Counting lattice walks by winding angle Sminaire de combinatoire - - PowerPoint PPT Presentation

counting lattice walks by winding angle
SMART_READER_LITE
LIVE PREVIEW

Counting lattice walks by winding angle Sminaire de combinatoire - - PowerPoint PPT Presentation

Counting lattice walks by winding angle Sminaire de combinatoire Philippe Flajolet Andrew Elvey Price CNRS, Universit de Tours October 2020 Counting lattice walks by winding angle Andrew Elvey Price L ATTICE WALKS BY WINDING ANGLE The


slide-1
SLIDE 1

Counting lattice walks by winding angle

Séminaire de combinatoire Philippe Flajolet Andrew Elvey Price

CNRS, Université de Tours

October 2020

Counting lattice walks by winding angle Andrew Elvey Price

slide-2
SLIDE 2

LATTICE WALKS BY WINDING ANGLE

The model: count walks starting at by end point and winding angle around . Cell-centred lattices:

Kreweras lattice Triangular Lattice Square Lattice King Lattice

Counting lattice walks by winding angle Andrew Elvey Price

slide-3
SLIDE 3

LATTICE WALKS BY WINDING ANGLE

The model: count walks starting at by end point and winding angle around . Vertex-centred lattices:

Kreweras lattice Triangular Lattice Square Lattice King Lattice

Counting lattice walks by winding angle Andrew Elvey Price

slide-4
SLIDE 4

LATTICE WALKS BY WINDING ANGLE

The model: count walks starting at (by end point). Left: Cell-centred triangular lattice Right: Vertex-centred square lattice

Counting lattice walks by winding angle Andrew Elvey Price

slide-5
SLIDE 5

WHY STUDY WALKS BY WINDING ANGLE?

Physics motivation: Models a long-chain polymer growing in the vicinity of a rod

Bélisle, Berger, Brereton, Butler, Duplantier, Durrett, Faraway, Fisher, Frish, Grosberg, Hu, Le Gall, Privman, Redner, Roberts, Rudnick, Saluer, Shi, Spitzer, . . .

Counting lattice walks by winding angle Andrew Elvey Price

slide-6
SLIDE 6

WHY STUDY WALKS BY WINDING ANGLE?

Physics motivation: Models a long-chain polymer growing in the vicinity of a rod

Bélisle, Berger, Brereton, Butler, Duplantier, Durrett, Faraway, Fisher, Frish, Grosberg, Hu, Le Gall, Privman, Redner, Roberts, Rudnick, Saluer, Shi, Spitzer, . . .

More real world applications:

Counting lattice walks by winding angle Andrew Elvey Price

slide-7
SLIDE 7

SQUARE LATTICE WALKS BY WINDING ANGLE

[Timothy Budd, 2017]: enumeration of square lattice walks (starting and ending on an axis or diagonal) by winding angle Method: Matrices counting paths, eigenvalue decomposition etc. Solution: Jacobi theta function expressions Corollaries:

Square lattice walks in cones (eg. Gessel walks) Loops around the origin (without a fixed starting point) Algebraicity results, asymptotic results, etc.

Counting lattice walks by winding angle Andrew Elvey Price

slide-8
SLIDE 8

SQUARE LATTICE WALKS BY WINDING ANGLE

[Timothy Budd, 2017]: enumeration of square lattice walks (starting and ending on an axis or diagonal) by winding angle Method: Matrices counting paths, eigenvalue decomposition etc. Solution: Jacobi theta function expressions Corollaries:

Square lattice walks in cones (eg. Gessel walks) Loops around the origin (without a fixed starting point) Algebraicity results, asymptotic results, etc.

This work: Completely different method Slightly different set of results Extension to three other lattices

Counting lattice walks by winding angle Andrew Elvey Price

slide-9
SLIDE 9

SQUARE LATTICE WALKS BY WINDING ANGLE

[Timothy Budd, 2017]: enumeration of square lattice walks (starting and ending on an axis or diagonal) by winding angle Method: Matrices counting paths, eigenvalue decomposition etc. Solution: Jacobi theta function expressions Corollaries:

Square lattice walks in cones (eg. Gessel walks) Loops around the origin (without a fixed starting point) Algebraicity results, asymptotic results, etc.

This work: Completely different method Slightly different set of results Extension to three other lattices (more coming)

Counting lattice walks by winding angle Andrew Elvey Price

slide-10
SLIDE 10

SQUARE LATTICE WALKS BY WINDING ANGLE

[Timothy Budd, 2017]: enumeration of square lattice walks (starting and ending on an axis or diagonal) by winding angle Method: Matrices counting paths, eigenvalue decomposition etc. Solution: Jacobi theta function expressions Corollaries:

Square lattice walks in cones (eg. Gessel walks) Loops around the origin (without a fixed starting point) Algebraicity results, asymptotic results, etc.

This work: Completely different method Slightly different set of results Extension to three other lattices (more coming) This talk: Kreweras lattice (mostly)

Counting lattice walks by winding angle Andrew Elvey Price

slide-11
SLIDE 11

JACOBI THETA FUNCTION

All results are in terms of the series: Tk(u, q) =

  • n=0

(−1)n(2n + 1)kqn(n+1)/2(un+1 − (−1)ku−n) = (u ± 1) − 3kq(u2 ± u−1) + 5kq3(u3 ± u−2) + O(q6). Related to Jacobi Theta function ϑ(z, τ) ≡ ϑ11(z, τ) by ϑ(k)(z, τ) ≡ ∂ ∂z k ϑ(z, τ) = e

(πτ−2z)i 2

ikTk(e2iz, e2iπτ).

Counting lattice walks by winding angle Andrew Elvey Price

slide-12
SLIDE 12

PREVIEW: KREWERAS ALMOST-EXCURSIONS

Vertex-centred Kreweras lattice Cell-centred Kreweras lattice

4π 3

Contributes s2t8 to E(t, s) Contributes s−1t6 to ˜ E(t, s)

On each lattice: count walks → ( or ). Walks with length n and winding angle 2πk

3 contribute tnsk.

