recent progress on tiling proofs of q--series identities · proof. construct tilings in the...
TRANSCRIPT
Recent Progress on Tiling Proofs of
q–Series Identities
David P. Little
November 11, 2008
www.math.psu.edu/dlittle
Weighted Tilings
Definition
A tiling is a covering of an infinitely long board:
· · ·1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Weighted Tilings
Definition
A tiling is a covering of an infinitely long board:
· · ·1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
using different types of tiles:
Weighted Tilings
Definition
A tiling is a covering of an infinitely long board:
· · ·1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
using different types of tiles:
Weighted Tilings
Definition
A tiling is a covering of an infinitely long board:
· · ·1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
using different types of tiles:
3 5
Weighted Tilings
Definition
A tiling is a covering of an infinitely long board:
· · ·1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
using different types of tiles:
3 5
The weight of a tiling T is given by
w(T ) =∏
t∈T
w(t)
where w(t) is the weight of the tile t. The weight of a white square willalways be 1. Each tiling will have a finite number of weighted tiles.
Lebesgue Identities
The weight of tile t:
w(t) =
qi if t is a in position i
zqi if t is a in position i
1 if t is a in position i
Theorem
∞∑
n=0
(−z; q)n(q; q)n
q(n+1
2 ) =
∞∏
n=1
(1 + qn)(1 + zq2n−1)
where (z; q)n = (1 − z)(1 − zq) · · · (1 − zqn−1).
Eight Identities of Rogers
∑
n≥0
znqn2
(q; q)n= (−zq2; q2)∞
∑
n≥0
znqn2
(q2; q2)n(−zq2; q2)n
∑
n≥0
znqn2+n
(q; q)n= (−zq2; q2)∞
∑
n≥0
znqn2+2n
(q2; q2)n(−zq2; q2)n
∑
n≥0
z2nq4n2+2n
(q4; q4)n= (zq2; q2)∞
∑
n≥0
znqn2+n
(q2; q2)n(zq2; q2)n
∑
n≥0
znq2n2+n
(q2; q2)n= (zq2; q2)∞
∑
n≥0
znq(3n2+n)/2
(q; q)n(zq2; q2)n
Five More Identities
∑
n≥0
znq2n2
(q2; q2)n= (zq; q2)∞
∑
n≥0
znq(3n2+n)/2
(q; q)n(zq; q2)n+1
∑
n≥0
znqn2
(q; q)n= (−zq; q)∞
∑
n≥0
(−1)nz2nq3n2
(q2; q2)n(−zq; q)2n
∑
n≥0
znqn2+n(1 − z2q2n+3)
(q; q)n= (−zq; q)∞
∑
n≥0
(−1)nz2nq3n2
(q2; q2)n(−zq; q)2n+1
An Example
The weight of tile t:
w(t) =
−zqi if t is a in position i
zqi if t is a in position i
1 if t is a in position i
An Example
The weight of tile t:
w(t) =
−zqi if t is a in position i
zqi if t is a in position i
1 if t is a in position i
Theorem (Cauchy)
∞∑
n=0
znqn2
(q; q)n(zq; q)n=
1
(zq; q)∞
An Example
The weight of tile t:
w(t) =
−zqi if t is a in position i
zqi if t is a in position i
1 if t is a in position i
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 1: Place n in positions 1, 3, 5, . . . , 2n − 1.
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 1: Place n in positions 1, 3, 5, . . . , 2n − 1.
· · ·
This accounts for a weight of
znq1+3+5+···+(2n−1) = znqn2
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 2: Pick j ≥ 0 and insert immediately after the jth tile.
︷ ︸︸ ︷j ≥ 0 tiles
· · ·
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 2: Pick j ≥ 0 and insert immediately after the jth tile.
︷ ︸︸ ︷j ≥ 0 tiles
· · ·︸ ︷︷ ︸
k dominoes
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 2: Pick j ≥ 0 and insert immediately after the jth tile.
︷ ︸︸ ︷j ≥ 0 tiles
· · ·︸ ︷︷ ︸
k dominoes︸ ︷︷ ︸
n − k dominoes
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 2: Pick j ≥ 0 and insert immediately after the jth tile.
︷ ︸︸ ︷j ≥ 0 tiles
· · ·︸ ︷︷ ︸
k dominoes︸ ︷︷ ︸
n − k dominoes
position j + k + 1
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 2: Pick j ≥ 0 and insert immediately after the jth tile.
