GENUS DISTRIBUTIONS OF CIRCULAR LADDERS:

A WEBSITE SUPPLEMENT
YICHAO CHEN, JONATHAN L. GROSS, AND TOUFIK MANSOUR
Abstract. The genus distribution of the circular ladder CLn is re-derived here, using overlap
matrices and Chebyshev polynomials.
1. The genus distributions of circular ladders
We dene two sets of matrices and two polynomials:
(1) A
n+2
as the set of all matrices over Z
2
of the form M
c,x,y,X,Y,Z
n+2
;
(2) A
n+2
(z) =

n+2
j=0
A
n+2
(j)z
j
as the rank-distribution polynomial of the set A
n+2
, that
is, of the overlap matrices for general rotation systems for the circular ladder CL
n
;
(3) B
n+2
as the set of all matrices of the form M
c,x,y,0,Y,Z
n+2
; and
(4) B
n+2
(z) =

n+2
j=0
B
n+2
(j)z
j
as the rank-distribution polynomial of the set B
n+2
, that
is, of the overlap matrices for pure rotation systems for the circular ladder CL
n
.
In a matrix of the form M
c,x,y,X,Y,Z
n+2
, suppose that we rst add the second row to the rst row
and next add the second column to the rst column. Without changing the rank of the matrix,
this produces a matrix of the following form:
_
_
_
_
_
_
_
_
_
_
_
_
_
_
x
e
c +x
f
x 0 0 0 y
c +x
f
x
f
z
1
z
2
z
n2
z
n1
z
n
x z
1
x
1
y
1
0 z
2
y
1
x
2
.
.
. 0
.
.
.
.
.
.
.
.
.
.
.
.
0 z
n2
0 y
n2
0 z
n1
0 y
n2
x
n1
y
n1
y z
n
y
n1
x
n
_
_
_
_
_
_
_
_
_
_
_
_
_
_
For xy {00, 01, 10, 11}, we dene two more sets and two more polynomials:
(1) A
xy
n+2
as the set of all matrices over Z
2
of the form M
c,x,y,X,Y,Z
n+2
;
(2) A
xy
n+2
(z) =

n+2
j=0
A
xy
n+2
(j)z
j
as the rank-distribution polynomial of the set A
xy
n+2
;
(3) B
xy
n+2
as the set of all matrices of the form M
c,x,y,0,Y,Z
n+2
; and
(4) B
xy
n+2
(z) =

n+2
j=0
B
xy
n+2
(j)z
j
as the rank-distribution polynomial of the set B
xy
n+2
.
2000 Mathematics Subject Classication. Primary: 05C10; Secondary: 30B70, 42C05.
Key words and phrases. graph embedding; total embedding distribution; circular ladders; overlap matrix;
Chebyshev polynomials.
1
2 YICHAO CHEN, JONATHAN L. GROSS, AND TOUFIK MANSOUR
Clearly, we have the following property.
Property 1.1. For all n 1,
A
n+2
(z) =

xy=00,01,10,11
A
xy
n+2
(z) and B
n+2
(z) =

xy=00,01,10,11
B
xy
n+2
(z).
For the vectors X = (x
0
, x
1
, . . . , x
n
), Y = (y
1
, y
2
, . . . , y
n1
), and Z = (z
1
, z
2
, . . . , z
n
), with
x
i
, y
j
, z
k
Z
2
, we dene the matrices
(1) M
X,Y
n
=
_
_
_
_
_
_
_
_
_
x
1
y
1
y
1
x
2
y
2
0
y
2
x
3
y
3
.
.
.
.
.
.
.
.
.
0 y
n2
x
n1
y
n1
y
n1
x
n
_
_
_
_
_
_
_
_
_
and
(2) M
X,Y,Z
n+1
=
_
_
_
_
_
_
_
_
_
_
_
_
x
0
z
1
z
2
z
3
. . . z
n1
z
n
z
1
x
1
y
1
z
2
y
1
x
2
y
2
0
z
3
y
2
x
3
.
.
.
.
.
.
.
.
.
.
.
. y
n2
z
n1
0 y
n2
x
n1
y
n1
z
n
y
n1
x
n
_
_
_
_
_
_
_
_
_
_
_
_
.
As described by [1] and [3], every overlap matrix of the closed-end ladder L
n1
has the form
M
X,Y
n+1
, and every overlap matrix of the Ringel ladder R
n1
has the form M
X,Y,Z
n+1
. (Note that
the subscripts of R
n1
and M
X,Y,Z
n+1
dier by two.) Now we further dene
(1) P
n
as the set of all matrices over Z
2
of the form M
0,Y,Z
n
;
(2) C
n
(j) as the number of overlap matrices for the Ringel ladder R
n
that are of rank j;
(3) P
n
(z) =