Cell-centred: E(t, s) = 1 + st +

  • s2 + s−1

t2 + . . . Vertex-centred: ˜ E(t, s) = 1 +

  • s−1 + 4 + s
  • t3 + . . .

Counting lattice walks by winding angle Andrew Elvey Price

slide-13
SLIDE 13

PREVIEW: KREWERAS ALMOST-EXCURSIONS

−2π 3

Vertex-centred Kreweras lattice Cell-centred Kreweras lattice

4π 3

Contributes s2t8 to E(t, s) Contributes s−1t6 to ˜ E(t, s)

On each lattice: count walks → ( or ). Walks with length n and winding angle 2πk

3 contribute tnsk.

Cell-centred: E(t, s) = 1 + st +

  • s2 + s−1

t2 + . . . Vertex-centred: ˜ E(t, s) = 1 +

  • s−1 + 4 + s
  • t3 + . . .

Counting lattice walks by winding angle Andrew Elvey Price

slide-14
SLIDE 14

PREVIEW: KREWERAS ALMOST-EXCURSIONS

Define Tk(u, q) =

  • n=0

(−1)n(2n + 1)kqn(n+1)/2(un+1 − (−1)ku−n) = (u ± 1) − 3kq(u2 ± u−1) + 5kq3(u3 ± u−2) + O(q6). Let q(t) ≡ q = t3 + 15t6 + 279t9 + · · · satisfy t = q1/3 T1(1, q3) 4T0(q, q3) + 6T1(q, q3). The gf for cell-centred Kreweras-lattice almost-excursions is: E(t, s) = s (1 − s3)t

  • s − q−1/3 T1(q2, q3)

T1(1, q3) − q−1/3 T0(q, q3)T1(sq−2/3, q) T1(1, q3)T0(sq−2/3, q)

  • .

The gf for vertex-centred Kreweras-lattice almost-excursions is:

˜ E(t, s) = s(1 − s)q− 2

3

t(1 − s3) T0(q, q3)2 T1(1, q3)2 T1(q, q3)2 T0(q, q3)2 − T2(q, q3) T0(q, q3) − T2(s, q) 2T0(s, q) + T3(1, q) 6T1(1, q) + T3(1, q3) 3T1(1, q3)

  • .

Counting lattice walks by winding angle Andrew Elvey Price

slide-15
SLIDE 15

TALK OUTLINE

Focus: Kreweras lattice (for parts 1 to 4). Part 1: Decomposition of lattice → functional equations Part 2: Solving the functional equations (with theta functions!) Part 3: Corollaries: walks restricted to cones

New result: Excursions with step set ✛✻ ❅ ❘ avoiding a quadrant

Part 4: Analysing the solution

Algebraicity results using modular forms Asymptotic results

Part 5: Square, triangular and king lattices Part 6: Final comments and open problems

Counting lattice walks by winding angle Andrew Elvey Price

slide-16
SLIDE 16

Part 1: Functional equations for Kreweras walks by winding angle

−2π 3

Vertex-centred Kreweras lattice Cell-centred Kreweras lattice

4π 3

Counting lattice walks by winding angle Andrew Elvey Price

slide-17
SLIDE 17

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at by end point and winding around .

Counting lattice walks by winding angle Andrew Elvey Price

slide-18
SLIDE 18

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at by end point.

Counting lattice walks by winding angle Andrew Elvey Price

slide-19
SLIDE 19

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at by end point.

Counting lattice walks by winding angle Andrew Elvey Price

slide-20
SLIDE 20

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at by end point.

×eiα (s) ×e−iα (s−1)

Definition: Q(t, α, x, y) ≡ Q(x, y) =

  • paths p

t|p|xx(p)yy(p)eiαn(p) Note: Q(0, 0) = E(t, eiα)

Counting lattice walks by winding angle Andrew Elvey Price

slide-21
SLIDE 21

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at by end point.

×eiα (s) ×e−iα (s−1)

This example contributes txy. Definition: Q(t, α, x, y) ≡ Q(x, y) =

  • paths p

t|p|xx(p)yy(p)eiαn(p) Note: Q(0, 0) = E(t, eiα)

Counting lattice walks by winding angle Andrew Elvey Price

slide-22
SLIDE 22

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at by end point.

×eiα (s) ×e−iα (s−1)

This example contributes t2y. Definition: Q(t, α, x, y) ≡ Q(x, y) =

  • paths p

t|p|xx(p)yy(p)eiαn(p) Note: Q(0, 0) = E(t, eiα)

Counting lattice walks by winding angle Andrew Elvey Price

slide-23
SLIDE 23

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at by end point.

×eiα (s) ×e−iα (s−1)

This example contributes t3xeiα. Definition: Q(t, α, x, y) ≡ Q(x, y) =

  • paths p

t|p|xx(p)yy(p)eiαn(p) Note: Q(0, 0) = E(t, eiα)

Counting lattice walks by winding angle Andrew Elvey Price

slide-24
SLIDE 24

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at by end point.

×eiα (s) ×e−iα (s−1)

This example contributes t4y2. Definition: Q(t, α, x, y) ≡ Q(x, y) =

  • paths p

t|p|xx(p)yy(p)eiαn(p) Note: Q(0, 0) = E(t, eiα)

Counting lattice walks by winding angle Andrew Elvey Price

slide-25
SLIDE 25

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at by end point.

×eiα (s) ×e−iα (s−1)

This example contributes t5xy3. Definition: Q(t, α, x, y) ≡ Q(x, y) =

  • paths p

t|p|xx(p)yy(p)eiαn(p) Note: Q(0, 0) = E(t, eiα)

Counting lattice walks by winding angle Andrew Elvey Price

slide-26
SLIDE 26

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at by end point.

×eiα (s) ×e−iα (s−1)

This example contributes t6xy2. Definition: Q(t, α, x, y) ≡ Q(x, y) =

  • paths p

t|p|xx(p)yy(p)eiαn(p) Note: Q(0, 0) = E(t, eiα)

Counting lattice walks by winding angle Andrew Elvey Price

slide-27
SLIDE 27

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at by end point.