︷ ︸︸ ︷j ≥ 0 tiles
· · ·︸ ︷︷ ︸
k dominoes︸ ︷︷ ︸
n − k dominoes
position j + k + 1
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 2: Pick j ≥ 0 and insert immediately after the jth tile.
︷ ︸︸ ︷j ≥ 0 tiles
· · ·︸ ︷︷ ︸
k dominoes︸ ︷︷ ︸
n − k dominoes
position j + k + 1
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 2: Pick j ≥ 0 and insert immediately after the jth tile.
︷ ︸︸ ︷j ≥ 0 tiles
· · ·︸ ︷︷ ︸
k dominoes︸ ︷︷ ︸
n − k dominoes
position j + k + 1
Increases the weight of the tiling by −zqj+k+1qn−k = −zqn+j+1
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 2: Pick j ≥ 0 and insert immediately after the jth tile.
︷ ︸︸ ︷j ≥ 0 tiles
· · ·︸ ︷︷ ︸
k dominoes︸ ︷︷ ︸
n − k dominoes
position j + k + 1
Increases the weight of the tiling by −zqj+k+1qn−k = −zqn+j+1
∏
j≥0
(1 − zqn+j+1) =(zq; q)∞(zq; q)n
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 3: Project the dominoes
· · ·· · · · · ·
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 3: Project the dominoes
· · ·· · · · · ·
↓
· · ·· · · · · ·
Increases the weight of the tiling by a factor of q.
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 3: Project the dominoes
· · ·· · · · · ·
↓
· · ·· · · · · ·
Increases the weight of the tiling by a factor of q. Therefore, the left–handside is the generating function for all weighted tilings.
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 3: Project the dominoes
When projecting tiles, always work in a right to left, weakly increasingmanner. In other words, make sure that each domino is projected at leastas many times as the dominoes to its left are projected.
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Construct tilings in the following manner:
STEP 3: Project the dominoes
When projecting tiles, always work in a right to left, weakly increasingmanner. In other words, make sure that each domino is projected at leastas many times as the dominoes to its left are projected.
This process accounts for a weight of
1
(1 − q)(1 − q2)(1 − q3) · · · (1 − qn)=
1
(q; q)n
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Cancel out all non-empty tilings:
STEP 4:Find first occurrence of: or
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Cancel out all non-empty tilings:
STEP 4:Find first occurrence of: or
and replace with: or (respectively)
Theorem (Cauchy)
(zq; q)∞
∞∑
n=0
znqn2
(q; q)n(zq; q)n= 1
Proof. Cancel out all non-empty tilings:
STEP 4:Find first occurrence of: or
and replace with: or (respectively)
The only remaining tiling is the empty tiling, which has weight 1.
q-analog of the binomial series
Weight tiles in the following manner:
w(t) =
aqi if t is a with i or to its left
bqi if t is a with i or to its left
1 if t is a
q-analog of the binomial series
Weight tiles in the following manner:
w(t) =
aqi if t is a with i or to its left
bqi if t is a with i or to its left
1 if t is a
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART I: Interpret infinite series
STEP 1: Place n black or gray squares in positions 1, 2, 3, . . . , n.
· · ·
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART I: Interpret infinite series
STEP 1: Place n black or gray squares in positions 1, 2, 3, . . . , n.
· · ·
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART I: Interpret infinite series
STEP 1: Place n black or gray squares in positions 1, 2, 3, . . . , n.
· · ·
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART I: Interpret infinite series
STEP 1: Place n black or gray squares in positions 1, 2, 3, . . . , n.
· · ·
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART I: Interpret infinite series
STEP 1: Place n black or gray squares in positions 1, 2, 3, . . . , n.
· · ·
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART I: Interpret infinite series
STEP 1: Place n black or gray squares in positions 1, 2, 3, . . . , n.
· · ·
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART I: Interpret infinite series
STEP 1: Place n black or gray squares in positions 1, 2, 3, . . . , n.
· · ·
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART I: Interpret infinite series
STEP 1: Place n black or gray squares in positions 1, 2, 3, . . . , n.
· · ·
A in position i accounts for a weight of a.
A in position i accounts for a weight of bqn−i.This process accounts for a weight of
n∏
i=1
(a + bqn−i) = (−b/a; q)nan
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART I: Interpret infinite series
STEP 2: Project the tiles.
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART I: Interpret infinite series
STEP 2: Project the tiles.
↓
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART I: Interpret infinite series
STEP 2: Project the tiles.
↓ ↓
This process increases the weight of the tiling by a factor of q.
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART I: Interpret infinite series
STEP 2: Project the tiles.