n+1
j=0
C
n
(j)z
j
as the rank-distribution polynomial of the set P
n
; and
(4) O
n
(z) as the rank-distribution polynomial of the overlap matrices of the closed-end
ladder L
n1
over the set of matrices of the form M
0,Y
n
.
Theorem 1.2. (see [2]) The rank-distribution polynomial O
n
(z) of the overlap matrix for the
closed-end ladder L
n1
satises the recurrence relation
O
n
(z) = O
n1
(z) + 2z
2
O
n2
(z)
with the initial conditions O
1
(z) = 1 and O
2
(z) = z
2
+ 1. Moreover, the generating function
O(t; z) =

n1
O
n
(z)t
n
is given by
O(t; z) =
t +z
2
t
2
1 t 2z
2
t
2
.
GENUS DISTRIBUTIONS OF CIRCULAR LADDERS: A WEBSITE SUPPLEMENT 3
Theorem 1.3. [3] Let O
n1
(z) be the rank-distribution polynomial of the closed-end ladder
L
n2
. Then the polynomial P
n
(z) (n 3) satises the recurrence relation
P
n+1
(z) = P
n
(z) + 8z
2
P
n1
(z) + 2
n1
z
2
O
n1
(z). (3)
with the initial condition P
2
(z) = z
2
+1, P
3
(z) = 7z
2
+1 and P
4
(z) = 12z
4
+19z
2
+1. Moreover,
the generating function P(t; z) =

n2
P
n
(z)t
n
is given by
P(t; z) =
t
2
(1 +z
2
2(1 2z
2
)t 4z
2
(4 +z
2
)t
2
32z
4
t
3
)
(1 2t 8z
2
t
2
)(1 t 8z
2
t
2
)
.
1.1. Rank-distribution polynomial for circular ladders. Let M
O,Y,Zeven
n+1
be the number
of matrices such that the number of elements of Z equals to 1 is even and M
O,Y,Z
odd
n+1
be the
number of matrices such that the number of elements of Z equals to 1 is odd. Let P
O
n+1
(z) be
the rank distribution polynomial over the set M
O,Y,Z
odd
n+1
and P
E
n+1
(z) be the rank distribution
polynomial over the set M
O,Y,Zeven
n+1
. By the denition, we have P
n
(z) = P
O
n
(z) +P
E
n
(z).
Lemma 1.4. The polynomial P
E
n
(z) (n 4) satises the recurrence relation
P
E
n+1
(z) = P
E
n
(z) + 4z
2
P
n1
(z) + 2
n2
z
2
O
n1
(z).
with the initial condition P
E
3
(z) = 3z
2
+1 and P
E
4
(z) = 6z
4
+9z
2
+1, where O
n1
(z) and P
n
(z)
are the rank-distribution polynomials of the closed-end ladder L
n2
and the Ringel ladder R
n2
,
respectively.
Proof. The proof has two cases.
Case 1: y
n1
= 0. If z
n
= 0, the matrix contributes a term P
E
n
(z); otherwise, if z
n
= 1,
it contributes a term 2
n2
z
2
O
n1
(z).
Case 2: y
n1
= 1. If z
n
= 0, we add the n+1-st column and row to the rst and n1-st
column and row, which transforms the above matrix (2) to the following form:
_
_
_
_
_
_
_
_
_
_
_
_
0 z
1
z
2
z
n2
0 0
z
1
0 y
1
z
2
y
1
0
.
.
. 0
.
.
.
.
.
.
.
.
.
z
n2
0 0 0 1
0 1 0
_
_
_
_
_
_
_
_
_
_
_
_
,
Depending whether z
n1
= 0 or 1, this matrix contributes either 2z
2
P
E
n1
(z) or 2z
2
P
O
n1
(z)
to P
E
n+1
(z). Otherwise z
n
= 1, and we add the last column and row to the rst column
4 YICHAO CHEN, JONATHAN L. GROSS, AND TOUFIK MANSOUR
and row, and thereby obtain the following matrix.
_
_
_
_
_
_
_
_
_
_
_
_
0 z
1
z
2
z
n2
0 1
z
1
0 y
1
z
2
y
1
0
.
.
. 0
.
.
.
.
.
.
.
.
.
z
n2
0 y
n2
0 0 y
n2
0 1
1 1 0
_
_
_
_
_
_
_
_
_
_
_
_
,
(1) y
n2
= 0. In this subcase, according to the values of z
n1
, it contributes to P
E
n+1
(z)
by a term z
2
P
O
n1
(z) or z
2
P
E
n1
(z)
(2) y
n2
= 1. We add the last row to the n 1-st row in this subcase. Depending on
the value of z
n1
, it contributes either z
2
P
O
n1
(z) or z
2
P
E
n1
(z):
_
_
_
_
_
_
_
_
_
_
_
_
0 z
1
z
2
z
n2
+ 1 0 1
z
1
0 y
1
z
2
y
1
0
.
.
. 0
.
.
.
.
.
.
.
.
.
z
n2
+ 1
0 0 0 1
1 1 0
_
_
_
_
_
_
_
_
_
_
_
_
,

