Kneser graph: Difference between revisions
(Created page with "==Definition== Suppose <matH>n</math> and <math>k</math> are positive integers. The '''Kneser graph''' <math>KG_{n,k}</math> is an undirected graph defined as follows. Fi...") |
|||
(10 intermediate revisions by the same user not shown) | |||
Line 7: | Line 7: | ||
==Particular cases== | ==Particular cases== | ||
===Classes of cases=== | |||
{| class="sortable" border="1" | {| class="sortable" border="1" | ||
Line 18: | Line 20: | ||
|- | |- | ||
| <math>k = n/2</math> || In this case, the graph is a [[matching graph]]: it is a disjoint union of 2-cliques, with one 2-clique for each partition of the <matH>n</matH>-set into two disjoint pieces of equal size. | | <math>k = n/2</math> || In this case, the graph is a [[matching graph]]: it is a disjoint union of 2-cliques, with one 2-clique for each partition of the <matH>n</matH>-set into two disjoint pieces of equal size. | ||
|- | |||
| <math>1 < k < n/2</math> || This is the interesting case. In this case, the graph is connected and non-empty, but is not a complete graph. | |||
|- | |- | ||
| <math>k = 1</math> || [[complete graph]] <math>K_n</math> | | <math>k = 1</math> || [[complete graph]] <math>K_n</math> | ||
|- | |- | ||
| <math>k = 0</math> || [[one-point graph]] | | <math>k = 0</math> || [[one-point graph]] | ||
|} | |||
===First few nontrivial cases=== | |||
As indicated above, the interesting cases are where <math>1 < k < n/2</math>. We list the first few of these: | |||
{| class="sortable" border="1" | |||
! <math>n</math> !! <math>k</math> !! Kneser graph <math>KG_{n,k}</math> | |||
|- | |||
| 5 || 2 || [[Petersen graph]] | |||
|- | |||
| 6 || 2 || {{fillin}} | |||
|} | |||
===Odd graph case=== | |||
A particular case of Kneser graphs of interest is where <math>n = 2k + 1</math>. This gives a one-parameter family of graphs called [[odd graph]]s. Specifically, the odd graph <math>O_m</math> is defined as the Kneser graph with <math>n = 2m - 1</math> and <math>k = m - 1</math>. | |||
==Arithmetic functions== | |||
===Size measures=== | |||
{| class="sortable" border="1" | |||
! Function !! Value !! Explanation | |||
|- | |||
| [[size of vertex set]] || <math>\binom{n}{k} = \frac{n!}{k!(n - k)!}</math> || By definition, it is the number of <math>k</math>-element subsets of a fixed <math>n</math>-element set. | |||
|- | |||
| [[size of edge set]] || <math>\frac{1}{2} \binom{n}{k,k,n-2k} = \frac{1}{2} \frac{n!}{k!k! (n - 2k)!}</math> || This is the number of ways of dividing a set of size <math>n</math> into two interchangeable substs of size <math>k</math> and a left-over set of size <math>n - 2k</math>. | |||
|} | |||
===Numerical invariants associated with vertices=== | |||
We restrict attention to the case that <math>1 \le k < n/2</math> for the expressions below. | |||
Since the graph is a [[vertex-transitive graph]], any numerical invariant associated to a vertex must be equal on all vertices of the graph. Below are listed some of these invariants: | |||
{| class="sortable" border="1" | |||
! Function !! Value !! Explanation | |||
|- | |||
| [[degree of a vertex]] || <math>\binom{n - k}{k}</math> || We need to determine, for a given <math>k</math>-element subset, the number of <math>k</math>-element subsets disjoint from it. This is the same as the number of <math>k</math>-element subsets of its set-theoretic complement, which has size <math>n - k</math>. | |||
|- | |||
| [[eccentricity of a vertex]] || <math>1 + \left\lceil\frac{k - 1}{n - 2k}\right\rceil</math> where <math>\lceil \rceil</math> denotes the ceiling function: the smallest integer greater than or equal to a given number || | |||
|} | |||
===Other numerical invariants=== | |||
We restrict attention to the case <math>1 \le k < n/2</math> for all the formulas below, though some of the formulas are also more generally valid. | |||
{| class="sortable" border="1" | |||
! Function !! Value !! Explanation | |||
|- | |||
| [[clique number]] || <math>\left\lfloor\frac{n}{k}\right\rfloor</math> where <math>\lfloor \rfloor</math> is the floor function (the greatest integer function) that returns the largest integer less than or equal to the number || For a subset of the vertex set to form a clique, the elements of the subset must be disjoint as subsets of the <math>n</math>-element set. If there are <math>r</math> such subsets, they have a total of <math>kr</math> elements, forcing <math>kr \le n</math>, giving the upper bound. It's also easy to see that this upper bound is achieved. | |||
|- | |||
| [[independence number]] || {{fillin}} || | |||
|- | |||
| [[chromatic number]] || <math>n - 2k + 2</math> if <math>k \le n/2</math><br>1 otherwise || | |||
|- | |||