↓ ↓
This process increases the weight of the tiling by a factor of q. Therefore,the left-hand side is the generating function for all weighted tilings.
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART II: Interpret infinite product
Each tiling can be broken up into segments:
· · · · · ·
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART II: Interpret infinite product
Each tiling can be broken up into segments:
· · · · · ·︷ ︸︸ ︷
j ≥ 0 black squares
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART II: Interpret infinite product
Each tiling can be broken up into segments:
· · · · · ·︷ ︸︸ ︷
j ≥ 0 black squares
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART II: Interpret infinite product
Each tiling can be broken up into segments:
· · · · · ·︷ ︸︸ ︷
j ≥ 0 black squares
The weight of the nth segment for n ≥ 0 is given by
(1 + bqn)
∞∑
j=0
ajqnj =1 + bqn
1 − aqn
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Proof. PART II: Interpret infinite product
Each tiling can be broken up into segments:
· · · · · ·︷ ︸︸ ︷
j ≥ 0 black squares
The weight of the nth segment for n ≥ 0 is given by
(1 + bqn)
∞∑
j=0
ajqnj =1 + bqn
1 − aqn
Multiplying over n ≥ 0 completes the construction.
A Few Observations
The generating function for all weighted tilings is given by
∞∑
n=0
(−b/a; q)nan
(q; q)n
A Few Observations
The generating function for all weighted tilings is given by
∞∑
n=0
(−b/a; q)nan
(q; q)n
Adding the parameter c in the following manner allows us to countnumber of white squares before the last weighted tile.
∞∑
n=0
(−b/a; q)nan
(cq; q)n
A Few Observations
The generating function for all weighted tilings is given by
∞∑
n=0
(−b/a; q)nan
(q; q)n
Adding the parameter c in the following manner allows us to countnumber of white squares before the last weighted tile.
∞∑
n=0
(−b/a; q)nan
(cq; q)n
Multiplication by a produces the generating function for tilings that startwith a black square.
∞∑
n=0
(−b/a; q)nan+1
(cq; q)n
Rogers-Fine Identity
Theorem
∞∑
n=0
(−b/a; q)nan
(cq; q)n=
∞∑
n=0
(−b/a; q)n(−b/c; q)nancnqn2
(1 + bq2n)
(a; q)n+1(cq; q)n
Rogers-Fine Identity
Theorem
∞∑
n=0
(−b/a; q)nan
(cq; q)n=
∞∑
n=0
(−b/a; q)n(−b/c; q)nancnqn2
(1 + bq2n)
(a; q)n+1(cq; q)n
Multiplying both sides by (1 − a) yields:
(1 − a)
∞∑
n=0
(−b/a; q)nan
(cq; q)n=
∞∑
n=0
(−b/a; q)n(−b/c; q)nancnqn2
(1 + bq2n)
(aq; q)n(cq; q)n
In this form, the left hand side counts tilings that do not start with a blacksquare where the power of c keeps track of the number of white squaresbefore the last weighted tile.
Theorem (Rogers-Fine)
(1 − a)∞∑
n=0
(−b/a; q)nan
(cq; q)n=
∞∑
n=0
(−b/a; q)n(−b/c; q)nancnqn2
(1 + bq2n)
(aq; q)n(cq; q)n
Proof. Interpret right–hand side
Front segments:
· · ·
Theorem (Rogers-Fine)
(1 − a)∞∑
n=0
(−b/a; q)nan
(cq; q)n=
∞∑
n=0
(−b/a; q)n(−b/c; q)nancnqn2
(1 + bq2n)
(aq; q)n(cq; q)n
Proof. Interpret right–hand side
Front segments:
· · ·︸ ︷︷ ︸
j ≥ 0 black squares
Theorem (Rogers-Fine)
(1 − a)∞∑
n=0
(−b/a; q)nan
(cq; q)n=
∞∑
n=0
(−b/a; q)n(−b/c; q)nancnqn2
(1 + bq2n)
(aq; q)n(cq; q)n
Proof. Interpret right–hand side
Front segments:
· · ·︸ ︷︷ ︸
j ≥ 0 black squares
Back segments:
Theorem (Rogers-Fine)
(1 − a)∞∑
n=0
(−b/a; q)nan
(cq; q)n=
∞∑
n=0
(−b/a; q)n(−b/c; q)nancnqn2
(1 + bq2n)
(aq; q)n(cq; q)n
Proof. Interpret right–hand side
Front segments:
· · ·︸ ︷︷ ︸
j ≥ 0 black squares
Back segments:
︸ ︷︷ ︸
k ≥ 0 white squares
Theorem (Rogers-Fine)
(1 − a)∞∑
n=0
(−b/a; q)nan
(cq; q)n=
∞∑
n=0
(−b/a; q)n(−b/c; q)nancnqn2
(1 + bq2n)
(aq; q)n(cq; q)n
Proof. Interpret right–hand side
Front segments:
· · ·︸ ︷︷ ︸
j ≥ 0 black squares
Back segments:
︸ ︷︷ ︸
k ≥ 0 white squares
· · ·
The center of a tiling marks the transition between front segments andback segments. The center can either be empty or a gray square.