Proposition 1.5. For all n 3,

P
E
n
(z) =
1
2
(P
n
(z) + 1 z
2
).
Moreover, the generating function P
E
(t; z) =

n3
P
E
n
(z)t
n
is given by
t
3
(1 + 3z
2
3(1 z
2
)(1 + 2z
2
)t + 2(1 10z
2
13z
4
)t
2
+ 8z
2
(3 2z
2
4z
4
)t
3
+ 64z
4
t
4
)
(1 2t 8z
2
t
2
)(1 t 8z
2
t
2
)(1 t)
.
Proof. By Lemma 1.4 and together with induction on n, we have
P
E
n
(z) = P
E
4
(z) +
n2

j=3
(4z
2
P
j
(z) + 2
j1
z
2
O
j
(z)).
Thus, by (3) we obtain
P
E
n
(z) = P
E
4
(z) +
1
2
n2

j=3
(P
j+2
(z) P
j+1
(z)) = P
E
4
(z) +
1
2
(P
n
(z) P
4
(z))
=
1
2
(P
n
(z) + 1 z
2
),
as claimed. The rest follows immediately from Theorem 1.3.
GENUS DISTRIBUTIONS OF CIRCULAR LADDERS: A WEBSITE SUPPLEMENT 5
Lemma 1.6. The polynomial B
00
n
(z) (n 4) equals
B
00
n+2
(z) = P
E
n+1
(z) + 2
n1
z
2
O
n
(z) =
1
2
(P
n+1
(z) + 1 z
2
) + 2
n1
z
2
O
n
(z). (4)
where O
n1
(z) is the rank-distribution polynomial of closed-end ladders L
n2
and P
E
n
(z) is the
rank-distribution polynomial over the set M
O,Y,Zeven
n
.
Proof. If c = 0, in this case, it contributes the term P
E
n+1
(z) to B
00
n+2
(z). Otherwise c = 1,
and since there are 2
n1
possible choices of z
1
, z
2
, , z
n
such that the number of variables
in {z
1
, z
2
, , z
n
} equal to 1 is odd, it contributes 2
n1
z
2
O
n
(z) to B
00
n+2
(z). The rest follows
immediately by Proposition 1.5.
Lemma 1.7. The polynomial B
10
n
(z) (n 4) is given by the formula
B
10
n+2
(z) = 4z
2
P
n
(z),
where P
n
(z) is the rank-distribution polynomial of the Ringel ladder R
n2
.
Proof. Multiply row 3 by the constant c and add to the second row, Similarly, Multiply column
3 by the constant c and add the result to the second row. If z
1
= 1, we add the rst row to
the third row and then add the rst column to the third column. There is a similar situation
for y
1
. We interchange row 2 with row 3 and column 2 with column 3, and thereby transform
M
c,1,0,0,Y,Z
n+2
to the following form:
_
_
_
_
_
_
_
_
_
_
_
_
_
_
0 1
1 0 0
0 z
2
z
3
z
n1
z
n
z
2
0 y
2
0 z
3
y
2
0
.
.
. 0
.
.
.
.
.
.
.
.
.
z
n1
0 0 y
n1
z
n
y
n1
0
_
_
_
_
_
_
_
_
_
_
_
_
_
_
.
Note that the lower-right n n submatrix has the form M
0,Y,Z
n
. There are 2
2
dierent as-
signments of the variables y
1
and z
1
in this case, so it contributes 4z
2
P
n
(z) to the polynomial
B
10
n+2
(z).
By symmetry, we have the following lemma.
Lemma 1.8. The polynomial B
01
n
(z) (n 4) satises this formula:
B
01
n+2
(z) = 4z
2
P
n
(z),
where P
n
(z) is the rank-distribution polynomial of the Ringel ladder R
n2
.
6 YICHAO CHEN, JONATHAN L. GROSS, AND TOUFIK MANSOUR
Let
M
X,Y,Z,1
n+1
=
_
_
_
_
_
_
_
_
_
_
_
_
x
0
z
1
z
2
z
3
. . . z
n1
z
n
z
1
x
1
y
1
0 . . . 0 1
z
2
y
1
x
2
y
2
z
3
0 y
2
x
3
.
.
. 0
.
.
.
.
.
.
.
.
.
.
.
. y
n2
z
n1
0 y
n2
x
n1
y
n1
z
n
1 y
n1
x
n
_
_
_
_
_
_
_
_
_
_
_
_
, where
X = (x
0
, x
1
, . . . , x
n
) Z
n+1
2
, Y = (y
1
, y
2
, . . . , y
n1
) Z
n1
2
, and Z = (z
1
, z
2
, . . . , z
n
) Z
n
2
.
We let Q
n+1
be the set of all matrices over Z
2
of the form M
O,Y,Z,1
n+1
, and we let
Q
n+1
(z) =
n+1