| [[radius of a graph]] || <math>1 + \left\lceil\frac{k - 1}{n - 2k}\right\rceil</math> where <math>\lceil \rceil</math> denotes the ceiling function: the smallest integer greater than or equal to a given number || Due to vertex-transitivity, the radius equals the eccentricity of any vertex, which has been computed above. | |||
|- | |||
| [[diameter of a graph]] || <math>1 + \left\lceil\frac{k - 1}{n - 2k}\right\rceil</math> where <math>\lceil \rceil</math> denotes the ceiling function: the smallest integer greater than or equal to a given number|| Due to vertex-transitivity, the diameter equals the eccentricity of any vertex, which has been computed above. | |||
|- | |||
| [[odd girth]] || <math>1 + 2\left\lceil\frac{k}{n - 2k}\right \rceil</math> where <math>\lceil \rceil</math> denotes the ceiling function: the smallest integer greater than or equal to a given number || | |||
|- | |||
| [[even girth]] || 6 if <math>n = 2k + 1</math><br>4 if <math>n \ge 2k + 2</math> || | |||
|} | |} |
Latest revision as of 22:11, 29 May 2012
Definition
Suppose and are positive integers. The Kneser graph is an undirected graph defined as follows. Fix a set of size (we usually take the set as for convenience). Then:
- The vertex set of is the collection of -element subsets of the fixed set of size .
- The edge set of is defined as follows: two vertices of are adjacent if and only if they are disjoint when viewed as subsets of the -element set.
Particular cases
Classes of cases
Condition on and | Conclusion |
---|---|
The vertex set is empty, so the graph is a graph on no vertices | |
one-point graph | |
The vertex set is non-empty, but the edge set is empty, so the graph is an empty graph | |
In this case, the graph is a matching graph: it is a disjoint union of 2-cliques, with one 2-clique for each partition of the -set into two disjoint pieces of equal size. | |
This is the interesting case. In this case, the graph is connected and non-empty, but is not a complete graph. | |
complete graph | |
one-point graph |
First few nontrivial cases
As indicated above, the interesting cases are where . We list the first few of these:
Kneser graph | ||
---|---|---|
5 | 2 | Petersen graph |
6 | 2 | Fill this in later |
Odd graph case
A particular case of Kneser graphs of interest is where . This gives a one-parameter family of graphs called odd graphs. Specifically, the odd graph is defined as the Kneser graph with and .
Arithmetic functions
Size measures
Function | Value | Explanation |
---|---|---|
size of vertex set | By definition, it is the number of -element subsets of a fixed -element set. | |
size of edge set | This is the number of ways of dividing a set of size into two interchangeable substs of size and a left-over set of size . |
Numerical invariants associated with vertices
We restrict attention to the case that for the expressions below.
Since the graph is a vertex-transitive graph, any numerical invariant associated to a vertex must be equal on all vertices of the graph. Below are listed some of these invariants:
Function | Value | Explanation |
---|---|---|
degree of a vertex | We need to determine, for a given -element subset, the number of -element subsets disjoint from it. This is the same as the number of -element subsets of its set-theoretic complement, which has size . | |
eccentricity of a vertex | where denotes the ceiling function: the smallest integer greater than or equal to a given number |
Other numerical invariants
We restrict attention to the case for all the formulas below, though some of the formulas are also more generally valid.
Function | Value | Explanation |
---|---|---|
clique number | where is the floor function (the greatest integer function) that returns the largest integer less than or equal to the number | For a subset of the vertex set to form a clique, the elements of the subset must be disjoint as subsets of the -element set. If there are such subsets, they have a total of elements, forcing , giving the upper bound. It's also easy to see that this upper bound is achieved. |
independence number | Fill this in later | |
chromatic number | if 1 otherwise |
|
radius of a graph | where denotes the ceiling function: the smallest integer greater than or equal to a given number | Due to vertex-transitivity, the radius equals the eccentricity of any vertex, which has been computed above. |
diameter of a graph | where denotes the ceiling function: the smallest integer greater than or equal to a given number | Due to vertex-transitivity, the diameter equals the eccentricity of any vertex, which has been computed above. |
odd girth | where denotes the ceiling function: the smallest integer greater than or equal to a given number | |
even girth | 6 if 4 if |