank m 1. If x^ is not in P, but is collinear with a point of C and a point of Ithen C U Icontains a ring of four long lines of rank 4 as a minor, by Lemma 7. If x^ is not in P, and is not collinear with a point of Iand a point of C, then x^ also is not in V = cl(^m_^ U / U /). (This follows from examination of the planes in V containing i> so that x^ would be collinear with a point of Iand a point of C, a contradiction.) If we contract C U Iby x^, the resulting geometry (C U Q/ X i satisfies the induction hypothesis (relabelling the points of C). *1 x. 3 t, y 3 Figure 7. C U Ihas rank 4, and f =4. LEMMA 9. If C is a chain of k long lines having rank k + 1, k is at least 3, C contains two skew 4-point lines, Iis a long line which meets C at a point of i^ which is not a joint of C, Idoes not depend on ^ U fk 1» an<^ ^depends on C, then C U Icontains a or a ST 4 as a minor.

46 42 PROOF. If k = 3, the two skew 4 point lines of C are I^ and Zg. The proof is identical to the case k = 3 in Lemma 8. Suppose that for k = 3, 4,..., m the lemma is true, and r(c) = m + 1. Let and I. be two skew 4 point lines of C, %J i < j 1. If Zmeets any other line in C, we are done by Lemma 3 and Lemma 7. If there is a point p in L = U ^ U... U such that p cl(z U U l^), then by Lemma 5 C U 1 has a or a as a minor. Otherwise, contract C U I by any point p in L\(Z^ U I.) which is not a joint, and use induction. Proof of the Main Theorem For each n > 3 there is a geometry H in = 1C (q) fl 8a( whose size is 6n 5. Thus the size function h(n) satisfies h(n) > 6n 5. For the proof of the main theorem, it will be shown that a geometry G in «(i) whose size is at least 6n 4 will contain a or a as a minor. Assuming that G = 6n 4 is sufficient, since a larger geometry would contain a subgeometry with 6n 4 points. The proof is by induction on the rank n. Cases 1 and 2 below are ad hoc arguments. The proof for n > 6 has two main cases, corresponding to whether or not G contains any 4 point lines.

47 43 Case 1: The Rank 4 Case Suppose that G is in %C (3), r(g) = 4, and G > 20. We will prove that G contains a F 4 or a as a minor. Ve may as well assume that G = 20, since we may achieve this by a restriction. First observe that G/x < 13 for any x in G, since the extremal geometry having rank 3 in It (3) is PG(2,3). It follows that for each point x in G, at least six points (not including x) are destroyed upon contraction by x. Also, each line Iof G is on at most four planes in G. Since at most 12 points are destroyed in a plane when contracting by a line of that plane, each line Iin G is on at least two planes and P 2 such that P^V > 2. Let P be the largest rank 3 subgeometry of G. If IP > 10, then for any point x not in P, there are two long lines containing x which meet P. By Lemma 1, this is sufficient to prove that G contains a or a F 4 as a subgeometry. Suppose that P < 9. Let Ibe a long line in P. By the fact that the planes of G containing Ipartition the points of G not in /, it is clear that P > 8, from the Pigeonhole Principle. Whether P is 8 or 9, and whether l is 3 or 4, there is a second plane P' containing I, such that P'V > 4. By Lemma 2, P U P' contains a ^ or a T 4 ".

TUTTE POLYNOMIALS OF q-cones

TUTTE POLYNOMIALS OF q-cones TUTTE POLYNOMIALS OF q-cones JOSEPH E. BONIN AND HONGXUN QIN ABSTRACT. We derive a formula for the Tutte polynomial t(g ; x, y) of a q-cone G of a GF (q)-representable geometry G in terms of t(g; x, y).

More information

ON SIZE, CIRCUMFERENCE AND CIRCUIT REMOVAL IN 3 CONNECTED MATROIDS

ON SIZE, CIRCUMFERENCE AND CIRCUIT REMOVAL IN 3 CONNECTED MATROIDS ON SIZE, CIRCUMFERENCE AND CIRCUIT REMOVAL IN 3 CONNECTED MATROIDS MANOEL LEMOS AND JAMES OXLEY Abstract. This paper proves several extremal results for 3-connected matroids. In particular, it is shown

More information

THE NUMBER OF POINTS IN A COMBINATORIAL GEOMETRY WITH NO 8-POINT-LINE MINORS

THE NUMBER OF POINTS IN A COMBINATORIAL GEOMETRY WITH NO 8-POINT-LINE MINORS THE NUMBER OF POINTS IN A COMBINATORIAL GEOMETRY WITH NO 8-POINT-LINE MINORS JOSEPH E. BONIN AND JOSEPH P. S. KUNG ABSTRACT. We show that when n is greater than 3, the number of points in a combinatorial

More information

TOWARDS A SPLITTER THEOREM FOR INTERNALLY 4-CONNECTED BINARY MATROIDS III

TOWARDS A SPLITTER THEOREM FOR INTERNALLY 4-CONNECTED BINARY MATROIDS III TOWARDS A SPLITTER THEOREM FOR INTERNALLY 4-CONNECTED BINARY MATROIDS III CAROLYN CHUN, DILLON MAYHEW, AND JAMES OXLEY Abstract. This paper proves a preliminary step towards a splitter theorem for internally

More information

MATROID PACKING AND COVERING WITH CIRCUITS THROUGH AN ELEMENT

MATROID PACKING AND COVERING WITH CIRCUITS THROUGH AN ELEMENT MATROID PACKING AND COVERING WITH CIRCUITS THROUGH AN ELEMENT MANOEL LEMOS AND JAMES OXLEY Abstract. In 1981, Seymour proved a conjecture of Welsh that, in a connected matroid M, the sum of the maximum

More information

HOW IS A CHORDAL GRAPH LIKE A SUPERSOLVABLE BINARY MATROID?