j=0
Q
n+1
(j)z
j
be the rank-distribution polynomial of the set Q
n+1
.
We now introduce some further notations:
M
O,Y,Zeven,1
n+1
is the number of matrices with evenly many elements of Z equal to 1;
M
O,Y,Z
odd
,1
n+1
is the number of matrices with oddly many elements of Z equal to 1;
Q
O
n+1
(z) is the rank-distribution polynomial over the set M
O,Y,Z
odd
n+1
;
Q
E
n+1
(z) is the rank-distribution polynomial over the set M
O,Y,Zeven,1
n+1
.
Its clear that Q
n+1
(z) = Q
E
n+1
(z) +Q
O
n+1
(z).
Similarly, we have further notations:
H
n+1
is the set of all matrices over Z
2
tof the form M
X,Y,Z,1
n+1
;
H
n+1
(z) =

n+1
j=0
H
n+1
(j)z
j
is the rank-distribution polynomial of the set Q
n+1
;
M
X,Y,Zeven,1
n+1
is the number of matrices with evenly many elements of Z equal to 1;
Q
E
n+1
(z) is the rank-distribution polynomial of the set M
X,Y,Zeven,1
n+1
;
M
X,Y,Z
odd
,1
n+1
is the number of matrices such that oddly many elements of Z equal 1; and
H
O
n+1
(z) is the rank-distribution polynomial of the set M
X,Y,Z
odd
n+1
.
Lemma 1.9. The polynomial Q
E
n
(z) (n 4) satises the recurrence relation
Q
E
n+1
(z) = 6z
2
P
n1
(z) + 4z
2
Q
E
n1
(z)
with the initial conditions Q
E
2
(z) = 1 and Q
E
3
(z) = 1+3z
2
, where P
n1
(z) is the rank-distribution
polynomial of the Ringel ladder R
n3
.
Proof. There are four cases.
(1) Case 1: z
n
= 0, y
n1
= 0. If z
1
= 1, we rst add the last row to the rst row, and then
add the last column to the rst column. By a similar discussion for y
1
, we can transform
GENUS DISTRIBUTIONS OF CIRCULAR LADDERS: A WEBSITE SUPPLEMENT 7
M
O,Y,Z,1
n+1
into the following forms:
_
_
_
_
_
_
_
_
_
_
_
_
0 z
1
z
2
z
3
. . . z
n1
0
z
1
0 y
1
0 . . . 0 1
z
2
y
1
0 y
2
z
3
y
2
0
.
.
. 0
.
.
. 0
.
.
.
.
.
. y
n2
z
n1
y
n2
0 0
0 1 0 . . . 0 0 0
_
_
_
_
_
_
_
_
_
_
_
_
and
_
_
_
_
_
_
_
_
_
_
_
_
0 0 z
2
z
3
. . . z
n1
0
0 0 0 0 . . . 0 1
z
2
0 0 y
2
z
3
y
2
0
.
.
. 0
.
.
. 0
.
.
.
.
.
. y
n2
z
n1
y
n2
0 0
0 1 0 . . . 0 0 0
_
_
_
_
_
_
_
_
_
_
_
_
.
Note that there are four dierent ways to assign the variables y
1
and z
1
in the matrix
M
O,Y,Z,1
n+1
. If z
1
= 0, the matrix contributes 2z
2
P
E
n1
(z) to the polynomial Q
E
n+1
(z).
Otherwise z
1
= 1, and it contributes a term 2z
2
P
O
n1
(z). Since P
n1
(z) = P
E
n1
(z) +
P
O
n1
(z), it contributes 2z
2
P
n1
(z) to the polynomial Q
E
n+1
(z).
(2) Case 2: z
n
= 1, y
n1
= 0. We rst add the second row to the rst row, and then add the
second column to the rst column. This transforms M
O,Y,Z,1
n+1
into the following form:
_
_
_
_
_
_
_
_
_
_
_
_
0 z
1
z
2
+y
1
z
3
. . . z
n1
0
z
1
0 y
1
0 . . . 0 1
z
2
+y
1
y
1
0 y
2
z
3
y
2
0
.
.
. 0
.
.
. 0
.
.
.
.
.
. y
n2
z
n1
y
n2
0 0
0 1 0 . . . 0 0 0
