Some Problems and Generalizations onErdos-Ko-Rado Theorem
Huajun Zhang
Department of Mathematics, Zhejiang Normal UniversityZhejiang 321004, P. R. China
December 27, 2013, Shanghai Jiao Tong University
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Theorem (EKR Theorem)
If A is an intersecting family of k-subsets of [n] = {1, 2, . . . , n},i.e., A ∩ B 6= ∅ for any A,B ∈ A, then
|A| ≤(
n − 1
k − 1
)subject to n ≥ 2k. Equality holds if and only if every subset in Acontains a common element of [n] except for n = 2k.
P. Erdos, C. Ko and R. Rado, Intersection theorems for systems offinite sets, Quart. J. Math. Oxford Ser., 2 (1961), 313-318.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Theorem (EKR Theorem for Finite Vector Spaces)
If A is an intersecting family of k-dimensional subspaces of ann-dimensional vector space over the q-element field, i.e.,dim(A ∩ B) ≥ 1 for any A,B ∈ A, then
|A| ≤[
n − 1
k − 1
]subject to n ≥ 2k. Equality holds if and only if every subset in Acontains a common nonzero vector except the case n = 2k.
W. N. Hsieh, Intersection theorems for systems of finite vectorspaces, Discrete Math., 12 (1975), 1-16.
C. Greene and D. J. Kleitman, Proof techniques in the ordered sets,in: G.-C. Rota, ed., “Studies in Combinatorics” 1978, 22-79.
P. Frankl, R. M. Wilson, The Erdos-Ko-Rado theorem for vectorspaces, JCTA 43 (1986), 228-236.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Theorem (EKR Theorem for Permutations)
If A is an intersecting family in Sn (the symmetric group on [n]),i.e., for each pair σ, τ ∈ Sn there is an i ∈ [n] with σ(i) = τ(i),then
|A| ≤ (n − 1)!.
Equality holds if and only if A is a coset of the stabilizer of a point.
M. Deza and P. Frankl, On the maximum number of permutationswith given maximal or minimal distance, JCTA 22(1977) 352-362.
P. Cameron and C.Y. Ku, Intersecting families of permutations,EuJC 24 (2003), 881-890.
JW, J. Zhang, An Erdos-Ko-Rado-type theorem in Coxeter groups,EuJC 29 (2008), 1112-1115.
D. Ellis, E. Friedgut, H. Pilpel, Intersecting families of permutations,Journal of the American Mathematical Society 24 (2011) 649-682.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
A q-signed k-set is a pair (A, f ), where A ⊆ [n] is a k-set andf is a function from A to [q]. A family F of q-signed k-sets isintersecting if for any (A, f ), (B, g) ∈ F there existsx ∈ A ∩ B such that f (x) = g(x).
Set Bkn(q) = {(A, f ) : A ∈([n]k
)} and Bn(q) =
⋃ni=0 Bkn(q).
A r -partial permutation of [n] is a pair (A, f ) with A ∈([n]r
)and f is an injective map from A to [n].
the set of all r -partial permutations of [n] denoted by Pr ,n.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Theorem (EKR Theorem for Signed Sets)
(Bollobas and Leader) Fix a positive integer k ≤ n, and let F bean intersecting family of q-signed k-sets on [n], where q ≥ 2. Then|F| ≤
(n−1k−1
)qk−1. Unless q = 2 and k = n, equality holds if and
only if F consists of all q-signed k-sets (A, f ) such that x0 ∈ Aand f (x0) = ε0 for some fixed x0 ∈ [n], ε0 ∈ [q].
Bollobas B. Bollobas and I. Leader, An Erdos-Ko-Rado theorem forsigned sets, Comput. Math. Applic. 34 (11) (1997) 9-13.
Y.S. Li, J. Wang, Erdos-Ko-Rado-Type Theorems for Colored Sets,Electron. J. Combin. 14 (1) (2007).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Theorem (EKR Theorem for Partial Permutation)
Fix a positive integer r < n, and let F be an intersecting family of
Pr ,n. Then |F| ≤(n−1r−1
)(n − 1)!
(n − r)!. Equality holds if and only if F
consists of all r -partial permutations (A, f ) such that i ∈ A andf (i) = j for some fixed i , j ∈ [n].
