Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Basic conceptsInequalitiesConvexity
On the Convexity of the Tutte Polynomial of aPaving Matroid Along Line Segments
Laura Chavez Lomelı† C.Merino‡ Steven D. Noble§
Marcelino Ramırez-Ibanez‡
†UAM-Azcapotzalco
§Departament of Mathematical SciencesBrunel University
‡Instituto de Matematicas UNAM,sede Oaxaca
ACCOTA 2010
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Summary
1 Basic conceptsMatroidsMatroid constructionsTutte Polynomial
2 InequalitiesMerino-Welsh conjectureFirst result
3 ConvexityMain resultPaving Matroidstechnical resultsProof
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Definition
A matroid M is a pair (E , r), where E is a finite set andr : ℘(E ) → N, the rank function, is such that
1 0 ≤ r(A) ≤ |A|, for all A ⊆ E .
2 If A ⊆ B, then r(A) ≤ r(B)
3 For all A,B ⊆ E ,
r(A ∪ B) + r(A ∩ B) ≤ r(A) + r(B) (submodularity)
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Definition
A matroid M is a pair (E , r), where E is a finite set andr : ℘(E ) → N, the rank function, is such that
1 0 ≤ r(A) ≤ |A|, for all A ⊆ E .
2 If A ⊆ B, then r(A) ≤ r(B)
3 For all A,B ⊆ E ,
r(A ∪ B) + r(A ∩ B) ≤ r(A) + r(B) (submodularity)
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Definition
A matroid M is a pair (E , r), where E is a finite set andr : ℘(E ) → N, the rank function, is such that
1 0 ≤ r(A) ≤ |A|, for all A ⊆ E .
2 If A ⊆ B, then r(A) ≤ r(B)
3 For all A,B ⊆ E ,
r(A ∪ B) + r(A ∩ B) ≤ r(A) + r(B) (submodularity)
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Bases and circuits
Let M = (E , r) be a matroid,
A ⊆ E is an independent set if |A| = r(A).
B ⊆ E is a basis if B is a maximal independent set.
C ⊆ E is a circuit if it is a minimal dependent set.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Bases and circuits
Let M = (E , r) be a matroid,
A ⊆ E is an independent set if |A| = r(A).
B ⊆ E is a basis if B is a maximal independent set.
C ⊆ E is a circuit if it is a minimal dependent set.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Bases and circuits
Let M = (E , r) be a matroid,
A ⊆ E is an independent set if |A| = r(A).
B ⊆ E is a basis if B is a maximal independent set.
C ⊆ E is a circuit if it is a minimal dependent set.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Bases and circuits
Let M = (E , r) be a matroid,
A ⊆ E is an independent set if |A| = r(A).
B ⊆ E is a basis if B is a maximal independent set.
C ⊆ E is a circuit if it is a minimal dependent set.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Example: Graphic matroids
For a connected graph G = (V ,E ) we define the matroidM(G ) = (E , r), where r(A) is the size of a maximal spanningforest in (V ,A).
An independent set in M(G ) is just a spanning forest of G .
A base is a spanning tree of G .
A circuit is a cycle of G .
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Example: Graphic matroids
For a connected graph G = (V ,E ) we define the matroidM(G ) = (E , r), where r(A) is the size of a maximal spanningforest in (V ,A).
An independent set in M(G ) is just a spanning forest of G .
A base is a spanning tree of G .
A circuit is a cycle of G .
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Example: Graphic matroids
For a connected graph G = (V ,E ) we define the matroidM(G ) = (E , r), where r(A) is the size of a maximal spanningforest in (V ,A).
An independent set in M(G ) is just a spanning forest of G .
A base is a spanning tree of G .
A circuit is a cycle of G .
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Example: Graphic matroids
For a connected graph G = (V ,E ) we define the matroidM(G ) = (E , r), where r(A) is the size of a maximal spanningforest in (V ,A).
An independent set in M(G ) is just a spanning forest of G .
A base is a spanning tree of G .
A circuit is a cycle of G .
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Example: Graphic matroids
For a connected graph G = (V ,E ) we define the matroidM(G ) = (E , r), where r(A) is the size of a maximal spanningforest in (V ,A).
An independent set in M(G ) is just a spanning forest of G .
A base is a spanning tree of G .
A circuit is a cycle of G .