_
_
_
_
_
_
_
_
_
_
_
_
.
If z
1
= 1, we rst add the last row to the rst row, and then add the last column to
the rst column. By similar discussion regarding y
1
, we can transform the above matrix
into the following form:
_
_
_
_
_
_
_
_
_
_
_
_
0 0 z
2
+y
1
z
3
. . . z
n1
0
0 0 0 0 . . . 0 1
z
2
+y
1
0 0 y
2
z
3
y
2
0
.
.
. 0
.
.
. 0
.
.
.
.
.
. y
n2
z
n1
y
n2
0 0
0 1 0 . . . 0 0 0
_
_
_
_
_
_
_
_
_
_
_
_
.
Accordingly, this case contributes 2z
2
P
n1
(z) to the polynomial Q
E
n+1
(z).
8 YICHAO CHEN, JONATHAN L. GROSS, AND TOUFIK MANSOUR
(3) Case 3: z
n
= 0, y
n1
= 1.
_
_
_
_
_
_
_
_
_
_
_
_
_
_
0 z
1
z
2
z
3
. . . z
n2
z
n1
0
z
1
0 y
1
0 . . . 0 0 1
z
2
y
1
0 y
2
0
z
3
0 y
2
0
.
.
. 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
z
n2
0 0 y
n2
0
z
n1
0 0 y
n2
0 1
0 1 0 . . . 0 1 0
_
_
_
_
_
_
_
_
_
_
_
_
_
_
,
We add the n-th row to the second row, then add the n-th column to the second column,
the resulted matrix has the following form:
_
_
_
_
_
_
_
_
_
_
_
_
_
_
0 z
1
+z
n1
z
2
z
3
. . . z
n2
z
n1
0
z
1
+z
n1
0 y
1
0 . . . y
n2
0 0
z
2
y
1
0 y
2
0
z
3
0 y
2
0
.
.
. 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
z
n2
y
n2
0 y
n2
0
z
n1
0 0 y
n2
0 1
0 0 0 . . . 0 1 0
_
_
_
_
_
_
_
_
_
_
_
_
_
_
,
No matter what the assignments of the values of z
n1
and y
n2
, we can transfer the
above matrix to the following form:
_
_
_
_
_
_
_
_
_
_
_
_
_
_
0 z
1
+z
n1
z
2
z
3
. . . z
n2
0 0
z
1
+z
n1
0 y
1
0 . . . y
n2
0 0
z
2
y
1
0 y
2
0
z
3
0 y
2
0
.
.
. 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
z
n2
y
n2
0 0 0
0 0 0 0 0 1
0 0 0 . . . 0 1 0
_
_
_
_
_
_
_
_
_
_
_
_
_
_
,
If y
n2
= 0, it contributes to the polynomial Q
E
n+1
(z) by a term 2z
2
P
E
n1
(z). Otherwise
y
n2
= 1, this case contributes to the polynomial Q
E
n+1
(z) by a term 2z
2
Q
E
n1
(z).
(4) Case 4: z
n
= 1, y
n1
= 1.
_
_
_
_
_
_
_
_
_
_
_
_
0 z
1
z
2
z
3
. . . z
n1
1
z
1
0 y
1
0 . . . 0 1
z
2
y
1
0 y
2
z
3
y
2
0
.
.
. 0
.
.
. 0
.
.
.
.
.
. y
n2
z
n1
y
n2
0 1
1 1 0 . . . 0 1 0
_
_
_
_
_
_
_
_
_
_
_
_
.
GENUS DISTRIBUTIONS OF CIRCULAR LADDERS: A WEBSITE SUPPLEMENT 9
We rst add the nth row to the rst and second row, then add the nth column to
the rst and second column, the resulting matrix has the following form:
_
_
_
_
_
_
_
_
_
_
_
_
_
_
0 z
1
+z
n1
z
2
z
3
. . . z
n2
+y
n2
z
n1
0
z
1
+z
n1
0 y
1
0 . . . y
n2
0 0
z
2
y
1
0 y
2
0
z
3
0 y
2
0
.
.
. 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
z
n2
+y
n2
y
n2
0 y
n2
0
z
n1
0 0 y
n2
0 1
0 0 0 . . . 0 1 0
_
_
_
_
_
_
_
_
_
_
_
_
_
_
.
If y
n2
= 0, it contributes 2z
2
P
O
n1
(z) to the polynomial Q
E
n+1
(z). Otherwise y
n2
= 1,
this case contributes 2z
2
Q
E
n1
(z) to the polynomial Q
E
n+1
(z).