C. Y. Ku and I. Leader, An Erdos-Ko-Rado theorem for partialpermutations, Disc. Math. 306 (2006) 74-86.
Y.S. Li, J. Wang, Erdos-Ko-Rado-Type Theorems for Colored Sets,Electron. J. Combin. 14 (1) (2007).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Theorem (Hilton, 1977)
Let A1,A2, . . . ,Am be cross-intersecting families of k subsets of[n] with A1 6= ∅, i.e., for any Ai ∈ Ai and Aj ∈ Aj , i 6= j ,Ai ∩ Aj 6= ∅. If k ≤ n/2, then
m∑i=1
|Ai | ≤{ (n
k
), if m ≤ n/k;
m(n−1k−1
), if m ≥ n/k.
Unless m = 2 = n/k, the bound is attained if and only if one ofthe following holds:
(i) m ≤ n/k and A1 =([n]k
), and A2 = · · · = Am = ∅;
(ii) m ≥ n/k and |A1| = |A2| = . . . = |Am| =(n−1k−1
).
A.J.W. Hilton, An intersection theorem for a collection of families ofsubsets of a finite set, J. London Math. Soc. 2 (1977) 369-384.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Results
The Hilton Theorem was generalized to partial permutation, signedsets and labled sets.
P. Borg, Cross-intersecting families of permutations, J. Combin.Theory Ser. A, 117 (2010) 483-487.
P. Borg, Intersecting and cross-independent families of labeled sets,Electron. J. Combin. 15 (2008) N9.
P. Borg and I. Leader, Multiple cross-intersecting families of signedsets, J. Combin. Theory Ser. A, 117 (2010) 583-588.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Results
The Hilton Theorem was generalized to general case.
J. Wang, H.J. Zhang, Cross-intersecting families and primitivity ofsymmetric systems, J. Combin. Theory Ser. A 118 (2011) 455-462.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Theorem (Hilton and Milner 1967)
Let n and a be two positive integers with n ≥ 2a. If A,B ⊆([n]a
)with A ∩ B 6= ∅ for all A ∈ A and B ∈ B, then
|A|+ |B| ≤(
n
a
)−(
n − a
a
)+ 1.
A.J.W. Hilton and E.C. Milner, Some intersection theorems forsystems of finite sets, Quart. J. Math. Oxford 18 (1967) 369-384.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Theorem ( Frankl and Tohushige)
Let n, a and b be three positive integers with n ≥ a + b and a ≤ b.If A ⊆
([n]a
)and B ⊆
([n]b
)with A ∩ B 6= ∅ for all A ∈ A and
B ∈ B, then
|A|+ |B| ≤(
n
b
)−(
n − a
b
)+ 1.
P. Frankl and N. Tohushige, Some best possible inequalitiesconcerning cross-intersecting families, J. Combin. Theory Ser. A 61(1992) 87-97.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Results
The Hilton-Milner Theorem was generalized to the general cases.
J. Wang, H.J. Zhang, Nontrivial independent sets of bipartitegraphs and cross-intersecting families, J. Combin. Theory Ser. A,120 (2013) 129-141.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Theorem
If A ⊆([n]k
)and B ⊆
([n]`
)are cross-intersecting with k , ` ≤ n/2,
then
|A||B| ≤(
n − 1
k − 1
)(n − 1
`− 1
).
Moreover, the equality holds if and only if A = {A ∈([n]k
): i ∈ A}
and B = {B ∈([n]
`
): i ∈ B} for some i ∈ [n], unless n = 2k = 2`.
L. Pyber, A new generalization of the Erdos-Rado-Ko theorem, J.Combin. Theory Ser. A, 43 (1986) 85-90.
M. Matsumoto, N. Tokushige, The exact bound in theErdos-Rado-Ko theorem for cross-intersecting families, J. Combin.Theory Ser. A, 52 (1989) 90-97.
C. Bey, On cross-intersecting families of sets, Graphs Combin., 21(2005) 161-168.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Theorem (Tokushige)
Let p be a real with 0 < p < 0.114, and let t be an integer with1 ≤ t ≤ 1/(2p). For fixed p and t there exist positive constants ε,
n1 such that for all integers n, k with n > n1 and |kn− p| < ε,the
following is true: if two families A1 ⊂([n]k
)and A2 ⊂
([n]k
)are
cross t-intersecting, then
|A1||A2| ≤(
n − t
k − t
)2
with equality holding iff A1 = A2 = {F ∈([n]k
): [t] ⊂ F} (up to
isomorphism).
N. Tokushige, On cross t-intersecting families of sets, J. Combin.Theory Ser. A, 117 (2010)1167-1177.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Theorem (Ellis, Friedgut, Pilpel)
For any positive integer k and any n sufficiently large depending onk, if I , J ⊂ Sn are k-cross-intersecting, then |I ||J| ≤ ((n − k)!)2.Equality holds if and only if I = J and I is a k-coset of Sn.