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Example: Graphic matroids
For a connected graph G = (V ,E ) we define the matroidM(G ) = (E , r), where r(A) is the size of a maximal spanningforest in (V ,A).
An independent set in M(G ) is just a spanning forest of G .
A base is a spanning tree of G .
A circuit is a cycle of G .
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Example: Uniform matroids
Um,n = ([n], r), where [n] = {1, 2 . . . , n} and the rank function isgiven by
r(A) = min{|A|,m}, whereA ⊆ [n].
An independent set is a any set with cardinality at most m. Abasis is a set of cardinality m and a circuit is any set of cardinalitym + 1.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Example: Uniform matroids
1 2 3 4 5
12 45
E U3,5
∅
123 345
Um,n = ([n], r), where [n] = {1, 2 . . . , n} and the rank function isgiven by
r(A) = min{|A|,m}, whereA ⊆ [n].
An independent set is a any set with cardinality at most m. Abasis is a set of cardinality m and a circuit is any set of cardinalitym + 1.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Example: Uniform matroids
1 2 3 4 5
12 45
E U3,5
∅
123 345
Um,n = ([n], r), where [n] = {1, 2 . . . , n} and the rank function isgiven by
r(A) = min{|A|,m}, whereA ⊆ [n].
An independent set is a any set with cardinality at most m.
Abasis is a set of cardinality m and a circuit is any set of cardinalitym + 1.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Example: Uniform matroids
1 2 3 4 5
12 45
E U3,5
∅
123 345
Um,n = ([n], r), where [n] = {1, 2 . . . , n} and the rank function isgiven by
r(A) = min{|A|,m}, whereA ⊆ [n].
An independent set is a any set with cardinality at most m. Abasis is a set of cardinality m
and a circuit is any set of cardinalitym + 1.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Example: Uniform matroids
1 2 3 4 5
12 45
E U3,5
∅
123 345
Um,n = ([n], r), where [n] = {1, 2 . . . , n} and the rank function isgiven by
r(A) = min{|A|,m}, whereA ⊆ [n].
An independent set is a any set with cardinality at most m. Abasis is a set of cardinality m and a circuit is any set of cardinalitym + 1.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Duality
If M = (E , r) is a matroid, then we can define its dual matroidM∗ = (E , r∗) where r∗(A) = |A| − r(E ) + r(E − A).
Figure: M(G ) and its dual M(G )∗
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Duality
If M = (E , r) is a matroid, then we can define its dual matroidM∗ = (E , r∗) where r∗(A) = |A| − r(E ) + r(E − A).
Figure: M(G ) and its dual M(G )∗
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Deletion-contraction, loops and coloops
Let M = (E , r) be a matroid and e ∈ E . The matroid obtained bydeleting e is defined by M\e=(E\e, r |E\e). The matroid obtainedby contracting e is defined by M/e=(M∗\e)∗.
An element e ∈ E with rank r(e) = 0 is called a loop, and duallyan element with r∗(e) = 0 is called an isthmus or coloop.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Deletion-contraction, loops and coloops
Let M = (E , r) be a matroid and e ∈ E . The matroid obtained bydeleting e is defined by M\e=(E\e, r |E\e). The matroid obtainedby contracting e is defined by M/e=(M∗\e)∗.An element e ∈ E with rank r(e) = 0 is called a loop, and duallyan element with r∗(e) = 0 is called an isthmus or coloop.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Definition
For a matroid M, the 2-variable polynomial
TM(x , y) =∑A⊆E
(x − 1)r(E)−r(A)(y − 1)|A|−r(A), (1)
is called the Tutte polynomial. It is easy to prove that
TM(1, 1) equals the number of bases of M.
TM(2, 2) = 2|E |.
TM∗(x , y) = TM(y , x).
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Definition
For a matroid M, the 2-variable polynomial
TM(x , y) =∑A⊆E
(x − 1)r(E)−r(A)(y − 1)|A|−r(A), (1)
is called the Tutte polynomial. It is easy to prove that
TM(1, 1) equals the number of bases of M.
TM(2, 2) = 2|E |.
TM∗(x , y) = TM(y , x).