HOW IS A CHORDAL GRAPH LIKE A SUPERSOLVABLE BINARY MATROID? HOW IS A CHORDAL GRAPH LIKE A SUPERSOLVABLE BINARY MATROID? RAUL CORDOVIL, DAVID FORGE AND SULAMITA KLEIN To the memory of Claude Berge Abstract. Let G be a finite simple graph. From the pioneering work

More information

Characterizing binary matroids with no P 9 -minor

Characterizing binary matroids with no P 9 -minor 1 2 Characterizing binary matroids with no P 9 -minor Guoli Ding 1 and Haidong Wu 2 1. Department of Mathematics, Louisiana State University, Baton Rouge, Louisiana, USA Email: ding@math.lsu.edu 2. Department

More information

ON CONTRACTING HYPERPLANE ELEMENTS FROM A 3-CONNECTED MATROID

ON CONTRACTING HYPERPLANE ELEMENTS FROM A 3-CONNECTED MATROID ON CONTRACTING HYPERPLANE ELEMENTS FROM A 3-CONNECTED MATROID RHIANNON HALL Abstract. Let K 3,n, n 3, be the simple graph obtained from K 3,n by adding three edges to a vertex part of size three. We prove

More information

A MATROID EXTENSION RESULT

A MATROID EXTENSION RESULT A MATROID EXTENSION RESULT JAMES OXLEY Abstract. Adding elements to matroids can be fraught with difficulty. In the Vámos matroid V 8, there are four independent sets X 1, X 2, X 3, and X 4 such that (X

More information

TOWARDS A SPLITTER THEOREM FOR INTERNALLY 4-CONNECTED BINARY MATROIDS IV

TOWARDS A SPLITTER THEOREM FOR INTERNALLY 4-CONNECTED BINARY MATROIDS IV TOWARDS A SPLITTER THEOREM FOR INTERNALLY 4-CONNECTED BINARY MATROIDS IV CAROLYN CHUN, DILLON MAYHEW, AND JAMES OXLEY Abstract. In our quest to find a splitter theorem for internally 4-connected binary

More information

arxiv: v1 [math.co] 3 Aug 2009

arxiv: v1 [math.co] 3 Aug 2009 GRAPHS WHOSE FLOW POLYNOMIALS HAVE ONLY INTEGRAL ROOTS arxiv:0908.0181v1 [math.co] 3 Aug 009 JOSEPH P.S. KUNG AND GORDON F. ROYLE Abstract. We show if the flow polynomial of a bridgeless graph G has only

More information

Relaxations of GF(4)-representable matroids

Relaxations of GF(4)-representable matroids Relaxations of GF(4)-representable matroids Ben Clark James Oxley Stefan H.M. van Zwam Department of Mathematics Louisiana State University Baton Rouge LA United States clarkbenj@myvuw.ac.nz oxley@math.lsu.edu

More information

THE STRUCTURE OF 3-CONNECTED MATROIDS OF PATH WIDTH THREE

THE STRUCTURE OF 3-CONNECTED MATROIDS OF PATH WIDTH THREE THE STRUCTURE OF 3-CONNECTED MATROIDS OF PATH WIDTH THREE RHIANNON HALL, JAMES OXLEY, AND CHARLES SEMPLE Abstract. A 3-connected matroid M is sequential or has path width 3 if its ground set E(M) has a

More information

Cyclic Flats, Sticky Matroids, and Intertwines

Cyclic Flats, Sticky Matroids, and Intertwines Cyclic Flats, Sticky Matroids, and Intertwines Joseph E. Bonin The George Washington University Part I Essential Matroid Background and Cyclic Flats Matroids via Flats There are many equivalent formulations

More information

TOWARDS A SPLITTER THEOREM FOR INTERNALLY 4-CONNECTED BINARY MATROIDS VII

TOWARDS A SPLITTER THEOREM FOR INTERNALLY 4-CONNECTED BINARY MATROIDS VII TOWARDS A SPLITTER THEOREM FOR INTERNALLY 4-CONNECTED BINARY MATROIDS VII CAROLYN CHUN AND JAMES OXLEY Abstract. Let M be a 3-connected binary matroid; M is internally 4- connected if one side of every

More information

An Introduction to Transversal Matroids

An Introduction to Transversal Matroids An Introduction to Transversal Matroids Joseph E Bonin The George Washington University These slides and an accompanying expository paper (in essence, notes for this talk, and more) are available at http://homegwuedu/