D. Ellis, E. Friedgut, H. Pilpel, Intersecting families of permutations,Journal of the American Mathematical Society 24 (2011) 649-682.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
1. Erdos-Ko-Rado Theorem
Theorem
Let n and p be two positive integer with p ≥ 4. If A and B arecross-intersecting families inLp =
{{(1, `1), (2, `2), . . . , (n, `n)} : `i ∈ [p], i = 1, 2 . . . , n
}, then
|A||B| ≤ p2n−2,
and equality holds if and only ifA = B =
{{(1, `1), (2, `2), . . . , (n, `n)} : `i = j
}for some i ∈ [n]
and j ∈ [p].
H.J. Zhang, Cross-intersecting families in labeled sets, Electron. J.Combin. 20 (2013).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
2. 2-independent sets in Kneser graphs
Let G be a simple graph with vertex set V (G ) and edge set E (G ).S ⊆ V (G ).
S is an independent set of G if no two elements of S areadjacent in G .
S is a k-independent set of G if S can be expressed as a unionof k independent sets of G .
k-independence number αk(G ): the maximum size ofk-independent sets of G ,
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
2. 2-independent sets in Kneser graphs
Let G be a simple graph with vertex set V (G ) and edge set E (G ).S ⊆ V (G ).
S is an independent set of G if no two elements of S areadjacent in G .
S is a k-independent set of G if S can be expressed as a unionof k independent sets of G .
k-independence number αk(G ): the maximum size ofk-independent sets of G ,
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
2. 2-independent sets in Kneser graphs
Let G be a simple graph with vertex set V (G ) and edge set E (G ).S ⊆ V (G ).
S is an independent set of G if no two elements of S areadjacent in G .
S is a k-independent set of G if S can be expressed as a unionof k independent sets of G .
k-independence number αk(G ): the maximum size ofk-independent sets of G ,
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
2. 2-independent sets in Kneser graphs
The Kneser graph K (n, k)(2k ≤ n): vertex set([n]k
)and A ∼ B iff
A ∩ B = ∅.Erdos-Ko-Rado: α1(K (n, k)) =
(n−1k−1
).
Given a graph G , what is α1(G )?
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
2. 2-independent sets in Kneser graphs
The Kneser graph K (n, k)(2k ≤ n): vertex set([n]k
)and A ∼ B iff
A ∩ B = ∅.Erdos-Ko-Rado: α1(K (n, k)) =
(n−1k−1
).
Given a graph G , what is α1(G )?
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
2. 2-independent sets in Kneser graphs
Given a graph G , what is αk(G )?
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
2. 2-independent sets in Kneser graphs
Note that A = {A ∈([n]k
): A ∩ {1, 2} 6= ∅} is a 2-independent set
of K (n, k).
Someone conjectured that α2(K (n, k)) =(n−1k−1
)+(n−2k−1
)for
sufficient larger n.
C.Godsil et al: α2(K (9, 4)) = 95 or 96
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
2. 2-independent sets in Kneser graphs
Note that A = {A ∈([n]k
): A ∩ {1, 2} 6= ∅} is a 2-independent set
of K (n, k).
Someone conjectured that α2(K (n, k)) =(n−1k−1
)+(n−2k−1
)for
sufficient larger n.
C.Godsil et al: α2(K (9, 4)) = 95 or 96
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
2. 2-independent sets in Kneser graphs
Example
A =
{A ∈
([9]
4
): 1 ∈ A and A ∩ {2, 3, 4} 6= ∅
}B =
{B ∈
([9]
4
): 1 6∈ A and {2, 3, 4} ⊂ B
}C =
{C ∈
([9]
4
): |C ∩ {5, 6, 7, 8, 9}| ≥ 3
}Clearly, A ∪ B and C are two disjoint independent sets of K (9, 4),and so A ∪ B ∪ C is a 2-independent set.
|A|+ |B|+ |C| =
(8
3
)−(
5
3
)+
(5
1
)+
(5
4
)+
(5
3
)(4
1
)= 96.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
2. 2-independent sets in Kneser graphs
Example
A =
{A ∈
([9]
4
): 1 ∈ A and A ∩ {2, 3} 6= ∅
}B =
{B ∈
([9]
4
): 1 6∈ A and {2, 3} ⊂ B
}C =
{C ∈
([9]
4
): |C ∩ {5, 6, 7, 8, 9}| ≥ 3
}A ∪ B ∪ C is also a 2-independent set.
|A|+ |B|+ |C| = 96.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
2. 2-independent sets in Kneser graphs
For n = 2k + 1, we conjecture
α2(K (n, k)) =
(n − 1
k − 1
)+
(n − 2
k − 1
)
+
∑
i≥dk/2e+1
(k
i
)(k
k − i
), if k is odd;
∑i≥dk/2e+1
(k − 1
i
)(k + 1
k − i
), if k is even.