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
MatroidsMatroid constructionsTutte Polynomial
Linear recursion definition
If M is a matroid and e is an element that is not a loop or anisthmus, then
TM(x , y) = TM\e(x , y) + TM/e(x , y). (2)
If no such an element e exists, TM(x , y) = x iy j where i and j arethe number of isthmuses and loops respectively.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Merino-Welsh conjectureFirst result
Old Conjecture
C.Merino y D.J.A. Welsh proposed the following in 1999
Conjecture
If G is a 2-connected graph with no loops, then
max{TG (2, 0),TG (0, 2)} ≥ TG (1, 1).
TG (2, 0) equals the number of acyclic orientations while TG (0, 2)equals the number of totally cyclic orientations.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Merino-Welsh conjectureFirst result
Old Conjecture
C.Merino y D.J.A. Welsh proposed the following in 1999
Conjecture
If G is a 2-connected graph with no loops, then
max{TG (2, 0),TG (0, 2)} ≥ TG (1, 1).
TG (2, 0) equals the number of acyclic orientations while TG (0, 2)equals the number of totally cyclic orientations.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Merino-Welsh conjectureFirst result
Old Conjecture
C.Merino y D.J.A. Welsh proposed the following in 1999
Conjecture
If G is a 2-connected graph with no loops, then
max{TG (2, 0),TG (0, 2)} ≥ TG (1, 1).
TG (2, 0) equals the number of acyclic orientations while TG (0, 2)equals the number of totally cyclic orientations.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Merino-Welsh conjectureFirst result
(0, 2)
(2, 0)
(1, 1)
TG(x, y)
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Merino-Welsh conjectureFirst result
First Result
Theorem
If a matroid M contains two disjoint bases, then
TM(0, 2a) ≥ TM(a, a), for all a ≥ 2.
Dually, if its ground set is the union of two bases, then
TM(2a, 0) ≥ TM(a, a), for all a ≥ 2.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Merino-Welsh conjectureFirst result
TM(x, y)
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Merino-Welsh conjectureFirst result
TM(x, y)
(0, 2a)
(a, a)
(2a, 0)
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Main theorem
Theorem
If M is a coloopless paving matroid, then TM(x , y) = T satisfies
tT (x1, y1)+(1−t)T (x2, y2) ≥ T (tx1+(1−t)x2, ty1+(1−t)y2), (3)
for 0 ≤ t ≤ 1 and x1 + y1 = x2 + y2 = p; where p ≥ 0 and0 ≤ x1, x2, y1, y2.
That is, TM is convex along the portion of the line x + y = p lyingin the positive quadrant.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Main theorem
Theorem
If M is a coloopless paving matroid, then TM(x , y) = T satisfies
tT (x1, y1)+(1−t)T (x2, y2) ≥ T (tx1+(1−t)x2, ty1+(1−t)y2), (3)
for 0 ≤ t ≤ 1 and x1 + y1 = x2 + y2 = p; where p ≥ 0 and0 ≤ x1, x2, y1, y2.
That is, TM is convex along the portion of the line x + y = p lyingin the positive quadrant.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
(x1, y1)
(x2, y2)
x + y = p
TM(x, y)
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Definition
A paving matroid M = (E, r) is a matroid whose circuits all havesize at least r(E ).
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Definition
A paving matroid M = (E, r) is a matroid whose circuits all havesize at least r(E ).
r(Mr(M)r(Mr(M)+1)+1
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Definition
A paving matroid M = (E, r) is a matroid whose circuits all havesize at least r(E ).
Note
Paving matroids are closed under minors
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Paving matroids and disjoint bases
Theorem
Let M = (E , r) be a rank-r paving matroid with n elements,
if 2r > n, then E is the union of two bases,
if 2r ≤ n and M is coloopless, then M contains two disjointbases.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Paving matroids and disjoint bases
Theorem
Let M = (E , r) be a rank-r paving matroid with n elements,
if 2r > n, then E is the union of two bases,
if 2r ≤ n and M is coloopless, then M contains two disjointbases.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Proof.
case: 2r > n
. . .
...
01
r − 1r
E
∅
...
n
2 . . .
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Proof.
case: 2r > n
. . .
...
01
r − 1r
E
∅
...
n
B
2 . . .
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Proof.
case: 2r > n
. . .
...
01
r − 1r
E
∅
...
n
B
I = E −B
2 . . .
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Proof.
case: 2r > n
. . .
...
01
r − 1r
E
∅
...
n
B
I = E −B
I ⊂ B′
2 . . .