Theorem (Rogers-Fine)
(1 − a)
∞∑
n=0
(−b/a; q)nan
(cq; q)n=
∞∑
n=0
(−b/a; q)n(−b/c; q)nancnqn2
(1 + bq2n)
(aq; q)n(cq; q)n
Proof. Interpret right–hand side
Tilings that start with or and have n front/back segments.
Theorem (Rogers-Fine)
(1 − a)
∞∑
n=0
(−b/a; q)nan
(cq; q)n=
∞∑
n=0
(−b/a; q)n(−b/c; q)nancnqn2
(1 + bq2n)
(aq; q)n(cq; q)n
Proof. Interpret right–hand side
Tilings that start with or and have n front/back segments.Generating function for n front segments:
(c + b)
(1 − aq)
(c + bq)
(1 − aq2)· · ·
(c + bqn−1)
(1 − aqn)=
(−b/c; q)ncn
(aq; q)n
Theorem (Rogers-Fine)
(1 − a)
∞∑
n=0
(−b/a; q)nan
(cq; q)n=
∞∑
n=0
(−b/a; q)n(−b/c; q)nancnqn2
(1 + bq2n)
(aq; q)n(cq; q)n
Proof. Interpret right–hand side
Tilings that start with or and have n front/back segments.Generating function for n back segments:
(aqn + bq2n−1)
(1 − cqn)· · ·
(aqn + bqn+1)
(1 − cq2)
(aqn + bqn)
(1 − cq)=
(−b/a; q)nanqn2
(cq; q)n
Theorem (Rogers-Fine)
(1 − a)
∞∑
n=0
(−b/a; q)nan
(cq; q)n=
∞∑
n=0
(−b/a; q)n(−b/c; q)nancnqn2
(1 + bq2n)
(aq; q)n(cq; q)n
Proof. Interpret right–hand side
The center can either be empty or a gray square.
· · · · · ·︸ ︷︷ ︸
n front segments︸ ︷︷ ︸
n back segments
If the center is a gray square, then it has weight bqn and increases theweight of the back segments by qn.
Theorem (Rogers-Fine)
(1 − a)
∞∑
n=0
(−b/a; q)nan
(cq; q)n=
∞∑
n=0
(−b/a; q)n(−b/c; q)nancnqn2
(1 + bq2n)
(aq; q)n(cq; q)n
Proof. Interpret right–hand side
The center can either be empty or a gray square.
· · · · · ·︸ ︷︷ ︸
n front segments︸ ︷︷ ︸
n back segments
If the center is a gray square, then it has weight bqn and increases theweight of the back segments by qn.In other words, the factor (1 + bq2n) represents the choice of the center.
Recall:
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Recall:
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Theorem
∞∑
n=0
(−a/b; q)nbnq(n
2)
(q; q)n(a; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Recall:
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Theorem
∞∑
n=0
(−c/a; q)nanq(n
2)
(q; q)n(c; q)n=
∞∏
n=0
1 + aqn
1 − cqn
Recall:
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Theorem
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
1 + aqn
1 − cqn
Recall:
Theorem (Cauchy)
∞∑
n=0
(−b/a; q)nan
(q; q)n=
∞∏
n=0
1 + bqn
1 − aqn
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
∞∑
n=0
(a + c)(a + cq) · · · (a + cqn−1)q(n+1
2 )
(q; q)n(1 − cqn+1)(1 − cqn+2) · · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
∞∑
n=0
(a + c)(a + cq) · · · (a + cqn−1)q(n+1
2 )
(q; q)n(1 − cqn+1)(1 − cqn+2) · · ·
Weight tiles in the following manner:
w(t) =
aqi if t is a in position i
cqi if t is a in position i
1 if t is a
−cqi if t is a in position i
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 1: Place in positions 1, 2, 3, . . . , n.
· · ·
A in position i accounts for a weight of aqi.