×eiα (s) ×e−iα (s−1)

This example contributes t7xy. Definition: Q(t, α, x, y) ≡ Q(x, y) =

  • paths p

t|p|xx(p)yy(p)eiαn(p) Note: Q(0, 0) = E(t, eiα)

Counting lattice walks by winding angle Andrew Elvey Price

slide-28
SLIDE 28

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at by end point.

×eiα (s) ×e−iα (s−1)

This example contributes t8x. Definition: Q(t, α, x, y) ≡ Q(x, y) =

  • paths p

t|p|xx(p)yy(p)eiαn(p) Note: Q(0, 0) = E(t, eiα)

Counting lattice walks by winding angle Andrew Elvey Price

slide-29
SLIDE 29

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at by end point.

×eiα (s) ×e−iα (s−1)

This example contributes t9y2e−iα. Definition: Q(t, α, x, y) ≡ Q(x, y) =

  • paths p

t|p|xx(p)yy(p)eiαn(p) Note: Q(0, 0) = E(t, eiα)

Counting lattice walks by winding angle Andrew Elvey Price

slide-30
SLIDE 30

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at by end point.

×eiα (s) ×e−iα (s−1)

This example contributes t10xy3e−iα. Definition: Q(t, α, x, y) ≡ Q(x, y) =

  • paths p

t|p|xx(p)yy(p)eiαn(p) Note: Q(0, 0) = E(t, eiα)

Counting lattice walks by winding angle Andrew Elvey Price

slide-31
SLIDE 31

FUNCTIONAL EQUATION

Recursion → functional equation: separate by type of final step.

Q(x, y) = 1 xytQ(x, y)

t x(Q(x, y) − Q(0, y)) t y(Q(x, y) − Q(x, 0))

+ + +

+ eiαtQ(0, x) + e−iαtyQ(y, 0)

(Final step goes through left wall) (Final step goes through bottom wall)

Counting lattice walks by winding angle Andrew Elvey Price

slide-32
SLIDE 32

KREWERAS WALKS BY WINDING NUMBER

The model: Count walks starting at the red point by end point.

×eiα (s) ×e−iα (s−1)

Definition: Q(t, α, x, y) ≡ Q(x, y) =

  • paths p

t|p|xx(p)yy(p)eiαn(p). Characterised by: Q(x, y) = 1 + txyQ(x, y) + tQ(x, y) − Q(0, y) x + tQ(x, y) − Q(x, 0) y +eiαtQ(0, x) + e−iαtyQ(y, 0).

Counting lattice walks by winding angle Andrew Elvey Price

slide-33
SLIDE 33

Part 2: Solution (using theta functions)

Counting lattice walks by winding angle Andrew Elvey Price

slide-34
SLIDE 34

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: Q(x, y) = 1 + txyQ(x, y) + tQ(x, y) − Q(0, y) x + tQ(x, y) − Q(x, 0) y + eiαtQ(0, x) + e−iαtyQ(y, 0).

Counting lattice walks by winding angle Andrew Elvey Price

slide-35
SLIDE 35

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: Q(x, y) = 1 + txyQ(x, y) + tQ(x, y) − Q(0, y) x + tQ(x, y) − Q(x, 0) y + eiαtQ(0, x) + e−iαtyQ(y, 0). Solution: Step 1: Fix t ∈ [0, 1/3), α ∈ R. All series converge for |x|, |y| < 1.

Counting lattice walks by winding angle Andrew Elvey Price

slide-36
SLIDE 36

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: Q(x, y) = 1 + txyQ(x, y) + tQ(x, y) − Q(0, y) x + tQ(x, y) − Q(x, 0) y + eiαtQ(0, x) + e−iαtyQ(y, 0). Solution: Step 1: Fix t ∈ [0, 1/3), α ∈ R. All series converge for |x|, |y| < 1. Step 2: Write equation as K(x, y)Q(x, y) = R(x, y), where K(x, y) = 1 − txy − t/y − t/x R(x, y) = 1 − t xQ(0, y) − t yQ(x, 0) + eiαtQ(0, x) + e−iαtyQ(y, 0).

Counting lattice walks by winding angle Andrew Elvey Price

slide-37
SLIDE 37

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: Q(x, y) = 1 + txyQ(x, y) + tQ(x, y) − Q(0, y) x + tQ(x, y) − Q(x, 0) y + eiαtQ(0, x) + e−iαtyQ(y, 0). Solution: Step 1: Fix t ∈ [0, 1/3), α ∈ R. All series converge for |x|, |y| < 1. Step 2: Write equation as K(x, y)Q(x, y) = R(x, y), where K(x, y) = 1 − txy − t/y − t/x R(x, y) = 1 − t xQ(0, y) − t yQ(x, 0) + eiαtQ(0, x) + e−iαtyQ(y, 0). Step 3: Consider the curve K(x, y) = 0 (Then R(x, y) = 0).

Counting lattice walks by winding angle Andrew Elvey Price

slide-38
SLIDE 38

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: Q(x, y) = 1 + txyQ(x, y) + tQ(x, y) − Q(0, y) x + tQ(x, y) − Q(x, 0) y + eiαtQ(0, x) + e−iαtyQ(y, 0). Solution: Step 1: Fix t ∈ [0, 1/3), α ∈ R. All series converge for |x|, |y| < 1. Step 2: Write equation as K(x, y)Q(x, y) = R(x, y), where K(x, y) = 1 − txy − t/y − t/x R(x, y) = 1 − t xQ(0, y) − t yQ(x, 0) + eiαtQ(0, x) + e−iαtyQ(y, 0). Step 3: Consider the curve K(x, y) = 0 (Then R(x, y) = 0). Parameterisation involves the Jacobi theta function ϑ(z, τ). So far: Similar to elliptic approaches to quadrant models [Bernardi, Bousquet-Mélou, Fayolle, Iasnogorodski, Kurkova, Malyshev, Raschel, Trotignon]

Counting lattice walks by winding angle Andrew Elvey Price

slide-39
SLIDE 39

JACOBI THETA FUNCTION ϑ(z, τ)