α2(K (9, 4)) = 96.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
2. 2-independent sets in Kneser graphs
For n > 2k + 1, we conjecture
α2(K (n, k)) =
(n − 1
k − 1
)+
(n − 2
k − 1
).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Theorem (Turan 1941)
Let G (V ,E ) be a graph on n vertices without k-clique, then
|E | ≤ (k − 2)n2
2(k − 1).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
A hypergraph H is a pair H = (X , E) where X is a set ofelements, called vertices, and E is a set of non-empty subsetsof X called edges.
H is said to be k-uniform if |E | = k for all E ∈ E .
H is said to be complete k-uniform if E =(Xk
).
A family {E1,E2, . . . ,Es} ⊂ E is called a matching if Ei ’s arepairwise disjoint.
For F ⊆ E , the matching number ν(F) is the size of themaximum matching contained in F .
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
A hypergraph H is a pair H = (X , E) where X is a set ofelements, called vertices, and E is a set of non-empty subsetsof X called edges.
H is said to be k-uniform if |E | = k for all E ∈ E .
H is said to be complete k-uniform if E =(Xk
).
A family {E1,E2, . . . ,Es} ⊂ E is called a matching if Ei ’s arepairwise disjoint.
For F ⊆ E , the matching number ν(F) is the size of themaximum matching contained in F .
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
A hypergraph H is a pair H = (X , E) where X is a set ofelements, called vertices, and E is a set of non-empty subsetsof X called edges.
H is said to be k-uniform if |E | = k for all E ∈ E .
H is said to be complete k-uniform if E =(Xk
).
A family {E1,E2, . . . ,Es} ⊂ E is called a matching if Ei ’s arepairwise disjoint.
For F ⊆ E , the matching number ν(F) is the size of themaximum matching contained in F .
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
A hypergraph H is a pair H = (X , E) where X is a set ofelements, called vertices, and E is a set of non-empty subsetsof X called edges.
H is said to be k-uniform if |E | = k for all E ∈ E .
H is said to be complete k-uniform if E =(Xk
).
A family {E1,E2, . . . ,Es} ⊂ E is called a matching if Ei ’s arepairwise disjoint.
For F ⊆ E , the matching number ν(F) is the size of themaximum matching contained in F .
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
A hypergraph H is a pair H = (X , E) where X is a set ofelements, called vertices, and E is a set of non-empty subsetsof X called edges.
H is said to be k-uniform if |E | = k for all E ∈ E .
H is said to be complete k-uniform if E =(Xk
).
A family {E1,E2, . . . ,Es} ⊂ E is called a matching if Ei ’s arepairwise disjoint.
For F ⊆ E , the matching number ν(F) is the size of themaximum matching contained in F .
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Erdos posed the following conjecture.
Conjecture
If F ⊂([n]k
), ν(F) = s and n ≥ (s + 1)k, then
|F| ≤ max
{((sk + k − 1
k
),
(n
k
)−(
n − s
k
)}. (1)
The case s = 1 is the classical Erdos-Ko-Rado Theorem.
P. Erdos, A problem on independent r-tuples, Ann. Univ. Sci.Budapest. Eotvos Sect. Math., 8 (1965) 93–95.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Erdos proved that the conjecture holds if n > n0(k , s).
Bollobas, Daykin and Erdos (1976) proved thatn0(k , s) = 2sk3.
Furedi and Frankl (unpublished) proved that n0(k , s) = 100k2.
Huang, Loh and Sudakov (2012) proved that n0(k , s) = 3k2.
Frankl(2013) proved that n0(k, s) = 2(s + 1)k − s.
The cases k = 2 and k = 3 were settled by Erdos and Gallai(1959) and Frankl (recently),resp.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Erdos proved that the conjecture holds if n > n0(k , s).
Bollobas, Daykin and Erdos (1976) proved thatn0(k , s) = 2sk3.
Furedi and Frankl (unpublished) proved that n0(k , s) = 100k2.