This process accounts for a weight of
anq1+2+···+n = anq(n+1
2 )
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 2: Place or in positions i > n.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 2: Place or in positions i > n.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 2: Place or in positions i > n.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 2: Place or in positions i > n.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 2: Place or in positions i > n.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 2: Place or in positions i > n.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 2: Place or in positions i > n.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 2: Place or in positions i > n.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 2: Place or in positions i > n.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 2: Place or in positions i > n.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 2: Place or in positions i > n.
· · ·
A in position i accounts for a weight of 1.
A in position i accounts for a weight of −cqi.This process accounts for a weight of
∏
i>n
(1 − cqi) =(cq; q)∞(cq; q)n
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 3: Project the black tiles.
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 3: Project the black tiles.
↓
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 3: Project the black tiles.
↓ ↓
This process increases the weight of the tiling by a factor of q.
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 3a: Convert into .If a square is converted into a circle, project each of the tiles to its right.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 3a: Convert into .If a square is converted into a circle, project each of the tiles to its right.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 3a: Convert into .If a square is converted into a circle, project each of the tiles to its right.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 3a: Convert into .If a square is converted into a circle, project each of the tiles to its right.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 3a: Convert into .If a square is converted into a circle, project each of the tiles to its right.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 3a: Convert into .If a square is converted into a circle, project each of the tiles to its right.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 3a: Convert into .If a square is converted into a circle, project each of the tiles to its right.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 3a: Convert into .If a square is converted into a circle, project each of the tiles to its right.
· · ·
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 3a: Convert into .If a square is converted into a circle, project each of the tiles to its right.
· · ·
The factor (1 + cqn−i/a) represents the choice of converting the ithsquare into a circle.
n∏
i=1
(1 + cqn−i/a) = (−c/a; q)n
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART I: Interpret infinite series
STEP 3b: Project the black tiles.
· · ·
Constructs all tilings where every circle is followed by a white tile.
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART II: Interpret infinite product
Cancel out any tilings with a or
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART II: Interpret infinite product
Cancel out any tilings with a or
Find first occurrence of: or
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART II: Interpret infinite product
Cancel out any tilings with a or
Find first occurrence of: or
and replace with: or (respectively)
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART II: Interpret infinite product
Cancel out any tilings with a or
Find first occurrence of: or
and replace with: or (respectively)
Remaining tilings cannot have any circles.
Theorem
(cq; q)∞
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(cq; q)n=
∞∏
n=1
(1 + aqn)
Proof. PART II: Interpret infinite product
Cancel out any tilings with a or
Find first occurrence of: or
and replace with: or (respectively)
Remaining tilings cannot have any circles.
Therefore, each tiling can be constructed by simply deciding whether ornot to place a black square in each position n ≥ 1.
q-analog of Gauss’s Theorem
Weight tiles in the following manner:
w(t) =
aqi if t is a in position i
cqi if t is a in position i
abqi if t is a with i or to its left
bcqi if t is a with i or to its left
1 if t is a
−cqi if t is a in position i
q-analog of Gauss’s Theorem
Weight tiles in the following manner:
w(t) =
aqi if t is a in position i
cqi if t is a in position i
abqi if t is a with i or to its left
bcqi if t is a with i or to its left
1 if t is a
−cqi if t is a in position i
Theorem (Heine)
(cq; q)∞
∞∑
n=0
(−c/a; q)n(−q/b; q)nanbn
(q; q)n(cq; q)n=
∞∏
n=1
(1 + bcqn−1)(1 + aqn)
(1 − abqn−1)
q-analog of Kummer’s Theorem
Weight tiles in the following manner:
w(t) =
qi if t is a in position i
cqi if t is a in position i
bqi if t is a with i or to its left
bcqi if t is a with i or to its left
−bcqi if t is a in position i + 1
1 if t is a
q-analog of Kummer’s Theorem
Weight tiles in the following manner:
w(t) =
qi if t is a in position i
cqi if t is a in position i
bqi if t is a with i or to its left
bcqi if t is a with i or to its left
−bcqi if t is a in position i + 1
1 if t is a
Theorem (Bailey)
(bc; q)∞
∞∑
n=0
(−c; q)n(−q/b; q)nbn
(q; q)n(bc; q)n=
∞∏
n=1
(1 + qn)(1 + cq2n−1)(1 + cb2q2n−2)
1 − bqn−1
A q-Series Symmetry Result
Theorem (Ramanujan)
(−bq; q)n
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(−bq; q)n= (−aq; q)n
∞∑
n=0
(−c/b; q)nbnq(n+1
2 )
(q; q)n(−aq; q)n
A q-Series Symmetry Result
Theorem (Ramanujan)
(−bq; q)n
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(−bq; q)n= (−aq; q)n
∞∑
n=0
(−c/b; q)nbnq(n+1
2 )
(q; q)n(−aq; q)n
Equivalently, the following function is symmetric in the variables a and b:
(−bq; q)n
∞∑
n=0
(−c/a; q)nanq(n+1
2 )
(q; q)n(−bq; q)n
Labeled Tilings
Weight tiles in the following manner:
w(t) =
ajqi if t is a j in position i, 1 ≤ j ≤ k + 1
cjqi if t is a j in position i, 1 ≤ j ≤ k
1 if t is a in position i
Labeled Tilings
Weight tiles in the following manner:
w(t) =
ajqi if t is a j in position i, 1 ≤ j ≤ k + 1
cjqi if t is a j in position i, 1 ≤ j ≤ k
1 if t is a in position i
Consider the following generating function:
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 1: Place n1 ≥ 0 1-squares in positions 1, 2, . . . , n1, immediatelyfollowed by n2 ≥ 0 2–squares, and so on.