Proposition 1.10. The generating function Q

E
(t; z) =

n2
Q
E
n
(z)t
n
is given by
t
2
(1 (2 3z
2
)t (1 + 19z
2
6z
4
)t
2
+ 2(1 + 4z
2
)(1 3z
2
)t
3
+ 8z
2
(3 + 5z
2
3z
4
)t
4
+ 64z
4
t
5
)
(1 2t 8z
2
t
2
)(1 t 8z
2
t
2
)(1 4z
2
t
2
)
.
Proof. Multiplying the recurrence relation in the statement of Lemma 1.9 by t
n+1
and summing
over all n 3 with using Theorem 1.3, we attain the result.
Lemma 1.11. The polynomial B
11
n
(z) (n 4) equals
B
11
n+2
(z) = 2z
2
P
n
(z) + 4z
2
Q
E
n
(z),
where P
n
(z) is the rank-distribution polynomial of the Ringel ladder R
n2
, and where Q
E
n
(z) is
the rank distribution polynomial over the set M
O,Y,Zeven,1
n
.
Proof. In this case, the matrix has the following form:
_
_
_
_
_
_
_
_
_
_
_
_
_
_
0 c 1 0 1
c 0 z
1
z
2
z
n2
z
n1
z
n
1 z
1
0 y
1
z
2
y
1
0
.
.
. 0
0
.
.
.
.
.
.
.
.
.
z
n2
0 y
n2
z
n1
0 y
n2
0 y
n1
1 z
n
y
n1
0
_
_
_
_
_
_
_
_
_
_
_
_
_
_
.
Multiply the last row constant c and add to the second row. Similarly, multiply the last column
constant c and add to the second column. A similar discussion to the row (column) 3. The
above matrix is transformed into this:
10 YICHAO CHEN, JONATHAN L. GROSS, AND TOUFIK MANSOUR
_
_
_
_
_
_
_
_
_
_
_
_
_
_
0 0 0 0 0 0 1
0 0 z
1
+z
n
z
2
z
n2
z
n1
+cy
n1
z
n
0 z
1
+z
n
0 y
1
0 y
n1
0
0 z
2
y
1
0
.
.
. 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 z
n2
0 0 y
n2
0 z
n1
+cy
n1
y
n1
0 y
n2
0 y
n1
1 z
n
0 y
n1
0
_
_
_
_
_
_
_
_
_
_
_
_
_
_
.
(1) Case 1: c = 0. There are four subcases.
Subcase 1: z
n
= 0, y
n1
= 0. Note that the down-right submatrix is the matrix
the form M
O,Y,Zeven
n
. In this case, it contributes to the polynomial B
11
n+2
(z) by a
term z
2
P
E
n
(z).
Subcase 2: z
n
= 1, y
n1
= 0. We rst add the rst row to the second row, then
add the rst column to the second column. Note that the down-right submatrix
is the matrix the form M
O,Y,Zeven
n
. In this case, it contributes to the polynomial
B
11
n+2
(z) by a term z
2
P
E
n
(z).
Subcase 3: z
n
= 0, y
n1
= 1. We rst delete the rst row and the last row, then
delete the rst column and the last column. At last we obtain a matrix of the form
M
O,Y,Zeven,1
n1
. In this case, it contributes to the polynomial B
11
n+2
(z) by a term
z
2
Q
E
n1
(z).
_
_
_
_
_
_
_
_
_
_
_
_
_
_
0 0 0 0 0 0 1
0 0 z
1
z
2
z
n2
z
n1
0
0 z
1
0 y
1
0 1 0
0 z
2
y
1
0
.
.
. 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 z
n2
0 0 y
n2
0 z
n1
1 0 y
n2
0 1
1 0 0 1 0
_
_
_
_
_
_
_
_
_
_
_
_
_
_
.
Subcase 4: z
n
= 1, y
n1
= 1. We rst delete the rst row and the last row, then
delete the rst column and the last column. At last we obtain a matrix of the
form M
O,Y,Zeven,1
n1
. In this case, it contributes to the polynomial B
11
n+2
(z) by a term
z
2
Q
E
n1
(z).
(2) Case 2: c = 1. In this case also, we have four subcases:
_
_
_
_
_
_
_
_
_
_
_
_
_
_
0 0 0 0 0 0 1
0 0 z
1
+z
n
z
2
z
n2
z
n1
+cy
n1
z
n
0 z
1
+z
n
0 y
1
0 y
n1
0
0 z
2
y
1
0
.
.
. 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 z
n2
0 0 y
n2
0 z
n1
+cy
n1
y
n1
0 y
n2
0 y
n1
1 z
n
0 y
n1
0
_
_
_
_
_
_
_
_
_
_
_
_
_
_
.
GENUS DISTRIBUTIONS OF CIRCULAR LADDERS: A WEBSITE SUPPLEMENT 11
Subcase 1: z
n
= 0, y
n1
= 0. Note that the down-right submatrix is the matrix the
form M
O,Y,Z
odd
n
. In this case, it contributes to the polynomial B
11
n+2
(z) by a term
z
2
P
O
n
(z).
Subcase 2: z
n
= 1, y
n1
= 0. We rst add the rst row to the second row, then add
the rst column to the second column. Note that the down-right submatrix is the
matrix the form M
O,Y,Z
odd
n
. In this case, it contributes to the polynomial B
11
n+2
(z)
by a term z
2
P
O
n
(z).
Subcase 3: z
n
= 0, y
n1
= 1. We rst delete the rst row and the last row, then
delete the rst column and the last column. At last we obtain a matrix of the form
M
O,Y,Zeven,1
n1
. In this case, it contributes to the polynomial B
11
n+2
(z) by a term
z
2
Q
E
n1
(z):
_
_
_
_
_
_
_
_
_
_
_
_
_
_
0 0 0 0 0 0 1
0 0 z
1
z
2
z
n2
z
n1
+ 1 0
0 z
1
0 y
1
0 1 0
0 z
2
y
1
0
.
.
. 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 z
n2
0 0 y
n2
0 z
n1
+ 1 1 0 y
n2
0 1
1 0 0 1 0
_
_
_
_
_
_
_
_
_
_
_
_
_
_
.
Subcase 4: z
n
= 1, y
n1
= 1. We rst delete the rst row and the last row, then
delete the rst column and the last column. At last we obtain a matrix of the form
M
O,Y,Zeven,1
n1
. In this case, it contributes to the polynomial B
11
n+2
(z) by a term
z
2
Q
E
n1
(z):
_
_
_
_
_
_
_
_
_
_
_
_
_
_
0 0 0 0 0 0 1
0 0 z
1
+ 1 z
2
z
n2
z
n1
+ 1 1
0 z
1
+ 1 0 y
1
0 1 0
0 z
2
y
1
0
.
.
. 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 z
n2
0 0 y
n2
0 z
n1
+ 1 1 0 y
n2
0 1
1 1 0 1 0
_
_
_
_
_
_
_
_
_
_
_
_
_
_
.