Definition: For τ, z ∈ C, im(τ) > 0, ϑ(z, τ) =

  • n=−∞

(−1)ne( 2n+1

2 ) 2iπτ+(2n+1)iz

Useful facts (for fixed τ): ϑ(z + π, τ) = −ϑ(z, τ) ϑ(z + πτ, τ) = −e−2iz−iπτϑ(z, τ)

Counting lattice walks by winding angle Andrew Elvey Price

slide-40
SLIDE 40

PARAMETERISATION OF K(x, y) = 0 USING ϑ(z, τ)

Definition: For τ, z ∈ C, im(τ) > 0, ϑ(z, τ) =

  • n=−∞

(−1)ne( 2n+1

2 ) 2iπτ+(2n+1)iz

Useful facts (for fixed τ): ϑ(z + π, τ) = −ϑ(z, τ) ϑ(z + πτ, τ) = −e−2iz−iπτϑ(z, τ) Parameterisation: The curve K(x, y) := 1 − txy − t/y − t/x = 0 is parameterised by X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ) and Y(z) = X(z + πτ), where τ is determined by t = e− πτi

3

ϑ′(0, 3τ) 4iϑ(πτ, 3τ) + 6ϑ′(πτ, 3τ).

Counting lattice walks by winding angle Andrew Elvey Price

slide-41
SLIDE 41

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: K(x, y)Q(x, y) = R(x, y), where K(x, y) = 1 − txy − t/y − t/x, R(x, y) = 1 − t xQ(0, y) − t yQ(x, 0) + eiαtQ(0, x) + e−iαtyQ(y, 0).

Counting lattice walks by winding angle Andrew Elvey Price

slide-42
SLIDE 42

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: K(x, y)Q(x, y) = R(x, y), where K(x, y) = 1 − txy − t/y − t/x, R(x, y) = 1 − t xQ(0, y) − t yQ(x, 0) + eiαtQ(0, x) + e−iαtyQ(y, 0). Define X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). Then K(X(z), X(z + πτ)) = 0.

Counting lattice walks by winding angle Andrew Elvey Price

slide-43
SLIDE 43

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: K(x, y)Q(x, y) = R(x, y), where K(x, y) = 1 − txy − t/y − t/x, R(x, y) = 1 − t xQ(0, y) − t yQ(x, 0) + eiαtQ(0, x) + e−iαtyQ(y, 0). Define X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). Then K(X(z), X(z + πτ)) = 0. Hence R(X(z), X(z + πτ)) = 0 (assuming |X(z)| ≤ 1 and |X(z + πτ)| ≤ 1).

Counting lattice walks by winding angle Andrew Elvey Price

slide-44
SLIDE 44

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: K(x, y)Q(x, y) = R(x, y), where K(x, y) = 1 − txy − t/y − t/x, R(x, y) = 1 − t xQ(0, y) − t yQ(x, 0) + eiαtQ(0, x) + e−iαtyQ(y, 0). Define X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). Then K(X(z), X(z + πτ)) = 0. Hence R(X(z), X(z + πτ)) = 0 (assuming |X(z)| ≤ 1 and |X(z + πτ)| ≤ 1). New equation to solve: R(X(z), X(z + πτ)) = 0,

Counting lattice walks by winding angle Andrew Elvey Price

slide-45
SLIDE 45

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: K(x, y)Q(x, y) = R(x, y), where K(x, y) = 1 − txy − t/y − t/x, R(x, y) = 1 − t xQ(0, y) − t yQ(x, 0) + eiαtQ(0, x) + e−iαtyQ(y, 0). Define X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). Then K(X(z), X(z + πτ)) = 0. Hence R(X(z), X(z + πτ)) = 0 (assuming |X(z)| ≤ 1 and |X(z + πτ)| ≤ 1). New equation to solve: R(X(z), X(z + πτ)) = 0,

Counting lattice walks by winding angle Andrew Elvey Price

slide-46
SLIDE 46

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Plot of

  • z : |X(z)| ∈
  • 0, 1

3

  • ,

1 3, 1

  • , (1, 3), (3, 9), (9, ∞]
  • .

Ω π 2π 3π −π πτ −πτ −2πτ 2πτ

For z ∈ Ω, |X(z)| < 1 ⇒ Q(X(z), 0) and Q(0, X(z)) are well defined.

Counting lattice walks by winding angle Andrew Elvey Price

slide-47
SLIDE 47

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Plot of

  • z : |X(z)| ∈
  • 0, 1

3

  • ,

1 3, 1

  • , (1, 3), (3, 9), (9, ∞]
  • .

Ω π 2π 3π −π πτ −πτ −2πτ 2πτ

For z ∈ Ω, |X(z)| < 1 ⇒ Q(X(z), 0) and Q(0, X(z)) are well defined. Near Re(z) = 0, we have z ∈ Ω and z + πτ ∈ Ω.

Counting lattice walks by winding angle Andrew Elvey Price

slide-48
SLIDE 48

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: (near Re(z) = 0) R(X(z), X(z + πτ)) = 0 where X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). R(x, y) = 1 − t xQ(0, y) − t yQ(x, 0) + eiαtQ(0, x) + e−iαtyQ(y, 0).

Counting lattice walks by winding angle Andrew Elvey Price

slide-49
SLIDE 49

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: (near Re(z) = 0) 1 = t X(z)Q(0, X(z + πτ)) + t X(z + πτ)Q(X(z), 0) − eiαtQ(0, X(z)) − e−iαtX(z + πτ)Q(X(z + πτ), 0), where X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ).

Counting lattice walks by winding angle Andrew Elvey Price

slide-50
SLIDE 50

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: (near Re(z) = 0) 1 = t X(z)Q(0, X(z + πτ)) + t X(z + πτ)Q(X(z), 0) − eiαtQ(0, X(z)) − e−iαtX(z + πτ)Q(X(z + πτ), 0), where X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). For z near 0, define L(z) = t X(z + πτ)Q(X(z), 0) − eiαtQ(0, X(z)). Both L(z) and L(z + πτ) converge.