More information

MATROIDS DENSER THAN A PROJECTIVE GEOMETRY

MATROIDS DENSER THAN A PROJECTIVE GEOMETRY MATROIDS DENSER THAN A PROJECTIVE GEOMETRY PETER NELSON Abstract. The growth-rate function for a minor-closed class M of matroids is the function h where, for each non-negative integer r, h(r) is the maximum

More information

Determining a Binary Matroid from its Small Circuits

Determining a Binary Matroid from its Small Circuits Determining a Binary Matroid from its Small Circuits James Oxley Department of Mathematics Louisiana State University Louisiana, USA oxley@math.lsu.edu Charles Semple School of Mathematics and Statistics

More information

TUTTE POLYNOMIALS OF GENERALIZED PARALLEL CONNECTIONS

TUTTE POLYNOMIALS OF GENERALIZED PARALLEL CONNECTIONS TUTTE POLYNOMIALS OF GENERALIZED PARALLEL CONNECTIONS JOSEPH E. BONIN AND ANNA DE MIER ABSTRACT. We use weighted characteristic polynomials to compute Tutte polynomials of generalized parallel connections

More information

JAMES OXLEY, CHARLES SEMPLE, AND GEOFF WHITTLE

JAMES OXLEY, CHARLES SEMPLE, AND GEOFF WHITTLE WILD TRIANGLES IN 3-CONNECTED MATROIDS JAMES OXLEY, CHARLES SEMPLE, AND GEOFF WHITTLE Abstract. Let {a, b, c} be a triangle in a 3-connected matroid M. In this paper, we describe the structure of M relative

More information

Modularity and Structure in Matroids

Modularity and Structure in Matroids Modularity and Structure in Matroids by Rohan Kapadia A thesis presented to the University of Waterloo in fulfilment of the thesis requirement for the degree of Doctor of Philosophy in Combinatorics and

More information

THE BINARY MATROIDS WITH NO 4-WHEEL MINOR

THE BINARY MATROIDS WITH NO 4-WHEEL MINOR TRANSACTIONS OF THE AMERICAN MATHEMATICAL SOCIETY Volume 301. Number I. May 1987 THE BINARY MATROIDS WITH NO 4-WHEEL MINOR JAMES G. OXLEY ABSTRACT. The cycle matroids of wheels are the fundamental building

More information

Semimatroids and their Tutte polynomials

Semimatroids and their Tutte polynomials Semimatroids and their Tutte polynomials Federico Ardila Abstract We define and study semimatroids, a class of objects which abstracts the dependence properties of an affine hyperplane arrangement. We

More information

An Introduction of Tutte Polynomial

An Introduction of Tutte Polynomial An Introduction of Tutte Polynomial Bo Lin December 12, 2013 Abstract Tutte polynomial, defined for matroids and graphs, has the important property that any multiplicative graph invariant with a deletion

More information

Matroids/1. I and I 2 ,I 2 > I 1

Matroids/1. I and I 2 ,I 2 > I 1 Matroids 1 Definition A matroid is an abstraction of the notion of linear independence in a vector space. See Oxley [6], Welsh [7] for further information about matroids. A matroid is a pair (E,I ), where

More information

Chordal Graphs, Interval Graphs, and wqo

Chordal Graphs, Interval Graphs, and wqo Chordal Graphs, Interval Graphs, and wqo Guoli Ding DEPARTMENT OF MATHEMATICS LOUISIANA STATE UNIVERSITY BATON ROUGE, LA 70803-4918 E-mail: ding@math.lsu.edu Received July 29, 1997 Abstract: Let be the

More information

A Decomposition Theorem for Binary Matroids with no Prism Minor

A Decomposition Theorem for Binary Matroids with no Prism Minor DOI 10.1007/s00373-013-1352-6 ORIGINAL PAPER A Decomposition Theorem for Binary Matroids with no Prism Minor S. R. Kingan Manoel Lemos Received: 21 March 2012 / Revised: 13 January 2013 Springer Japan

More information

PERFECT BINARY MATROIDS

PERFECT BINARY MATROIDS DEPARTMENT OF MATHEMATICS TECHNICAL REPORT PERFECT BINARY MATROIDS Allan Mills August 1999 No. 1999-8 TENNESSEE TECHNOLOGICAL UNIVERSITY Cookeville, TN 38505 PERFECT BINARY MATROIDS ALLAN D. MILLS Abstract.

More information

Diskrete Mathematik und Optimierung

Diskrete Mathematik und Optimierung Diskrete Mathematik und Optimierung Winfried Hochstättler Michael Wilhelmi: Sticky matroids and Kantor s Conjecture Technical Report feu-dmo044.17 Contact: {Winfried.Hochstaettler, Michael.Wilhelmi}@fernuni-hagen.de

More information

PROJECTIVE GEOMETRIES IN EXPONENTIALLY DENSE MATROIDS. II

PROJECTIVE GEOMETRIES IN EXPONENTIALLY DENSE MATROIDS. II PROJECTIVE GEOMETRIES IN EXPONENTIALLY DENSE MATROIDS. II PETER NELSON Abstract. We show for each positive integer a that, if M is a minor-closed class of matroids not containing all rank-(a + 1) uniform

More information

THE LARGEST INTERSECTION LATTICE OF A CHRISTOS A. ATHANASIADIS. Abstract. We prove a conjecture of Bayer and Brandt [J. Alg. Combin.