Huang, Loh and Sudakov (2012) proved that n0(k , s) = 3k2.
Frankl(2013) proved that n0(k, s) = 2(s + 1)k − s.
The cases k = 2 and k = 3 were settled by Erdos and Gallai(1959) and Frankl (recently),resp.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Erdos proved that the conjecture holds if n > n0(k , s).
Bollobas, Daykin and Erdos (1976) proved thatn0(k , s) = 2sk3.
Furedi and Frankl (unpublished) proved that n0(k , s) = 100k2.
Huang, Loh and Sudakov (2012) proved that n0(k , s) = 3k2.
Frankl(2013) proved that n0(k, s) = 2(s + 1)k − s.
The cases k = 2 and k = 3 were settled by Erdos and Gallai(1959) and Frankl (recently),resp.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Erdos proved that the conjecture holds if n > n0(k , s).
Bollobas, Daykin and Erdos (1976) proved thatn0(k , s) = 2sk3.
Furedi and Frankl (unpublished) proved that n0(k , s) = 100k2.
Huang, Loh and Sudakov (2012) proved that n0(k , s) = 3k2.
Frankl(2013) proved that n0(k, s) = 2(s + 1)k − s.
The cases k = 2 and k = 3 were settled by Erdos and Gallai(1959) and Frankl (recently),resp.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Erdos proved that the conjecture holds if n > n0(k , s).
Bollobas, Daykin and Erdos (1976) proved thatn0(k , s) = 2sk3.
Furedi and Frankl (unpublished) proved that n0(k , s) = 100k2.
Huang, Loh and Sudakov (2012) proved that n0(k , s) = 3k2.
Frankl(2013) proved that n0(k, s) = 2(s + 1)k − s.
The cases k = 2 and k = 3 were settled by Erdos and Gallai(1959) and Frankl (recently),resp.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Erdos proved that the conjecture holds if n > n0(k , s).
Bollobas, Daykin and Erdos (1976) proved thatn0(k , s) = 2sk3.
Furedi and Frankl (unpublished) proved that n0(k , s) = 100k2.
Huang, Loh and Sudakov (2012) proved that n0(k , s) = 3k2.
Frankl(2013) proved that n0(k, s) = 2(s + 1)k − s.
The cases k = 2 and k = 3 were settled by Erdos and Gallai(1959) and Frankl (recently),resp.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
A linear path of length ` is a family of sets {F1,F2, . . . ,F`}such that |Fi ∩ Fi+1| = 1 for each i and Fi ∩ Fj = ∅ whenever
|i − j | > 1. Let P(k)` denote the k-uniform linear path of
length of `.
Theorem (Furedi, Jiang and Seiver)
Let k, t be positive integers, where k ≥ 3. For sufficiently larger n,we have
exk(n;P(k)2t+1) =
(n − 1
k − 1
)+
(n − 2
k − 1
)+ . . .+
(n − t
k − 1
).
The only extremal family consists of all the k-sets in [n] that meetsome fixed set S of t vertices.
Z. Furedi, T. Jiang and R. Seiver, Exact solution of thehypergraph turan problem for k-uniform linear paths,Combinatorica, to appear.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Theorem (Furedi, Jiang and Seiver)
Let k, t be positive integers, where k ≥ 3. For sufficiently larger n,we have
exk(n;P(k)2t+2) =
(n − 1
k − 1
)+
(n − 2
k − 1
)+. . .+
(n − t
k − 1
)+
(n − t − 2
k − 2
).
The only extremal family consists of all the k-sets in [n] that meetsome fixed set S of t vertices plus all the k-sets in [n] \ S thatcontain some two fixed elements.
Z. Furedi, T. Jiang and R. Seiver, Exact solution of thehypergraph turan problem for k-uniform linear paths,Combinatorica, to appear.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
A k-uniform minimal cycle of length ` is a cyclic list of k-setsA1,A2, . . . ,A` such that consecutive sets intersect in at leastone element and nonconsecutive sets are disjoint. The set of
all k-uniform minimal cycles of length ` denoted by C(k)` .
A k-uniform linear cycle of length `, denoted by C(k)` is a
cyclic list of k-sets A1,A2, . . . ,A` such that consecutive setsintersect in exactly one element and nonconsecutive sets aredisjoint.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
A k-uniform minimal cycle of length ` is a cyclic list of k-setsA1,A2, . . . ,A` such that consecutive sets intersect in at leastone element and nonconsecutive sets are disjoint. The set of
all k-uniform minimal cycles of length ` denoted by C(k)` .