Theorem 1.12. For all n 5,

B
n
(z) =
1
2
(P
n
(z) + 1 z
2
) +Q
E
n
(z).
Moreover, the generating function B(t; z) =

n5
B
n
(z)t
n
is given by
t
5
f(t; z)
(1 2t 8z
2
t
2
)(1 t 8z
2
t
2
)(1 t)(1 4z
2
t
2
)
,
12 YICHAO CHEN, JONATHAN L. GROSS, AND TOUFIK MANSOUR
where
f(t; z) = 100z
4
+ 27z
2
+ 1 + (z
2
1)(160z
4
+ 76z
2
+ 3)t + (30z
2
640z
4
+ 2 1312z
6
)t
2
4z
2
(213z
2
7 + 60z
4
+ 352z
6
)t
3
+ 8z
2
(15z
2
+ 448z
4
+ 528z
6
1)t
4
+ 32z
4
(96z
6
3 + 68z
4
65z
2
)t
5
256z
6
(1 + 19z
2
+ 12z
4
)t
6
.
Proof. By Property 1.1 and Lemmas 1.6, 1.7, 1.8 and 1.11, we obtain B
n
(z) = P
E
n
(z) +Q
E
n
(z),
and by Proposition 1.5, we obtain B
n
(z) =
1
2
(P
n
(z) + 1 z
2
) + Q
E
n
(z). Multiplying by t
n
and summing over all n 5 with using Theorem 1.3 and Proposition 1.10, we complete the
proof.
1.2. The genus polynomial of circular ladders.
Theorem 1.13. We have

n4
g
CLn
(x)t
n
=
2
t
B(t;

x). Moreover, for all n 4, the number

of distinct cellular imbeddings of CL
n
in a surface of genus j is
_

_
7n+j
j
2
3j3
_
nj1
j1
_
+
n4j+2
j
2
3j3
_
nj+1
j1
_
+
n
j1
2
n+j1
_
nj
j2
_
+2
n1

n,2j+2
+ 2
n

n,2j+1
3 2
n1

n,2j
j 2,
2
n
+ 8n 2 + 8
n,4
j = 1,
2 j = 0.
Proof. Let GCL(t; x) =

n4
g
CLn
(x)t
n
. Then by Theorem 1.12 we have
GCL(t; x) =
2t
4
f(t; x)
(1 2t 8xt
2
)(1 t 8xt
2
)(1 t)(1 4xt
2
)
,
where
f(t; z) = 100x
2
+ 27x + 1 + (x 1)(160x
2
+ 76x + 3)t + (30x 640x
2
+ 2 1312x
3
)t
2
4x(213x 7 + 60x
2
+ 352x
3
)t
3
+ 8x(15x + 448x
2
+ 528x
3
1)t
4
+ 32x
2
(96x
3
3 + 68x
2
65x)t
5
256x
3
(1 + 19x + 12x
2
)t
6
.
By several algebraic operations, we calculate that the coecient of x
j
in GCL(t; x) is given by
_