THE LARGEST INTERSECTION LATTICE OF A CHRISTOS A. ATHANASIADIS. Abstract. We prove a conjecture of Bayer and Brandt [J. Alg. Combin. THE LARGEST INTERSECTION LATTICE OF A DISCRIMINANTAL ARRANGEMENT CHRISTOS A. ATHANASIADIS Abstract. We prove a conjecture of Bayer and Brandt [J. Alg. Combin. 6 (1997), 229{246] about the \largest" intersection

More information

On the intersection of infinite matroids

On the intersection of infinite matroids On the intersection of infinite matroids Elad Aigner-Horev Johannes Carmesin Jan-Oliver Fröhlich University of Hamburg 9 July 2012 Abstract We show that the infinite matroid intersection conjecture of

More information

Diskrete Mathematik und Optimierung

Diskrete Mathematik und Optimierung Diskrete Mathematik und Optimierung Winfried Hochstättler, Robert Nickel: Joins of Oriented Matroids Technical Report feu-dmo009.07 Contact: winfried.hochstaettler@fernuni-hagen.de robert.nickel@fernuni-hagen.de

More information

Non-Recursively Constructible Recursive Families of Graphs

Non-Recursively Constructible Recursive Families of Graphs Non-Recursively Constructible Recursive Families of Graphs Colleen Bouey Department of Mathematics Loyola Marymount College Los Angeles, CA 90045, USA cbouey@lion.lmu.edu Aaron Ostrander Dept of Math and

More information

Rota s Conjecture. Jim Geelen, Bert Gerards, and Geoff Whittle

Rota s Conjecture. Jim Geelen, Bert Gerards, and Geoff Whittle Rota s Conjecture Jim Geelen, Bert Gerards, and Geoff Whittle Rota s Conjecture For each finite field field F, there are at most a finite number of excluded minors for F-representability. Ingredients of

More information

On the interplay between graphs and matroids

On the interplay between graphs and matroids On the interplay between graphs and matroids James Oxley Abstract If a theorem about graphs can be expressed in terms of edges and circuits only it probably exemplifies a more general theorem about matroids.

More information

MATROID THEORY, Second edition

MATROID THEORY, Second edition MATROID THEORY, Second edition James Oxley Oxford University Press, New York, 2011 Errata and Update on Conjectures, Problems, and References Latest update: August 16, 2016 The reader is encouraged to

More information

Excluding a Long Double Path Minor

Excluding a Long Double Path Minor journal of combinatorial theory, Series B 66, 1123 (1996) article no. 0002 Excluding a Long Double Path Minor Guoli Ding Department of Mathematics, Louisiana State University, Baton Rouge, Louisiana 70803

More information

TOWARDS A SPLITTER THEOREM FOR INTERNALLY 4-CONNECTED BINARY MATROIDS II

TOWARDS A SPLITTER THEOREM FOR INTERNALLY 4-CONNECTED BINARY MATROIDS II TOWARDS A SPLITTER THEOREM FOR INTERNALLY 4-CONNECTED BINARY MATROIDS II CAROLYN CHUN, DILLON MAYHEW, AND JAMES OXLEY Abstract. Let M and N be internally 4-connected binary matroids such that M has a proper

More information

The cocycle lattice of binary matroids

The cocycle lattice of binary matroids Published in: Europ. J. Comb. 14 (1993), 241 250. The cocycle lattice of binary matroids László Lovász Eötvös University, Budapest, Hungary, H-1088 Princeton University, Princeton, NJ 08544 Ákos Seress*

More information

Diskrete Mathematik und Optimierung

Diskrete Mathematik und Optimierung Diskrete Mathematik und Optimierung Winfried Hochstättler, Robert Nickel: On the Chromatic Number of an Oriented Matroid Technical Report feu-dmo007.07 Contact: winfried.hochstaettler@fernuni-hagen.de

More information

WHAT IS A MATROID? JAMES OXLEY

WHAT IS A MATROID? JAMES OXLEY WHAT IS A MATROID? JAMES OXLEY Abstract. Matroids were introduced by Whitney in 1935 to try to capture abstractly the essence of dependence. Whitney s definition embraces a surprising diversity of combinatorial

More information

What Do Lattice Paths Have To Do With Matrices, And What Is Beyond Both?

What Do Lattice Paths Have To Do With Matrices, And What Is Beyond Both? What Do Lattice Paths Have To Do With Matrices, And What Is Beyond Both? Joseph E. Bonin The George Washington University These slides are available at blogs.gwu.edu/jbonin Some ideas in this talk were

More information

Unless otherwise specified, V denotes an arbitrary finite-dimensional vector space.