Counting lattice walks by winding angle Andrew Elvey Price

slide-51
SLIDE 51

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: (near Re(z) = 0) 1 = t X(z)Q(0, X(z + πτ))+ t X(z + πτ)Q(X(z), 0) −eiαtQ(0, X(z)) − e−iαtX(z + πτ)Q(X(z + πτ), 0), where X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). For z near 0, define L(z) = t X(z + πτ)Q(X(z), 0) − eiαtQ(0, X(z)). Both L(z) and L(z + πτ) converge.

Counting lattice walks by winding angle Andrew Elvey Price

slide-52
SLIDE 52

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: (near Re(z) = 0) 1 = t X(z)Q(0, X(z + πτ)) + L(z) − e−iαtX(z + πτ)Q(X(z + πτ), 0), where X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). For z near 0, define L(z) = t X(z + πτ)Q(X(z), 0) − eiαtQ(0, X(z)). Both L(z) and L(z + πτ) converge.

Counting lattice walks by winding angle Andrew Elvey Price

slide-53
SLIDE 53

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: (near Re(z) = 0) 1 = t X(z)Q(0, X(z + πτ)) + L(z) − e−iαtX(z + πτ)Q(X(z + πτ), 0), where X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). For z near 0, define L(z) = t X(z + πτ)Q(X(z), 0) − eiαtQ(0, X(z)). Both L(z) and L(z + πτ) converge.

Counting lattice walks by winding angle Andrew Elvey Price

slide-54
SLIDE 54

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: (near Re(z) = 0) 1 = t X(z)Q(0, X(z + πτ)) + L(z) − e−iαtX(z + πτ)Q(X(z + πτ), 0), where X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). For z near 0, define L(z) = t X(z + πτ)Q(X(z), 0) − eiαtQ(0, X(z)). Both L(z) and L(z + πτ) converge.

Counting lattice walks by winding angle Andrew Elvey Price

slide-55
SLIDE 55

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: (near Re(z) = 0) 1 = t X(z)Q(0, X(z + πτ)) + L(z) − e−iαt X(z)X(z + 2πτ)Q(X(z + πτ), 0), where X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). For z near 0, define L(z) = t X(z + πτ)Q(X(z), 0) − eiαtQ(0, X(z)). Both L(z) and L(z + πτ) converge.

Counting lattice walks by winding angle Andrew Elvey Price

slide-56
SLIDE 56

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: (near Re(z) = 0) 1 = t X(z)Q(0, X(z + πτ)) + L(z) − e−iαt X(z)X(z + 2πτ)Q(X(z + πτ), 0), where X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). For z near 0, define L(z) = t X(z + πτ)Q(X(z), 0) − eiαtQ(0, X(z)). Both L(z) and L(z + πτ) converge.

Counting lattice walks by winding angle Andrew Elvey Price

slide-57
SLIDE 57

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: (near Re(z) = 0) 1 = −e−iα X(z)L(z + πτ) + L(z). where X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). For z near 0, define L(z) = t X(z + πτ)Q(X(z), 0) − eiαtQ(0, X(z)). Both L(z) and L(z + πτ) converge.

Counting lattice walks by winding angle Andrew Elvey Price

slide-58
SLIDE 58

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: (near Re(z) = 0) 1 = −e−iα X(z)L(z + πτ) + L(z). where X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). For z near 0, define L(z) = t X(z + πτ)Q(X(z), 0) − eiαtQ(0, X(z)). Both L(z) and L(z + πτ) converge.

Counting lattice walks by winding angle Andrew Elvey Price

slide-59
SLIDE 59

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: (near Re(z) = 0) 1 = −e−iα X(z)L(z + πτ) + L(z). where X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ).

Counting lattice walks by winding angle Andrew Elvey Price

slide-60
SLIDE 60

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: (near Re(z) = 0) 1 = −e−iα X(z)L(z + πτ) + L(z). where X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). We can solve this exactly: L(z) = − e3iα 1 − e3iα

  • 1 + e−iα

X(z) + e−2iαX(z − πτ)

eiα+ 5iπτ

3 ϑ(πτ, 3τ)ϑ′(0, τ)

(1 − e3iα)ϑ( α

2 − 2πτ 3 , τ)ϑ′(0, 3τ)

ϑ(z − 2πτ, 3τ)ϑ(z − α

2 + 2πτ 3 , τ)

ϑ(z, τ)ϑ(z, 3τ)

Counting lattice walks by winding angle Andrew Elvey Price

slide-61
SLIDE 61

SOLUTION TO KREWERAS WALKS BY WINDING NUMBER

Equation to solve: (near Re(z) = 0) 1 = −e−iα X(z)L(z + πτ) + L(z). where X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ). We can solve this exactly: L(z) = − e3iα 1 − e3iα

  • 1 + e−iα

X(z) + e−2iαX(z − πτ)

eiα+ 5iπτ

3 ϑ(πτ, 3τ)ϑ′(0, τ)

(1 − e3iα)ϑ( α

2 − 2πτ 3 , τ)ϑ′(0, 3τ)

ϑ(z − 2πτ, 3τ)ϑ(z − α

2 + 2πτ 3 , τ)

ϑ(z, τ)ϑ(z, 3τ) We can extract E(t, eiα) = Q(0, 0)...

Counting lattice walks by winding angle Andrew Elvey Price

slide-62
SLIDE 62

KREWERAS WALKS BY WINDING NUMBER: SOLUTION

Recall: τ is determined by t = e− πτi

3

ϑ′(0, 3τ) 4iϑ(πτ, 3τ) + 6ϑ′(πτ, 3τ). The gf E(t, eiα) = Q(0, 0) ≡ Q(t, α, 0, 0) is given by:

E(t, eiα) = eiα t(1 − e3iα)

  • eiα − e

4πτi 3

ϑ′(2πτ, 3τ) ϑ′(0, 3τ) − e

πτi 3 ϑ(πτ, 3τ)ϑ′( α

2 − 2πτ 3 , τ)

ϑ′(0, 3τ)ϑ( α

2 − 2πτ 3 , τ)

  • .