· · ·
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 1: Place n1 ≥ 0 1-squares in positions 1, 2, . . . , n1, immediatelyfollowed by n2 ≥ 0 2–squares, and so on.
· · ·1 1 1
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 1: Place n1 ≥ 0 1-squares in positions 1, 2, . . . , n1, immediatelyfollowed by n2 ≥ 0 2–squares, and so on.
· · ·1 1 1 2 2
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 1: Place n1 ≥ 0 1-squares in positions 1, 2, . . . , n1, immediatelyfollowed by n2 ≥ 0 2–squares, and so on.
· · ·1 1 1 2 2 3 3 3 3
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 1: Place n1 ≥ 0 1-squares in positions 1, 2, . . . , n1, immediatelyfollowed by n2 ≥ 0 2–squares, and so on.
· · ·1 1 1 2 2 3 3 3 3 4 4
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 1: Place n1 ≥ 0 1-squares in positions 1, 2, . . . , n1, immediatelyfollowed by n2 ≥ 0 2–squares, and so on.
· · ·1 1 1 2 2 3 3 3 3 4 4
A j in position i accounts for a weight of ajqi.
This accounts for a weight of
an1
1 · · · ank
k q(n1+n2+···+nk+1
2)
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 2: Arbitrarily place k +1-squares in positions i > n1 +n2 + · · ·+nk.
· · ·1 1 1 2 2 3 3 3 3 4 4
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 2: Arbitrarily place k +1-squares in positions i > n1 +n2 + · · ·+nk.
· · ·1 1 1 2 2 3 3 3 3 4 4 5
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 2: Arbitrarily place k +1-squares in positions i > n1 +n2 + · · ·+nk.
· · ·1 1 1 2 2 3 3 3 3 4 4 5
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 2: Arbitrarily place k +1-squares in positions i > n1 +n2 + · · ·+nk.
· · ·1 1 1 2 2 3 3 3 3 4 4 5 5
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 2: Arbitrarily place k +1-squares in positions i > n1 +n2 + · · ·+nk.
· · ·1 1 1 2 2 3 3 3 3 4 4 5 5 5
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 2: Arbitrarily place k +1-squares in positions i > n1 +n2 + · · ·+nk.
· · ·1 1 1 2 2 3 3 3 3 4 4 5 5 5
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 2: Arbitrarily place k +1-squares in positions i > n1 +n2 + · · ·+nk.
· · ·1 1 1 2 2 3 3 3 3 4 4 5 5 5
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 2: Arbitrarily place k +1-squares in positions i > n1 +n2 + · · ·+nk.
· · ·1 1 1 2 2 3 3 3 3 4 4 5 5 5
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 2: Arbitrarily place k +1-squares in positions i > n1 +n2 + · · ·+nk.
· · ·1 1 1 2 2 3 3 3 3 4 4 5 5 5 5
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 2: Arbitrarily place k +1-squares in positions i > n1 +n2 + · · ·+nk.
· · ·1 1 1 2 2 3 3 3 3 4 4 5 5 5 5
A j in position i accounts for a weight of aqi.
This accounts for a weight of
∏
i>n1+n2+···+nk
(1 + ak+1qi) =
(−ak+1q; q)∞(−ak+1q; q)n1+n2+···+nk
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 3: Projectiles, j and j.
j > j ≥ j j > j > j
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 3: Projectiles, j and j.
j > j ≥ j j > j > j
↓
> j j ≥ j
↓
> j j > j
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 3: Projectiles, j and j.
j > j ≥ j j > j > j
↓
> j j ≥ j
↓
> j j > j
This process increases the weight of a tiling by a factor of q.