Unless otherwise specified, V denotes an arbitrary finite-dimensional vector space. MAT 90 // 0 points Exam Solutions Unless otherwise specified, V denotes an arbitrary finite-dimensional vector space..(0) Prove: a central arrangement A in V is essential if and only if the dual projective

More information

Regular matroids without disjoint circuits

Regular matroids without disjoint circuits Regular matroids without disjoint circuits Suohai Fan, Hong-Jian Lai, Yehong Shao, Hehui Wu and Ju Zhou June 29, 2006 Abstract A cosimple regular matroid M does not have disjoint circuits if and only if

More information

AN INTRODUCTION TO TRANSVERSAL MATROIDS

AN INTRODUCTION TO TRANSVERSAL MATROIDS AN INTRODUCTION TO TRANSVERSAL MATROIDS JOSEPH E BONIN October 26, 2010 CONTENTS 1 Prefatory Remarks 1 2 Several Perspectives on Transversal Matroids 2 21 Set systems, transversals, partial transversals,

More information

EXCLUDING SUBDIVISIONS OF INFINITE CLIQUES. Neil Robertson* Department of Mathematics Ohio State University 231 W. 18th Ave. Columbus, Ohio 43210, USA

EXCLUDING SUBDIVISIONS OF INFINITE CLIQUES. Neil Robertson* Department of Mathematics Ohio State University 231 W. 18th Ave. Columbus, Ohio 43210, USA EXCLUDING SUBDIVISIONS OF INFINITE CLIQUES Neil Robertson* Department of Mathematics Ohio State University 231 W. 18th Ave. Columbus, Ohio 43210, USA P. D. Seymour Bellcore 445 South St. Morristown, New

More information

Finite affine planes in projective spaces

Finite affine planes in projective spaces Finite affine planes in projective spaces J. A.Thas H. Van Maldeghem Ghent University, Belgium {jat,hvm}@cage.ugent.be Abstract We classify all representations of an arbitrary affine plane A of order q

More information

Submodular Functions, Optimization, and Applications to Machine Learning

Submodular Functions, Optimization, and Applications to Machine Learning Submodular Functions, Optimization, and Applications to Machine Learning Spring Quarter, Lecture http://www.ee.washington.edu/people/faculty/bilmes/classes/eeb_spring_0/ Prof. Jeff Bilmes University of

More information

Finite connectivity in infinite matroids

Finite connectivity in infinite matroids Finite connectivity in infinite matroids Henning Bruhn Paul Wollan Abstract We introduce a connectivity function for infinite matroids with properties similar to the connectivity function of a finite matroid,

More information

arxiv: v1 [math.co] 19 Oct 2018

arxiv: v1 [math.co] 19 Oct 2018 On the structure of spikes arxiv:1810.08416v1 [math.co] 19 Oct 2018 Abstract Vahid Ghorbani, Ghodratollah Azadi and Habib Azanchiler Department of mathematics, University of Urmia, Iran Spikes are an important

More information

Combinatorial Geometries Representable over GF(3) and GF(q). I. The Number of Points*

Combinatorial Geometries Representable over GF(3) and GF(q). I. The Number of Points* Discrete Comput Geom 5:83-95 (1990) rete 8, Comlmtatk~ml omet Combinatorial Geometries Representable over GF(3) and GF(q). I. The Number of Points* Joseph P. S. Kung Department of Mathematics, North Texas

More information

AN UPGRADED WHEELS-AND-WHIRLS THEOREM FOR 3-CONNECTED MATROIDS. 1. Introduction

AN UPGRADED WHEELS-AND-WHIRLS THEOREM FOR 3-CONNECTED MATROIDS. 1. Introduction AN UPGRADED WHEELS-AND-WHIRLS THEOREM FOR 3-CONNECTED MATROIDS JAMES OXLEY, CHARLES SEMPLE, AND GEOFF WHITTLE Abstract. Let M be a 3-connected matroid that is not a wheel or a whirl. In this paper, we

More information

ON A FAMILY OF PLANAR BICRITICAL GRAPHS

ON A FAMILY OF PLANAR BICRITICAL GRAPHS ON A FAMILY OF PLANAR BICRITICAL GRAPHS By L. LOVASZf and M. D. PLUMMER [Received 19 August 1973] 1. Introduction A l-factor of a graph G is a set of independent lines in 0 which span V(O). Tutte ([7])

More information

Linear Point Sets and Rédei Type k-blocking

Linear Point Sets and Rédei Type k-blocking Journal of Algebraic Combinatorics 14 (2001), 221 228 c 2001 Kluwer Academic Publishers. Manufactured in The Netherlands. Linear Point Sets and Rédei Type k-blocking Sets in PG(n, q) L. STORME ls@cage.rug.ac.be

More information

Tree-width and planar minors

Tree-width and planar minors Tree-width and planar minors Alexander Leaf and Paul Seymour 1 Princeton University, Princeton, NJ 08544 May 22, 2012; revised March 18, 2014 1 Supported by ONR grant N00014-10-1-0680 and NSF grant DMS-0901075.