Counting lattice walks by winding angle Andrew Elvey Price

slide-63
SLIDE 63

KREWERAS WALKS BY WINDING NUMBER: SOLUTION

Recall: τ is determined by t = e− πτi

3

ϑ′(0, 3τ) 4iϑ(πτ, 3τ) + 6ϑ′(πτ, 3τ). The gf E(t, eiα) = Q(0, 0) ≡ Q(t, α, 0, 0) is given by:

E(t, eiα) = eiα t(1 − e3iα)

  • eiα − e

4πτi 3

ϑ′(2πτ, 3τ) ϑ′(0, 3τ) − e

πτi 3 ϑ(πτ, 3τ)ϑ′( α

2 − 2πτ 3 , τ)

ϑ′(0, 3τ)ϑ( α

2 − 2πτ 3 , τ)

  • .

Equivalently: Let q(t) ≡ q = t3 + 15t6 + 279t9 + · · · satisfy t = q1/3 T1(1, q3) 4T0(q, q3) + 6T1(q, q3). The gf for cell-centred Kreweras-lattice almost-excursions is: E(t, s) = s (1 − s3)t

  • s − q−1/3 T1(q2, q3)

T1(1, q3) − q−1/3 T0(q, q3)T1(sq−2/3, q) T1(1, q3)T0(sq−2/3, q)

  • .

Counting lattice walks by winding angle Andrew Elvey Price

slide-64
SLIDE 64

Part 3: Walks in cones

Counting lattice walks by winding angle Andrew Elvey Price

slide-65
SLIDE 65

WALKS IN CONES WITH SMALL STEPS

Quarter plane walks: Completely classified into rational, algebraic, D-finite, D-algebraic cases.

[Mishna, Rechnitzer 09], [Bousquet-Mélou, Mishna 10], [Bostan, Kauers 10], [Fayolle, Raschel 10], [Kurkova, Raschel 12], [Melczer, Mishna 13], [Bostan, Raschel, Salvy 14], [Bernardi, Bousquet-Mélou, Raschel 17], [Dreyfus, Hardouin, Roques, Singer 18]

Half plane walks: Easy Three quarter plane walks: Active area of research (Previously) solved in 6-12 of the 74 non-trivial cases

[Bousquet-Mélou 16], [Raschel-Trotignon 19], [Budd 20], [Bousquet-Mélou, Wallner 20+]

Walks on the slit plane C \ R<0: solved in all cases

[Bousquet-Mélou, 01], [Bousquet-Mélou, Schaeffer, 02], [Rubey 05]

Counting lattice walks by winding angle Andrew Elvey Price

slide-66
SLIDE 66

WALKS IN THE 3/4-PLANE: SOLVED CASES

Not D-finite D-finite [Budd 20] [B-M 16] [B-M, W 20+] This work [R,T 19] [D,T 20]

[Bousquet-Mélou 16],[Raschel, Trotignon 19], [Budd 20], [Bousquet-Mélou, Wallner 20+]

Counting lattice walks by winding angle Andrew Elvey Price

slide-67
SLIDE 67

WALKS IN THE 5/4-PLANE: SOLVED CASES

D-finite [Budd 20] This work

[Budd 20]

Counting lattice walks by winding angle Andrew Elvey Price

slide-68
SLIDE 68

WALKS IN THE 6/4-PLANE: SOLVED CASES

D-finite [Budd 20] This work

[Budd 20]

Counting lattice walks by winding angle Andrew Elvey Price

slide-69
SLIDE 69

WALKS IN THE 7/4-PLANE: SOLVED CASES

D-finite [Budd 20] This work

[Budd 20]

Counting lattice walks by winding angle Andrew Elvey Price

slide-70
SLIDE 70

COUNTING KREWERAS WALKS IN A CONE

B B A R

In the upper half plane: Use reflection principle #(Walks from A to B above R) = #(Walks from A to B) − #(Walks from A to B through R) = #(Walks from A to B) − #(Walks from A to B)

Counting lattice walks by winding angle Andrew Elvey Price

slide-71
SLIDE 71

COUNTING KREWERAS EXCURSIONS IN 5/6-PLANE

New result:

✛✻ ❅ ❘-excursions avoiding a quadrant.

Counting lattice walks by winding angle Andrew Elvey Price

slide-72
SLIDE 72

COUNTING KREWERAS EXCURSIONS IN 5/6-PLANE

New result:

✛✻ ❅ ❘-excursions avoiding a quadrant.

Equivalently: Walks avoiding the blue and green lines

Counting lattice walks by winding angle Andrew Elvey Price

slide-73
SLIDE 73

COUNTING KREWERAS EXCURSIONS IN 5/6-PLANE

New result:

✛✻ ❅ ❘-excursions avoiding a quadrant.

Equivalently: Walks avoiding the blue and green lines

Counting lattice walks by winding angle Andrew Elvey Price

slide-74
SLIDE 74

COUNTING KREWERAS EXCURSIONS IN 5/6-PLANE

New result:

✛✻ ❅ ❘-excursions avoiding a quadrant.

Equivalently: Walks avoiding the blue and green lines

Reflection principle: For walks passing through at least one such line: reflect walk after first intersection. Winding angle α → − 4π

3 − α or 2π − α.

Counting lattice walks by winding angle Andrew Elvey Price

slide-75
SLIDE 75

COUNTING KREWERAS EXCURSIONS IN 5/6-PLANE

New result:

✛✻ ❅ ❘-excursions avoiding a quadrant.

Equivalently: Walks avoiding the blue and green lines

Reflection principle: For walks passing through at least one such line: reflect walk after first intersection. Winding angle α → − 4π

3 − α or 2π − α.

Counting lattice walks by winding angle Andrew Elvey Price

slide-76
SLIDE 76

COUNTING KREWERAS EXCURSIONS IN 5/6-PLANE

New result:

✛✻ ❅ ❘-excursions avoiding a quadrant.