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Duality and convexity
Lemma
Let M be a matroid. Either, both TM(x , y) and TM∗(x , y) areconvex along the portion of the line x + y = p lying in the positivequadrant or neither is.
Proof.
This follows directly from the equality TM(x , y) = TM∗(y , x).
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Deletion-contraction and convexity
Lemma
Let M be a matroid and e in M be neither a loop nor a coloop. IfTM\e and TM/e are both convex along the portion of the linex + y = p lying in the positive quadrant, then TM is also convexon the same domain.
Proof.
This follows directly from linear recursive definition.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Base case
Lemma
If M is isomorphic to the paving matroid U1,k+1 ⊕ U0,l , wherel ≥ 0 and k ≥ 1, then TM is convex along the portion of the linex + y = p lying in the positive quadrant.
Proof.
TM(x , y) = py l +l+k∑
m=l+2
ym.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Base case
Lemma
If M is isomorphic to the paving matroid U1,k+1 ⊕ U0,l , wherel ≥ 0 and k ≥ 1, then TM is convex along the portion of the linex + y = p lying in the positive quadrant.
Proof.
TM(x , y) = py l +l+k∑
m=l+2
ym.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Uniform matroids and convexity
Theorem
The Tutte polynomial TM is a convex function in the positivequadrant when M is a uniform matroid. In particular, TM isconvex along the portion of the line x + y = p lying in the positivequadrant.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Proof.
The Tutte polynomial of Ur ,n can be computed using (1)
TUr,n(x , y) =r−1∑i=0
(n
i
)(x − 1)r−i +
(n
r
)+
n∑i=r+1
(n
i
)(y − 1)i−r ,
This can be expanded into the following expression,
TUr,n(x , y) =n−r∑j=1
(n − j − 1
r − 1
)y j +
r∑i=1
(n − i − 1
n − r − 1
)x i .
TUn,n(x , y) = xn y TU0,n(x , y) = yn.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Proof.
The Tutte polynomial of Ur ,n can be computed using (1)
TUr,n(x , y) =r−1∑i=0
(n
i
)(x − 1)r−i +
(n
r
)+
n∑i=r+1
(n
i
)(y − 1)i−r ,
This can be expanded into the following expression,
TUr,n(x , y) =n−r∑j=1
(n − j − 1
r − 1
)y j +
r∑i=1
(n − i − 1
n − r − 1
)x i .
TUn,n(x , y) = xn y TU0,n(x , y) = yn.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Proof.
The Tutte polynomial of Ur ,n can be computed using (1)
TUr,n(x , y) =r−1∑i=0
(n
i
)(x − 1)r−i +
(n
r
)+
n∑i=r+1
(n
i
)(y − 1)i−r ,
This can be expanded into the following expression,
TUr,n(x , y) =n−r∑j=1
(n − j − 1
r − 1
)y j +
r∑i=1
(n − i − 1
n − r − 1
)x i .
TUn,n(x , y) = xn y TU0,n(x , y) = yn.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
rank-2 case
Theorem
If M is a rank-2 loopless and coloopless matroid, then TM isconvex along the portion of the line x + y = p lying in the positivequadrant.
Proof.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
rank-2 case
Theorem
If M is a rank-2 loopless and coloopless matroid, then TM isconvex along the portion of the line x + y = p lying in the positivequadrant.
Proof.
If M ∼= U2,n, the result follows from applying the previous lemma.Else, M is isomorphic to a matroid with parallel elements whosesimplification is isomorphic to U2,n.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
rank-2 case
Theorem
If M is a rank-2 loopless and coloopless matroid, then TM isconvex along the portion of the line x + y = p lying in the positivequadrant.
Proof.
If n ≥ 3 or there is a parallel class of size at least 3, there existse ∈ M such that M/e ∼= U1,k+1⊕U0,l and M\e is a rank-2 looplessand coloopless matroid. The geometric representation of M is
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
rank-2 case
Theorem
If M is a rank-2 loopless and coloopless matroid, then TM isconvex along the portion of the line x + y = p lying in the positivequadrant.
Proof.
Otherwise, the simplification of M is isomorphic to U2,2 and everyelement is in a parallel class of size 2. Then M ∼= U1,2 ⊕ U1,2 andTM = (x +y)2 which is convex (in fact is constant) along x +y = pfor p > 0 and 0 ≤ y ≤ p.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Structural result
In order to establish or main result, we need the followingstructural result about coloopless paving matroids.