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 3a: Decide whether or not to convert each j into j for eachj = k, . . . , 2, 1. If so, project every j-tile that appears to its right.This accounts for a weight of
k∏
i=1
nj∏
j=1
(1 + ciqj−1/ai) = (−c1/a1; q)n1
· · · (−ck/ak; q)nk
Note that every circle must be followed by a white square or a tile with alarger label.
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 3b: Project the j-tiles for j = k, . . . , 2, 1. As usual, work in a rightto left manner and make sure to project each j-tile at least as many timesas the j-tiles to its left are projected.
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 3b: Project the j-tiles for j = k, . . . , 2, 1. As usual, work in a rightto left manner and make sure to project each j-tile at least as many timesas the j-tiles to its left are projected. This accounts for a weight of
1
(q; q)n1· · · (q; q)nk
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
STEP 3b: Project the j-tiles for j = k, . . . , 2, 1. As usual, work in a rightto left manner and make sure to project each j-tile at least as many timesas the j-tiles to its left are projected. This accounts for a weight of
1
(q; q)n1· · · (q; q)nk
Remark: We have constructed all labeled tilings that consist of weaklyincreasing sequences of weighted tiles separated by a white square wherethe label must strictly increase after a circle.
Labeled Tilings
Theorem
The generating function
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
is symmetric in the variables (a1, a2, . . . , ak+1) and (c1, c2, . . . , ck).
Labeled Tilings
Theorem
The generating function
(−ak+1q)∞∑
n1,...,nk≥0
(−c1/a1)n1· · · (−ck/ak)nk
an1
1 · · · ank
k
(q)n1· · · (q)nk
(−ak+1q)n1+···+nk
q(n1+···+nk+1
2).
is symmetric in the variables (a1, a2, . . . , ak+1) and (c1, c2, . . . , ck).
Note that the following sequences of tiles have the same weight:
· · · · · ·1 1 2 2 2 3 3 3 3 4 4 5 5 5
· · · · · ·3 3 3 4 1 2 5 5 5 1 2 2 3 4
Tiling Statistics
Let Pn denote the set of Pell tilings, (i.e., tilings with , , ) ofa 1 × n board.
Definition
The number of or that are immediately followed by a intiling T is the number of descents in T , denoted des(T ).
Tiling Statistics
Let Pn denote the set of Pell tilings, (i.e., tilings with , , ) ofa 1 × n board.
Definition
The number of or that are immediately followed by a intiling T is the number of descents in T , denoted des(T ).
Find the generating function for descents
∑
T∈Pn
xdes(T )
Tiling Statistics
Let Pn denote the set of Pell tilings, (i.e., tilings with , , ) ofa 1 × n board.
Definition
The number of or that are immediately followed by a intiling T is the number of descents in T , denoted des(T ).
Find the generating function for descents
∑
T∈Pn
xdes(T )
Weight tilings in the following manner:
w(t) =
{
x if t is the last tile on the board
1 or − x otherwise
Tilings with no Descents
Let Fn(x) denote the G.F. for tilings of a 1×n board using and .
Tilings with no Descents
Let Fn(x) denote the G.F. for tilings of a 1×n board using and .
F0(x) = 1 ∅
F1(x) = x
F2(x) = (1 − x)x + x
Tilings with no Descents
Let Fn(x) denote the G.F. for tilings of a 1×n board using and .
F0(x) = 1 ∅
F1(x) = x
F2(x) = (1 − x)x + x
For n ≥ 3,
Fn(x) = (1 − x)Fn−1(x) + (1 − x)Fn−2(x)
F (x, t) =∞∑
n=0
Fn(x)tn
=1 + (2x − 1)t + (2x − 1)t2
1 − (1 − x)t + (1 − x)t2
F (x, t) =∞∑
n=0
Fn(x)tn
=1 + (2x − 1)t + (2x − 1)t2
1 − (1 − x)t + (1 − x)t2
Let Gn(x) denote the G.F. for tilings of a 1 × n board using , and
, where no or is followed by a .
F (x, t) =∞∑
n=0
Fn(x)tn
=1 + (2x − 1)t + (2x − 1)t2
1 − (1 − x)t + (1 − x)t2
Let Gn(x) denote the G.F. for tilings of a 1 × n board using , and
, where no or is followed by a .