Equivalently: Walks avoiding the blue and green lines

Reflection principle: For walks passing through at least one such line: reflect walk after first intersection. Winding angle α → − 4π

3 − α or 2π − α.

Winding angle 10πk

3

→ − 4π

3 + 10πj 3 .

#(Walks → avoiding lines) =

  • k∈Z

[s5k]˜ E(t, s)

  • k∈Z

[s5k−2]˜ E(t, s)

  • = 1

5

4

  • j=1
  • 1 − e

4πij 5

  • ˜

E

  • t, e

2πi 5

  • Counting lattice walks by winding angle

Andrew Elvey Price

slide-77
SLIDE 77

COUNTING KREWERAS EXCURSIONS IN 5/6-PLANE

New result:

✛✻ ❅ ❘-excursions avoiding a quadrant.

Equivalently: Walks avoiding the blue and green lines

Reflection principle: For walks passing through at least one such line: reflect walk after first intersection. Winding angle α → − 4π

3 − α or 2π − α.

Winding angle 10πk

3

→ − 4π

3 + 10πj 3 .

#(Walks → avoiding lines) =

  • k∈Z

[s5k]˜ E(t, s)

  • k∈Z

[s5k−2]˜ E(t, s)

  • = 1

5

4

  • j=1
  • 1 − e

4πij 5

  • ˜

E

  • t, e

2πi 5

  • Counting lattice walks by winding angle

Andrew Elvey Price

slide-78
SLIDE 78

COUNTING KREWERAS EXCURSIONS IN k/6-PLANE

More generally: Let Ck,r(t) count whole-plane Kreweras excursions... Starting adjacent to the origin, Avoiding the origin, Having winding angle 0, Having intermediate winding angles restricted to

  • − rπ

3 , (k−r)π 3

  • I.e., Kreweras excursions in the k/6-plane

Counting lattice walks by winding angle Andrew Elvey Price

slide-79
SLIDE 79

COUNTING KREWERAS EXCURSIONS IN k/6-PLANE

More generally: Let Ck,r(t) count whole-plane Kreweras excursions... Starting adjacent to the origin, Avoiding the origin, Having winding angle 0, Having intermediate winding angles restricted to

  • − rπ

3 , (k−r)π 3

  • I.e., Kreweras excursions in the k/6-plane

Previous slide: C5,2(t) = 1 5

4

  • j=1
  • 1 − e

4πij 5

  • ˜

E

  • t, e

2πi 5

  • .

More generally: Ck,r(t) = 1 k

k−1

  • j=1
  • 1 − e

2πijr k

  • ˜

E

  • t, e

2πij k

  • .

Counting lattice walks by winding angle Andrew Elvey Price

slide-80
SLIDE 80

Part 4: Analysis of solutions

Counting lattice walks by winding angle Andrew Elvey Price

slide-81
SLIDE 81

ANALYSIS OF SOLUTION

From the exact solution we extract: Asymptotic distribution ([Bélisle, 1989]): For random excursions of length n, winding angle

c log(n)

has asymptotic density 4(x − 1)ex + (x + 1)e−x (ex − e−x)2 . Asymptotics ([Denisov, Wachtel, 2015]): Let cn count Kreweras-lattice excursions in a cone of angle α ∈ π

3 N.

cn ∼ − 2 · 35− 6

k sin2 π

k

  • πk2

1 + 2 cos 2π

k

  • Γ
  • − 3

k

n−1− 3

k 3n.

Conditions for algebraicity: Let Cα(t) count Kreweras-lattice excursions in a cone of angle α ∈ π

3 N. This satisfies a non-trivial

polynomial equation P(Cα(t), t) = 0 if and only if α / ∈ πZ.

(uses modular forms as in [Zagier, 08] and [E.P., Zinn-Justin, 20])

Counting lattice walks by winding angle Andrew Elvey Price

slide-82
SLIDE 82

ANALYSIS OF SOLUTION: ASYMPTOTICS

Fix α. Writing ˆ τ = − 1

3τ and ˆ

q = e2πiˆ

τ, the dominant singularity t = 1/3 of

˜ E(t, eiα) corresponds to ˆ q = 0.

Counting lattice walks by winding angle Andrew Elvey Price

slide-83
SLIDE 83

ANALYSIS OF SOLUTION: ASYMPTOTICS

Fix α. Writing ˆ τ = − 1

3τ and ˆ

q = e2πiˆ

τ, the dominant singularity t = 1/3 of

˜ E(t, eiα) corresponds to ˆ q = 0. Series in ˆ q:

t = 1 3 − 3ˆ q + 18ˆ q2 + O(ˆ q3) t˜ E(t, eiα) = a0 + a1ˆ q − 27αeiα 2π(1 + eiα + e2iα)ˆ q

3α 2π + o

  • ˆ

q

3α 2π

  • ,

→ ˜ E(t, eiα) as a series in (1 − 3t),

Counting lattice walks by winding angle Andrew Elvey Price

slide-84
SLIDE 84

ANALYSIS OF SOLUTION: ASYMPTOTICS

Fix α. Writing ˆ τ = − 1

3τ and ˆ

q = e2πiˆ

τ, the dominant singularity t = 1/3 of

˜ E(t, eiα) corresponds to ˆ q = 0. Series in ˆ q:

t = 1 3 − 3ˆ q + 18ˆ q2 + O(ˆ q3) t˜ E(t, eiα) = a0 + a1ˆ q − 27αeiα 2π(1 + eiα + e2iα)ˆ q

3α 2π + o

  • ˆ

q

3α 2π

  • ,

→ ˜ E(t, eiα) as a series in (1 − 3t), → [tn]˜ E(t, eiα) ∼ − 35− 3α

π eαiα

2π(1 + eαi + e2αi)Γ

  • − 3α

n− 3α

2π −13n,

[tn]Ck,r(t) ∼ − 2 · 35− 6

k sin2 rπ

k

  • πk2

1 + 2 cos 2π

k

  • Γ
  • − 3

k

n−1− 3

k 3n. Counting lattice walks by winding angle Andrew Elvey Price

slide-85
SLIDE 85

ANALYSIS OF SOLUTION: ASYMPTOTICS

Fix α. Writing ˆ τ = − 1

3τ and ˆ

q = e2πiˆ

τ, the dominant singularity t = 1/3 of

˜ E(t, eiα) corresponds to ˆ q = 0. Series in ˆ q:

t = 1 3 − 3ˆ q + 18ˆ q2 + O(ˆ q3) t˜ E(t, eiα) = a0 + a1ˆ q − 27αeiα 2π(1 + eiα + e2iα)ˆ q