A k-uniform linear cycle of length `, denoted by C(k)` is a
cyclic list of k-sets A1,A2, . . . ,A` such that consecutive setsintersect in exactly one element and nonconsecutive sets aredisjoint.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Theorem (Furedi and Jiang)
Let t b e a positive integer, k ≥ 4. For sufficiently larger n, wehave
exk(n, C(k)2t+1) =
(n
k
)−(
n − t
k
)and
exk(n, C(k)2t+2) =
(n
k
)−(
n − t
k
)+ 1.
The only extremal family consists of all the k-sets in [n] that meet
some fixed k-set S. For C(k)2t+2, the only extremal family consists ofall the k-sets in [n] that meet some fixed k-set S plus oneadditional k-set outside S.
Z. Furedi and T. Jiang, Hypergraph turan numbers of linearcycles, arXiv:1302.2387.[math.CO]2013.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Theorem (Furedi and Jiang)
Let t b e a positive integer, k ≥ 5. For sufficiently larger n, wehave
exk(n,C(k)2t+1) =
(n
k
)−(
n − t
k
)and
exk(n,C(k)2t+2) =
(n
k
)−(
n − t
k
)+
(n − t − 2
k − 2
).
For C(k)2t+1, the only extremal family consists of all the k-sets in [n]
that meet some fixed k-set S. For C(k)2t+2, the only extremal family
consists of all the k-sets in [n] that meet some fixed k-set S plusall the k-sets in [n] \ S that contain some two two fixed elements.
Z. Furedi and T. Jiang, Hypergraph turan numbers of linearcycles, arXiv:1302.2387.[math.CO]2013.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
S = {C1 ∪ . . . ∪ Cr : Ci ∈ C(k)`i
for i ∈ [r ]}
Theorem (Gu, Li and Shi)
Let integers k ≥ 4, r ≥ 1, `1, . . . , `r ≥ 3, t =r∑
i=1
b`i + 1
2c − 1, and
I = 1 if all the `1, . . . , `r are even, I = 0 otherwise. For sufficientlylarge n,
exk(n;S(`1, . . . , `r )) =
(n
k
)−(
n − t
k
)+ I .
Ran Gu, Xueliang Li and Yongtang Shi, Hypergraph turannumbers of vertex disjoint cycles, arXive: 1305.5372v1[math.CO]2013.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Theorem (Gu, Li and Shi)
Let integers k ≥ 5, r ≥ 1, `1, . . . , `r ≥ 3, t =r∑
i=1
b`i + 1
2c − 1, and
J =(n−t−2
k
)if all the `1, . . . , `r are even, J = 0 otherwise. For
sufficiently large n,
exk(n;C(k)`1, . . . ,C(k)
`r) =
(n
k
)−(
n − t
k
)+ J.
Ran Gu, Xueliang Li and Yongtang Shi, Hypergraph turannumbers of vertex disjoint cycles, arXive: 1305.5372v1[math.CO]2013.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Bkn(q) = {(A, f ) : A ∈([n]k
)}.
Theorem
For positive integers q, n and k with k ≤ n, if F ⊂ Bkn(q) withν(F) = s where s < q, then
|F| ≤ sqk−1
(n − 1
k − 1
),
and equality holds if and only if F is isomorphic toF1 = {(A, f ) ∈ Bkn(q) : 1 ∈ A, f (1) ∈ [s]}.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Bkn(q) = {(A, f ) : A ∈([n]k
)}.
Theorem
For positive integers q, n and k with k ≤ n, if F ⊂ Bkn(q) withν(F) = s where s < q, then
|F| ≤ sqk−1
(n − 1
k − 1
),
and equality holds if and only if F is isomorphic toF1 = {(A, f ) ∈ Bkn(q) : 1 ∈ A, f (1) ∈ [s]}.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Let V be a finite set and let p be an ideal of 2V , that is, pconsists of subsets of V such that B ∈ p if B ⊆ A for someA ∈ p.
Define αp(V ) := max{|A| : A ∈ p} and write
M(V ) = {A ∈ p : |A| = αp(V )}
For H ⊂ V , write αp(H) := max{|A| : A ∈ p and A ⊆ H}.For positive integer s, set
p = {F ⊆ Bkn(q) : ν(F) ≤ s}.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Let V be a finite set and let p be an ideal of 2V , that is, pconsists of subsets of V such that B ∈ p if B ⊆ A for someA ∈ p.