More information

MENGER'S THEOREM AND MATROIDS

MENGER'S THEOREM AND MATROIDS MENGER'S THEOREM AND MATROIDS R. A. BRUALDI 1. Introduction Let G be a finite directed graph with X, Y disjoint subsets of the nodes of G. Menger's theorem [6] asserts that the maximum cardinal number

More information

Sequences of height 1 primes in Z[X]

Sequences of height 1 primes in Z[X] Sequences of height 1 primes in Z[X] Stephen McAdam Department of Mathematics University of Texas Austin TX 78712 mcadam@math.utexas.edu Abstract: For each partition J K of {1, 2,, n} (n 2) with J 2, let

More information

Parity Versions of 2-Connectedness

Parity Versions of 2-Connectedness Parity Versions of 2-Connectedness C. Little Institute of Fundamental Sciences Massey University Palmerston North, New Zealand c.little@massey.ac.nz A. Vince Department of Mathematics University of Florida

More information

LECTURE 3 Matroids and geometric lattices

LECTURE 3 Matroids and geometric lattices LECTURE 3 Matroids and geometric lattices 3.1. Matroids A matroid is an abstraction of a set of vectors in a vector space (for us, the normals to the hyperplanes in an arrangement). Many basic facts about

More information

AN ALGORITHM FOR CONSTRUCTING A k-tree FOR A k-connected MATROID

AN ALGORITHM FOR CONSTRUCTING A k-tree FOR A k-connected MATROID AN ALGORITHM FOR CONSTRUCTING A k-tree FOR A k-connected MATROID NICK BRETTELL AND CHARLES SEMPLE Dedicated to James Oxley on the occasion of his 60th birthday Abstract. For a k-connected matroid M, Clark

More information

M INIM UM DEGREE AND F-FACTORS IN GRAPHS

M INIM UM DEGREE AND F-FACTORS IN GRAPHS NEW ZEALAND JOURNAL OF MATHEMATICS Volume 29 (2000), 33-40 M INIM UM DEGREE AND F-FACTORS IN GRAPHS P. K a t e r in is a n d N. T s ik o p o u l o s (Received June 1998) Abstract. Let G be a graph, a,

More information

Characterizations of the finite quadric Veroneseans V 2n

Characterizations of the finite quadric Veroneseans V 2n Characterizations of the finite quadric Veroneseans V 2n n J. A. Thas H. Van Maldeghem Abstract We generalize and complete several characterizations of the finite quadric Veroneseans surveyed in [3]. Our

More information

Solutions to Exercises Chapter 10: Ramsey s Theorem

Solutions to Exercises Chapter 10: Ramsey s Theorem Solutions to Exercises Chapter 10: Ramsey s Theorem 1 A platoon of soldiers (all of different heights) is in rectangular formation on a parade ground. The sergeant rearranges the soldiers in each row of

More information

THE MINIMALLY NON-IDEAL BINARY CLUTTERS WITH A TRIANGLE 1. INTRODUCTION

THE MINIMALLY NON-IDEAL BINARY CLUTTERS WITH A TRIANGLE 1. INTRODUCTION THE MINIMALLY NON-IDEAL BINARY CLUTTERS WITH A TRIANGLE AHMAD ABDI AND BERTRAND GUENIN ABSTRACT. It is proved that the lines of the Fano plane and the odd circuits of K 5 constitute the only minimally

More information

ON LINEAR CODES WHOSE WEIGHTS AND LENGTH HAVE A COMMON DIVISOR. 1. Introduction

ON LINEAR CODES WHOSE WEIGHTS AND LENGTH HAVE A COMMON DIVISOR. 1. Introduction ON LINEAR CODES WHOSE WEIGHTS AND LENGTH HAVE A COMMON DIVISOR SIMEON BALL, AART BLOKHUIS, ANDRÁS GÁCS, PETER SZIKLAI, AND ZSUZSA WEINER Abstract. In this paper we prove that a set of points (in a projective

More information

Codewords of small weight in the (dual) code of points and k-spaces of P G(n, q)

Codewords of small weight in the (dual) code of points and k-spaces of P G(n, q) Codewords of small weight in the (dual) code of points and k-spaces of P G(n, q) M. Lavrauw L. Storme G. Van de Voorde October 4, 2007 Abstract In this paper, we study the p-ary linear code C k (n, q),

More information

THE DIRECT SUM, UNION AND INTERSECTION OF POSET MATROIDS

THE DIRECT SUM, UNION AND INTERSECTION OF POSET MATROIDS SOOCHOW JOURNAL OF MATHEMATICS Volume 28, No. 4, pp. 347-355, October 2002 THE DIRECT SUM, UNION AND INTERSECTION OF POSET MATROIDS BY HUA MAO 1,2 AND SANYANG LIU 2 Abstract. This paper first shows how

More information

THE CLOSED-POINT ZARISKI TOPOLOGY FOR IRREDUCIBLE REPRESENTATIONS. K. R. Goodearl and E. S. Letzter

THE CLOSED-POINT ZARISKI TOPOLOGY FOR IRREDUCIBLE REPRESENTATIONS. K. R. Goodearl and E. S. Letzter THE CLOSED-POINT ZARISKI TOPOLOGY FOR IRREDUCIBLE REPRESENTATIONS K. R. Goodearl and E. S. Letzter Abstract. In previous work, the second author introduced a topology, for spaces of irreducible representations,

More information

Planar Ramsey Numbers for Small Graphs

Planar Ramsey Numbers for Small Graphs Planar Ramsey Numbers for Small Graphs Andrzej Dudek Department of Mathematics and Computer Science Emory University Atlanta, GA 30322, USA Andrzej Ruciński Faculty of Mathematics and Computer Science

More information

Sergey Norin Department of Mathematics and Statistics McGill University Montreal, Quebec H3A 2K6, Canada. and

Sergey Norin Department of Mathematics and Statistics McGill University Montreal, Quebec H3A 2K6, Canada. and NON-PLANAR EXTENSIONS OF SUBDIVISIONS OF PLANAR GRAPHS Sergey Norin Department of Mathematics and Statistics McGill University Montreal, Quebec H3A 2K6, Canada and Robin Thomas 1 School of Mathematics

More information

Irreducible subgroups of algebraic groups

Irreducible subgroups of algebraic groups Irreducible subgroups of algebraic groups Martin W. Liebeck Department of Mathematics Imperial College London SW7 2BZ England Donna M. Testerman Department of Mathematics University of Lausanne Switzerland

More information

Branchwidth of graphic matroids.