(8t
3
15t
2
3t+2)2
3j3
t
2j2
(1t)
j+1
+
(13t+2t
2
)2
3j2
t
2j2
(12t)
j+1
+ (2t
2
+ 2t + 3/2)2
2j
t
2j
j 2,
2t
4
(27+50t
2
8t
3
73t)
(12t)(1t)
2
j = 1,
2t
4
1t
j = 0.
Hence, the coecient of x
j
t
n
in GCL(t; x) is given by
_

_
2
3j3
_
8
_
nj1
j
_
+ 15
_
nj
j
_
+ 3
_
nj+1
j
_
2
_
nj+2
j
_
_
+2
n+j1
_
2
_
nj+2
j
_
3
_
nj+1
j
_
+
_
nj
j
_
_
+ 2
n1

n,2j+2
+ 2
n

n,2j+1
3 2
n1

n,2j
j 2,
2
n
+ 8n 2 + 8
n,4
j = 1,
2 j = 0.
GENUS DISTRIBUTIONS OF CIRCULAR LADDERS: A WEBSITE SUPPLEMENT 13
which is equivalent to
_

_
7n+j
j
2
3j3
_
nj1
j1
_
+
n4j+2
j
2
3j3
_
nj+1
j1
_
+
n
j1
2
n+j1
_
nj
j2
_
+2
n1

n,2j+2
+ 2
n

n,2j+1
3 2
n1

n,2j
j 2,
2
n
+ 8n 2 + 8
n,4
j = 1,
2 j = 0.
as claimed.
Note that it is not hard to check that our formula in the above theorem is equivalent to the
formula in [4, Theorem 3.10].
As a corollary of the above theorem we can obtain the following result.
Corollary 1.14. For all n 4,
g
CLn
(x) = 1 x +
1 3x 2

x
4x
(2

x)
n
+
1 3x + 2

x
4x
(2

x)
n
+ 2
n
x(i

2x)
n
_
U
n
_
1
2i

2x
_
U
n2
_
1
2i

2x
__
+ (1 x)(2i

2x)
n
_
U
n
_
1
4i

2x
_
U
n2
_
1
4i

2x
__
,
where U
s
is the s-th Chebyshev polynomial of the second kind and i
2
= 1
Proof. By Theorem 1.13 we infer that the generating function f =

n4
g
CLn
(x)t
n
can be
written as
f = 2(1 + 19x + 12x
2
)t
3
4(1 + 3x)t
2
4t
1
2
x(1 + 3x 2x
2
) +
1 3x + 4xt
2x(1 4xt
2
)
+
2x(1 t)
1 2t 8xt
2
+
2 2x +xt t
1 t 8xt
2
+
1 x
1 t
.
Let n 4, so the coecient g
CLn
(x) of t
n
in the generating function f is given by
g
CLn
(x) = 1 x +
1 3x 2

x
4x
(2

x)
n
+
1 3x + 2

x
4x
(2

x)
n
+ 2
n
x(i

2x)
n
_
U
n
_
1
2i

2x
_
U
n2
_
1
2i

2x
__
+ (1 x)(2i

2x)
n
_
U
n
_
1
4i

2x
_
U
n2
_
1
4i

2x
__
,
where the rst, second and third lines are the coecients of t
n
in the generating functions
1 x
1 t
+
1 3x + 4xt
2x(1 4xt
2
)
,
2x(1 t)
1 2t 8xt
2
, and
2 2x +xt t
1 t 8xt
2
,
respectively.
14 YICHAO CHEN, JONATHAN L. GROSS, AND TOUFIK MANSOUR
References
[1] J. Chen, J. L. Gross, and R. G. Rieper, Overlap matrices and total embeddings, Discrete Math. 128 (1994)
7394.
[2] Y. Chen, T. Mansour, and Q. Zou, Embedding distributions and Chebyshev polynomial, Graphs and Com-
bin., DOI: 10.1007/s00373-011-1075-5.
[3] Y. Chen, L. Ou, and Q. Zou, Total embedding distributions of Ringel ladders, Discrete Math. 311 (2011)
24632474.
[4] L. A. McGeoch, Genus distribution for circular and Mobius ladders, Technical report extracted from PhD
thesis, Carnegie-Mellon University, 1987.
College of Mathematics and Econometrics, Hunan University, 410082 Changsha, China
E-mail address: ycchen@hnu.edu.cn
Department of Computer Science, Columbia University, New York, NY 10027 USA
E-mail address: gross@cs.columbia.edu
Department of Mathematics, University of Haifa, 31905 Haifa, Israel
E-mail address: toufik@math.haifa.ac.il