Define αp(V ) := max{|A| : A ∈ p} and write
M(V ) = {A ∈ p : |A| = αp(V )}
For H ⊂ V , write αp(H) := max{|A| : A ∈ p and A ⊆ H}.For positive integer s, set
p = {F ⊆ Bkn(q) : ν(F) ≤ s}.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Let V be a finite set and let p be an ideal of 2V , that is, pconsists of subsets of V such that B ∈ p if B ⊆ A for someA ∈ p.
Define αp(V ) := max{|A| : A ∈ p} and write
M(V ) = {A ∈ p : |A| = αp(V )}
For H ⊂ V , write αp(H) := max{|A| : A ∈ p and A ⊆ H}.
For positive integer s, set
p = {F ⊆ Bkn(q) : ν(F) ≤ s}.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Let V be a finite set and let p be an ideal of 2V , that is, pconsists of subsets of V such that B ∈ p if B ⊆ A for someA ∈ p.
Define αp(V ) := max{|A| : A ∈ p} and write
M(V ) = {A ∈ p : |A| = αp(V )}
For H ⊂ V , write αp(H) := max{|A| : A ∈ p and A ⊆ H}.For positive integer s, set
p = {F ⊆ Bkn(q) : ν(F) ≤ s}.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
To complete the proof, we only need to determine αp(Bkn (q)) and
M(Bkn (q)).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Lemma
Let V be a finite set and p an idea on 2V . Suppose that there is atransitive permutation group Γ on V that keeps the ideal p, i.e.,σ(A) ∈ p for all A ∈ p and σ ∈ Γ. Then, for each H ⊆ V ,
αp(V )
|V |≤ αp(H)
|H|, (2)
and equality holds if and only if |H ∩ S | = αp(H) for eachS ∈M(V ).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Bkn(q) = {(A, f ) : A ∈([n]k
)and f is an injection from A to [q]}
1 2 3 . . . n 1 2 3 . . . n . . . 1 2 3 . . . n1 1 1 . . . 1 2 2 2 . . . 2 . . . q q q . . . q
(A, f ) is said to be contained in the above cycle if it consistsof k consecutive elements.
Let H be the set of (A, f ) which contained in the above cycle.Clearly |H| = nq.
Determine αp(H).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Bkn(q) = {(A, f ) : A ∈([n]k
)and f is an injection from A to [q]}
1 2 3 . . . n 1 2 3 . . . n . . . 1 2 3 . . . n1 1 1 . . . 1 2 2 2 . . . 2 . . . q q q . . . q
(A, f ) is said to be contained in the above cycle if it consistsof k consecutive elements.
Let H be the set of (A, f ) which contained in the above cycle.Clearly |H| = nq.
Determine αp(H).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Bkn(q) = {(A, f ) : A ∈([n]k
)and f is an injection from A to [q]}
1 2 3 . . . n 1 2 3 . . . n . . . 1 2 3 . . . n1 1 1 . . . 1 2 2 2 . . . 2 . . . q q q . . . q
(A, f ) is said to be contained in the above cycle if it consistsof k consecutive elements.
Let H be the set of (A, f ) which contained in the above cycle.Clearly |H| = nq.
Determine αp(H).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Bkn(q) = {(A, f ) : A ∈([n]k
)and f is an injection from A to [q]}
1 2 3 . . . n 1 2 3 . . . n . . . 1 2 3 . . . n1 1 1 . . . 1 2 2 2 . . . 2 . . . q q q . . . q
(A, f ) is said to be contained in the above cycle if it consistsof k consecutive elements.
Let H be the set of (A, f ) which contained in the above cycle.Clearly |H| = nq.
Determine αp(H).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Graph G [H]: V (G [H]) = H, two vertices (A, f ) and (B, g)are adjacent iff they are not intersecting.
G [H] is isomorphism to the well-known circular graph Knq:k .
Clique number ω(G ): the maximum number of pairwiseadjacent vertices in G . For S ⊆ V (G ), ω(G [S ]) written asω(S).
F of H with ν(F) = s iff ω(F) = s.
To determine αp(H) is equivalent to determine ω(G [H]).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Graph G [H]: V (G [H]) = H, two vertices (A, f ) and (B, g)are adjacent iff they are not intersecting.
G [H] is isomorphism to the well-known circular graph Knq:k .
Clique number ω(G ): the maximum number of pairwiseadjacent vertices in G . For S ⊆ V (G ), ω(G [S ]) written asω(S).
F of H with ν(F) = s iff ω(F) = s.