Branchwidth of graphic matroids. Branchwidth of graphic matroids. Frédéric Mazoit and Stéphan Thomassé Abstract Answering a question of Geelen, Gerards, Robertson and Whittle [2], we prove that the branchwidth of a bridgeless graph is

More information

Section Summary. Relations and Functions Properties of Relations. Combining Relations

Section Summary. Relations and Functions Properties of Relations. Combining Relations Chapter 9 Chapter Summary Relations and Their Properties n-ary Relations and Their Applications (not currently included in overheads) Representing Relations Closures of Relations (not currently included

More information

ON THE HIGHLY CONNECTED DYADIC, NEAR-REGULAR, AND SIXTH-ROOT-OF-UNITY MATROIDS

ON THE HIGHLY CONNECTED DYADIC, NEAR-REGULAR, AND SIXTH-ROOT-OF-UNITY MATROIDS ON THE HIGHLY CONNECTED DYADIC, NEAR-REGULAR, AND SIXTH-ROOT-OF-UNITY MATROIDS BEN CLARK, KEVIN GRACE, JAMES OXLEY, AND STEFAN H.M. VAN ZWAM arxiv:1903.04910v1 [math.co] 9 Mar 2019 Abstract. Subject to

More information

Generalized Quadrangles Weakly Embedded in Finite Projective Space

Generalized Quadrangles Weakly Embedded in Finite Projective Space Generalized Quadrangles Weakly Embedded in Finite Projective Space J. A. Thas H. Van Maldeghem Abstract We show that every weak embedding of any finite thick generalized quadrangle of order (s, t) in a

More information

Cycles in 4-Connected Planar Graphs

Cycles in 4-Connected Planar Graphs Cycles in 4-Connected Planar Graphs Guantao Chen Department of Mathematics & Statistics Georgia State University Atlanta, GA 30303 matgcc@panther.gsu.edu Genghua Fan Institute of Systems Science Chinese

More information

Maps of matroids with applications

Maps of matroids with applications Discrete Mathematics 303 (2005) 75 85 www.elsevier.com/locate/disc Maps of matroids with applications András Recski a,b, a Department of Computer Science and Information Theory, Budapest University of

More information

Course 311: Michaelmas Term 2005 Part III: Topics in Commutative Algebra

Course 311: Michaelmas Term 2005 Part III: Topics in Commutative Algebra Course 311: Michaelmas Term 2005 Part III: Topics in Commutative Algebra D. R. Wilkins Contents 3 Topics in Commutative Algebra 2 3.1 Rings and Fields......................... 2 3.2 Ideals...............................

More information

Hodge theory for combinatorial geometries

Hodge theory for combinatorial geometries Hodge theory for combinatorial geometries June Huh with Karim Adiprasito and Eric Katz June Huh 1 / 48 Three fundamental ideas: June Huh 2 / 48 Three fundamental ideas: The idea of Bernd Sturmfels that

More information

Page Line Change 80-5 Replace by Adjoin or delete a zero row Omit non-zero before column Replace the sentence beginning Clearly

Page Line Change 80-5 Replace by Adjoin or delete a zero row Omit non-zero before column Replace the sentence beginning Clearly MATROID THEORY James G. Oxley Oxford University Press, New York, 1992 Errata and Update on Conjectures, Problems, and References Latest update: December 10, 2005 The comments below apply to all printings

More information

ON THE CORE OF A GRAPHf

ON THE CORE OF A GRAPHf ON THE CORE OF A GRAPHf By FRANK HARARY and MICHAEL D. PLUMMER [Received 8 October 1965] 1. Introduction Let G be a graph. A set of points M is said to cover all the lines of G if every line of G has at

More information

CONVENIENT PRETOPOLOGIES ON Z 2

CONVENIENT PRETOPOLOGIES ON Z 2 TWMS J. Pure Appl. Math., V.9, N.1, 2018, pp.40-51 CONVENIENT PRETOPOLOGIES ON Z 2 J. ŠLAPAL1 Abstract. We deal with pretopologies on the digital plane Z 2 convenient for studying and processing digital

More information

Compatible Circuit Decompositions of 4-Regular Graphs

Compatible Circuit Decompositions of 4-Regular Graphs Compatible Circuit Decompositions of 4-Regular Graphs Herbert Fleischner, François Genest and Bill Jackson Abstract A transition system T of an Eulerian graph G is a family of partitions of the edges incident

More information

A necessary and sufficient condition for the existence of a spanning tree with specified vertices having large degrees

A necessary and sufficient condition for the existence of a spanning tree with specified vertices having large degrees A necessary and sufficient condition for the existence of a spanning tree with specified vertices having large degrees Yoshimi Egawa Department of Mathematical Information Science, Tokyo University of

More information

The Reduction of Graph Families Closed under Contraction

The Reduction of Graph Families Closed under Contraction The Reduction of Graph Families Closed under Contraction Paul A. Catlin, Department of Mathematics Wayne State University, Detroit MI 48202 November 24, 2004 Abstract Let S be a family of graphs. Suppose

More information

The Catalan matroid.