G(x, t) =∞∑
n=0
Gn(x)tn
=1
1 − (1 − x)tF (x, t) −
(1 − x)t
1 − (1 − x)t+
xt
1 − (1 − x)t
=1 + 2(2x − 1)t + x(2x − 1)t2 + (x − 1)(2x − 1)t3
(1 − (1 − x)t)(1 − (1 − x)t + (1 − x)t2)
Symmetric Functions
Definition
Let XN = (x1, x2, . . . , xn). We say that f(XN ) is symmetric if
f(x1, x2, . . . , xN ) = f(xσ1, . . . , xσN
)
for all σ ∈ SN .
Elementary Symmetric Functions:
ek(XN ) =∑
1≤i1<···<ik≤N
xi1 · · · xik
eλ = eλ1eλ2
· · · eλl
Symmetric Functions
Definition
Let XN = (x1, x2, . . . , xn). We say that f(XN ) is symmetric if
f(x1, x2, . . . , xN ) = f(xσ1, . . . , xσN
)
for all σ ∈ SN .
Elementary Symmetric Functions:
ek(XN ) =∑
1≤i1<···<ik≤N
xi1 · · · xik
eλ = eλ1eλ2
· · · eλl
Homogeneous Symmetric Functions
hk(XN ) =∑
1≤i1≤···≤ik≤N
xi1 · · · xik
hλ = hλ1hλ2
· · ·hλl
Theorem
hn(X) =∑
λ⊢n
(−1)n−l(λ)Bλeλ(X)
where Bλ is the number of compositions that are rearrangements of the
parts of λ.
For example:
h3(X3) = x1x1x1 + x1x1x2 + x1x1x3 + x1x2x2 + x1x2x3
+ x1x3x3 + x2x2x2 + x2x2x3 + x2x3x3 + x3x3x3
= x1x2x3 − 2(x1x2 + x1x3 + x2x3)(x1 + x2 + x3)
+ (x1 + x2 + x3)3
= e3(X3) − 2e2(X3)e1(X3) + e1(X3)3
= e3(X3) − 2e2,1(X3) + e1,1,1(X3)
Theorem
hn(X) =∑
λ⊢n
(−1)n−l(λ)Bλeλ(X)
where Bλ is the number of compositions that are rearrangements of the
parts of λ.
Consider B4,2,2,1 = 12
A Ring Homomorphism
Define ζ(en) = (−1)n−1G(x, t)∣∣∣tn
.
ζ(hn) =∑
λ⊢n
(−1)n−l(λ)Bλζ(eλ(X))
=∑
λ⊢n
(−1)n−l(λ)Bλ
l(λ)∏
i=1
(−1)λi−1G(x, t)∣∣∣tλi
=∑
λ⊢n
Bλ
l(λ)∏
i=1
G(x, t)∣∣∣tλi
An Involution
Find first tile weighted by −x or consecutive bricks with no descent.
1 1 1 −x −x x
An Involution
Find first tile weighted by −x or consecutive bricks with no descent.
1 1 1 −x −x x
l1 1 1 x −x x
An Involution
Find first tile weighted by −x or consecutive bricks with no descent.
1 1 1 −x −x x
l1 1 1 x −x x
Fixed Points:
1 1 1 x 1 1 x 1 1 x 1 1 1 x
An Involution
Find first tile weighted by −x or consecutive bricks with no descent.
1 1 1 −x −x x
l1 1 1 x −x x
Fixed Points:
1 1 1 x 1 1 x 1 1 x 1 1 1 x
Conclusion:
ζ(hn) =∑
T∈Pn
xdes(T )+1
Theorem
∞∑
n=0
hn(X)tn =1
1 +∑
n≥1(−1)nen(X)tn
∞∑
n=0
ζ(hn)tn =1
1 +∑∞
n=1(−1)nζ(en)tn
=1
1 +∑∞
n=1(−1)n(−1)n−1G(x, t)∣∣tn
tn
=1
1 −∑∞
n=1 G(x, t)∣∣tn
tn
=1
1 − (G(x, t) − 1)
=1
2 − G(x, t)
Recap
G(x, t) =1 + 2(2x − 1)t + x(2x − 1)t2 + (x − 1)(2x − 1)t3
(1 − (1 − x)t)(1 − (1 − x)t + (1 − x)t2)
ζ(hn) =∑
T∈Pn
xdes(T )+1
∞∑
n=0
ζ(hn)tn =1
2 − G(x, t)
Theorem
∑
T∈Pn
xdes(T )+1 =1
2 − G(x, t)
∣∣∣tn