To determine αp(H) is equivalent to determine ω(G [H]).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Graph G [H]: V (G [H]) = H, two vertices (A, f ) and (B, g)are adjacent iff they are not intersecting.
G [H] is isomorphism to the well-known circular graph Knq:k .
Clique number ω(G ): the maximum number of pairwiseadjacent vertices in G . For S ⊆ V (G ), ω(G [S ]) written asω(S).
F of H with ν(F) = s iff ω(F) = s.
To determine αp(H) is equivalent to determine ω(G [H]).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Graph G [H]: V (G [H]) = H, two vertices (A, f ) and (B, g)are adjacent iff they are not intersecting.
G [H] is isomorphism to the well-known circular graph Knq:k .
Clique number ω(G ): the maximum number of pairwiseadjacent vertices in G . For S ⊆ V (G ), ω(G [S ]) written asω(S).
F of H with ν(F) = s iff ω(F) = s.
To determine αp(H) is equivalent to determine ω(G [H]).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Graph G [H]: V (G [H]) = H, two vertices (A, f ) and (B, g)are adjacent iff they are not intersecting.
G [H] is isomorphism to the well-known circular graph Knq:k .
Clique number ω(G ): the maximum number of pairwiseadjacent vertices in G . For S ⊆ V (G ), ω(G [S ]) written asω(S).
F of H with ν(F) = s iff ω(F) = s.
To determine αp(H) is equivalent to determine ω(G [H]).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Kn:k : vertex set [n], i ∼ j iff k ≤ |i − j | ≤ n − k .
The maximum independent sets in Kn:k are the sets of kconsecutive elements, viewed cyclically.
Lemma
Let n, s and k be three integers with n ≥ k(s + 1). For any vertexsubset F of Kn:k with ω(F ) = s, then |F | ≤ ks. Moreover, ifn > k(s + 1), the equality holds if and only if F is the union of smaximum independent sets of Kn:k .
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Kn:k : vertex set [n], i ∼ j iff k ≤ |i − j | ≤ n − k .
The maximum independent sets in Kn:k are the sets of kconsecutive elements, viewed cyclically.
Lemma
Let n, s and k be three integers with n ≥ k(s + 1). For any vertexsubset F of Kn:k with ω(F ) = s, then |F | ≤ ks. Moreover, ifn > k(s + 1), the equality holds if and only if F is the union of smaximum independent sets of Kn:k .
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Kn:k : vertex set [n], i ∼ j iff k ≤ |i − j | ≤ n − k .
The maximum independent sets in Kn:k are the sets of kconsecutive elements, viewed cyclically.
Lemma
Let n, s and k be three integers with n ≥ k(s + 1). For any vertexsubset F of Kn:k with ω(F ) = s, then |F | ≤ ks. Moreover, ifn > k(s + 1), the equality holds if and only if F is the union of smaximum independent sets of Kn:k .
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
ω(G [H]) = ks
Let F be a family of Bkn(q) with ν(F) = s. Then by the formerlemmas we have
|F||Bkn(q)|
≤ sk
nq.
Therefore,
|F| ≤ sk
nqqk
(n
k
)= sqk−1
(n − 1
k − 1
).
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
For n positive numbers p1, p2, . . . , pn with p1 ≤ p2 ≤ · · · ≤ pn, letLp be the labeled n-sets given by
Ln,p = {(i1, i2, . . . , in) : ij ∈ [pj ] for j ∈ [n]}.
Theorem
If F is a family of Ln,p with ν(F) = s ≤ p1, then
|F| ≤ sp2p3 · · · pn,
and equality holds if and only ifF = {(i1, i2, . . . , in) ∈ Ln,p : ij ∈ S} for one s-subset S of [p1] andone j of [n] with pj = p1.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Theorem
Let n and s be two positive integers with s ≤ n. If F is a family ofSn with ν(F) = s, then |F| ≤ s(n − 1)!.
Theorem
Let n and s be two positive integers with s < n. If F is a family of
Pr ,n with ν(F) = s, then |F| ≤ s(n−1r−1
)(n − 1)!
(n − r)!.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
3. Matching Numbers
Theorem
Let n and s be two positive integers with s ≤ n. If F is a family ofSn with ν(F) = s, then |F| ≤ s(n − 1)!.
Theorem
Let n and s be two positive integers with s < n. If F is a family of
Pr ,n with ν(F) = s, then |F| ≤ s(n−1r−1
)(n − 1)!
(n − r)!.
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem
Many Thanks!
Huajun Zhang Some Problems and Generalizations on Erdos-Ko-Rado Theorem