The Catalan matroid. The Catalan matroid. arxiv:math.co/0209354v1 25 Sep 2002 Federico Ardila fardila@math.mit.edu September 4, 2002 Abstract We show how the set of Dyck paths of length 2n naturally gives rise to a matroid,

More information

The cycle polynomial of a permutation group

The cycle polynomial of a permutation group The cycle polynomial of a permutation group Peter J. Cameron School of Mathematics and Statistics University of St Andrews North Haugh St Andrews, Fife, U.K. pjc0@st-andrews.ac.uk Jason Semeraro Department

More information

GENERALIZED PIGEONHOLE PROPERTIES OF GRAPHS AND ORIENTED GRAPHS

GENERALIZED PIGEONHOLE PROPERTIES OF GRAPHS AND ORIENTED GRAPHS GENERALIZED PIGEONHOLE PROPERTIES OF GRAPHS AND ORIENTED GRAPHS ANTHONY BONATO, PETER CAMERON, DEJAN DELIĆ, AND STÉPHAN THOMASSÉ ABSTRACT. A relational structure A satisfies the n k property if whenever

More information

Matroid intersection, base packing and base covering for infinite matroids

Matroid intersection, base packing and base covering for infinite matroids Matroid intersection, base packing and base covering for infinite matroids Nathan Bowler Johannes Carmesin June 25, 2014 Abstract As part of the recent developments in infinite matroid theory, there have

More information

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution

More information

Planar Graphs (1) Planar Graphs (3) Planar Graphs (2) Planar Graphs (4)

Planar Graphs (1) Planar Graphs (3) Planar Graphs (2) Planar Graphs (4) S-72.2420/T-79.5203 Planarity; Edges and Cycles 1 Planar Graphs (1) Topological graph theory, broadly conceived, is the study of graph layouts. Contemporary applications include circuit layouts on silicon

More information

Matroids on graphs. Brigitte Servatius Worcester Polytechnic Institute. First Prev Next Last Go Back Full Screen Close Quit

Matroids on graphs. Brigitte Servatius Worcester Polytechnic Institute. First Prev Next Last Go Back Full Screen Close Quit on K n on graphs Brigitte Servatius Worcester Polytechnic Institute Page 1 of 35 on K n Page 2 of 35 1. Whitney [9] defined a matroid M on a set E: M = (E, I) E is a finite set I is a collection of subsets

More information

ORBIT-HOMOGENEITY IN PERMUTATION GROUPS

ORBIT-HOMOGENEITY IN PERMUTATION GROUPS Submitted exclusively to the London Mathematical Society DOI: 10.1112/S0000000000000000 ORBIT-HOMOGENEITY IN PERMUTATION GROUPS PETER J. CAMERON and ALEXANDER W. DENT Abstract This paper introduces the

More information

Publications of James G. Oxley. Books and edited books

Publications of James G. Oxley. Books and edited books Books and edited books Publications of James G. Oxley 1. Matroid Theory, Oxford University Press, New York, 1992 (532 pages). 2. Matroid Theory (edited with J.E. Bonin and B. Servatius), Proc. AMS- IMS-SIAM

More information

Minimal Paths and Cycles in Set Systems

Minimal Paths and Cycles in Set Systems Minimal Paths and Cycles in Set Systems Dhruv Mubayi Jacques Verstraëte July 9, 006 Abstract A minimal k-cycle is a family of sets A 0,..., A k 1 for which A i A j if and only if i = j or i and j are consecutive

More information

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution

More information

Compatible Circuit Decompositions of Eulerian Graphs

Compatible Circuit Decompositions of Eulerian Graphs Compatible Circuit Decompositions of Eulerian Graphs Herbert Fleischner, François Genest and Bill Jackson Septemeber 5, 2006 1 Introduction Let G = (V, E) be an Eulerian graph. Given a bipartition (X,

More information

Spanning cycles in regular matroids without M (K 5 ) minors

Spanning cycles in regular matroids without M (K 5 ) minors Spanning cycles in regular matroids without M (K 5 ) minors Hong-Jian Lai, Bolian Liu, Yan Liu, Yehong Shao Abstract Catlin and Jaeger proved that the cycle matroid of a 4-edge-connected graph has a spanning

More information

A CHAIN THEOREM FOR MATROIDS

A CHAIN THEOREM FOR MATROIDS A CHAIN THEOREM FOR MATROIDS JAMES OXLEY, CHARLES SEMPLE, AND GEOFF WHITTLE Abstract. Tutte s Wheels-and-Whirls Theorem proves that if M is a 3-connected matroid other than a wheel or a whirl, then M has

More information