Lemma
Let M be a rank-r coloopless paving matroid. If for every elemente of M, M \ e has a coloop, then one of the following three caseshappens.
1 M is isomorphic to Ur ,r+1.
2 M is the 2-stretching of a uniform matroid Us,s+2, for somes ≥ 1.
3 M is isomorphic to U1,2 ⊕ U1,2.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Structural result
Lemma
Let M be a rank-r coloopless paving matroid. If for every elemente of M, M \ e has a coloop, then TM is convex along the portionof the line x + y = p lying in the positive quadrant.
Proof.
We jus check the 4 cases of the previos theorem. The onlyinteresting one is when M is the 2-stretching of Us,s+2. In thiscase we use that M∗ is the 2-thickening of U2,n.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Proof of main theorem
Theorem
If M is a coloopless paving matroid, then TM is convex along theportion of the line x + y = p lying in the positive quadrant.
Proof.
M has a loop, M ∼= U1,k+1 ⊕ U0,l . (base case)
Otherwise, every element of M is neither a loop nor a coloop,
a) If there is an element e such that M \ e has no coloop, thenboth M/e and M \ e are coloopless paving matroids.(Induction)
b) So, we can assume that for all e, M \ e has acoloop.(Structural result)
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Proof of main theorem
Theorem
If M is a coloopless paving matroid, then TM is convex along theportion of the line x + y = p lying in the positive quadrant.
Proof.
M has a loop, M ∼= U1,k+1 ⊕ U0,l . (base case)
Otherwise, every element of M is neither a loop nor a coloop,
a) If there is an element e such that M \ e has no coloop, thenboth M/e and M \ e are coloopless paving matroids.(Induction)
b) So, we can assume that for all e, M \ e has acoloop.(Structural result)
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Proof of main theorem
Theorem
If M is a coloopless paving matroid, then TM is convex along theportion of the line x + y = p lying in the positive quadrant.
Proof.
M has a loop, M ∼= U1,k+1 ⊕ U0,l . (base case)
Otherwise, every element of M is neither a loop nor a coloop,
a) If there is an element e such that M \ e has no coloop, thenboth M/e and M \ e are coloopless paving matroids.(Induction)
b) So, we can assume that for all e, M \ e has acoloop.(Structural result)
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Proof of main theorem
Theorem
If M is a coloopless paving matroid, then TM is convex along theportion of the line x + y = p lying in the positive quadrant.
Proof.
M has a loop, M ∼= U1,k+1 ⊕ U0,l . (base case)
Otherwise, every element of M is neither a loop nor a coloop,
a) If there is an element e such that M \ e has no coloop, thenboth M/e and M \ e are coloopless paving matroids.(Induction)
b) So, we can assume that for all e, M \ e has acoloop.(Structural result)
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Proof of main theorem
Theorem
If M is a coloopless paving matroid, then TM is convex along theportion of the line x + y = p lying in the positive quadrant.
Proof.
M has a loop, M ∼= U1,k+1 ⊕ U0,l . (base case)
Otherwise, every element of M is neither a loop nor a coloop,
a) If there is an element e such that M \ e has no coloop, thenboth M/e and M \ e are coloopless paving matroids.(Induction)
b) So, we can assume that for all e, M \ e has acoloop.(Structural result)
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Corollary
If M or M∗ is a coloopless paving matroid, then TM is convex alongthe portion of the line x + y = p lying in the positive quadrant.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
The proportion of paving matroids
1976 Dominic Welsh ask if most matroids are paving, he used thetable of matroids of up to 8 elements of Blackburn, Crapo yHiggs from 1973.
2008 This table has been update recently up to 9 elements forMayhew and Royle.
2010 Mayhew, Newman, Welsh y Whittle pose as a conjecture thatasymptotically almost every matroid is paving.
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Our conjecture
Conjecture
If M contains two disjoint bases or its ground set is the union oftwo bases then
max{TM(2, 0),TM(0, 2)} ≥ TM(1, 1).
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Thanks
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid
Basic conceptsInequalitiesConvexity
Main resultPaving Matroidstechnical resultsProof
Gracias
Criel Merino Convexity of the Tutte Polynomial of a Paving Matroid