3α 2π + o

  • ˆ

q

3α 2π

  • ,

→ ˜ E(t, eiα) as a series in (1 − 3t), → [tn]˜ E(t, eiα) ∼ − 35− 3α

π eαiα

2π(1 + eαi + e2αi)Γ

  • − 3α

n− 3α

2π −13n,

[tn]Ck,r(t) ∼ − 2 · 35− 6

k sin2 rπ

k

  • πk2

1 + 2 cos 2π

k

  • Γ
  • − 3

k

n−1− 3

k 3n.

Previously: Terms 3n and n−1− 3

k known [Denisov, Wachtel, 2015]. Counting lattice walks by winding angle Andrew Elvey Price

slide-86
SLIDE 86

ANALYSIS OF SOLUTION: ALGEBRAICITY

Recall: ϑ(z, τ) is differentially algebraic → so are ˜ E(t, s) and Q(t, α, x, y). For α ∈ π

3 (Q \ Z) we get algebraicity (Ideas from [Zagier, 08] and

[E.P., Zinn-Justin, 20+]):

Counting lattice walks by winding angle Andrew Elvey Price

slide-87
SLIDE 87

ANALYSIS OF SOLUTION: ALGEBRAICITY

Recall: ϑ(z, τ) is differentially algebraic → so are ˜ E(t, s) and Q(t, α, x, y). For α ∈ π

3 (Q \ Z) we get algebraicity (Ideas from [Zagier, 08] and

[E.P., Zinn-Justin, 20+]): Q(t, α, X(z), 0) and X(z) are elliptic functions with the same periods ⇒ Q(t, α, x, 0) is algebraic in x.

Counting lattice walks by winding angle Andrew Elvey Price

slide-88
SLIDE 88

ANALYSIS OF SOLUTION: ALGEBRAICITY

Recall: ϑ(z, τ) is differentially algebraic → so are ˜ E(t, s) and Q(t, α, x, y). For α ∈ π

3 (Q \ Z) we get algebraicity (Ideas from [Zagier, 08] and

[E.P., Zinn-Justin, 20+]): Q(t, α, X(z), 0) and X(z) are elliptic functions with the same periods ⇒ Q(t, α, x, 0) is algebraic in x. E(t(τ), eiα) and t(τ) are modular functions of τ ⇒ E(t, eiα) is algebraic in t. Same for ˜ E(t(τ), eiα).

Counting lattice walks by winding angle Andrew Elvey Price

slide-89
SLIDE 89

ANALYSIS OF SOLUTION: ALGEBRAICITY

Recall: ϑ(z, τ) is differentially algebraic → so are ˜ E(t, s) and Q(t, α, x, y). For α ∈ π

3 (Q \ Z) we get algebraicity (Ideas from [Zagier, 08] and

[E.P., Zinn-Justin, 20+]): Q(t, α, X(z), 0) and X(z) are elliptic functions with the same periods ⇒ Q(t, α, x, 0) is algebraic in x. E(t(τ), eiα) and t(τ) are modular functions of τ ⇒ E(t, eiα) is algebraic in t. Same for ˜ E(t(τ), eiα). Combining these ideas: Q(t, α, x, y) is algebraic in t, x and y.

Counting lattice walks by winding angle Andrew Elvey Price

slide-90
SLIDE 90

ANALYSIS OF SOLUTION: ALGEBRAICITY

Recall: ϑ(z, τ) is differentially algebraic → so are ˜ E(t, s) and Q(t, α, x, y). For α ∈ π

3 (Q \ Z) we get algebraicity (Ideas from [Zagier, 08] and

[E.P., Zinn-Justin, 20+]): Q(t, α, X(z), 0) and X(z) are elliptic functions with the same periods ⇒ Q(t, α, x, 0) is algebraic in x. E(t(τ), eiα) and t(τ) are modular functions of τ ⇒ E(t, eiα) is algebraic in t. Same for ˜ E(t(τ), eiα). Combining these ideas: Q(t, α, x, y) is algebraic in t, x and y. Recall: The gf for excursions in the k/6-plane is Ck,r(t) = 1 k

k−1

  • j=1
  • 1 − e

2πijr k

  • ˜

E

  • t, e

2πij k

  • .

Counting lattice walks by winding angle Andrew Elvey Price

slide-91
SLIDE 91

ANALYSIS OF SOLUTION: ALGEBRAICITY

Recall: ϑ(z, τ) is differentially algebraic → so are ˜ E(t, s) and Q(t, α, x, y). For α ∈ π

3 (Q \ Z) we get algebraicity (Ideas from [Zagier, 08] and

[E.P., Zinn-Justin, 20+]): Q(t, α, X(z), 0) and X(z) are elliptic functions with the same periods ⇒ Q(t, α, x, 0) is algebraic in x. E(t(τ), eiα) and t(τ) are modular functions of τ ⇒ E(t, eiα) is algebraic in t. Same for ˜ E(t(τ), eiα). Combining these ideas: Q(t, α, x, y) is algebraic in t, x and y. Recall: The gf for excursions in the k/6-plane is Ck,r(t) = 1 k

k−1

  • j=1
  • 1 − e

2πijr k

  • ˜

E

  • t, e

2πij k

  • .

Algebraic iff 3 ∤ k. (always D-finite).

Counting lattice walks by winding angle Andrew Elvey Price

slide-92
SLIDE 92

Part 5: Other lattices

Kreweras lattice Triangular Lattice Square Lattice King Lattice

Counting lattice walks by winding angle Andrew Elvey Price

slide-93
SLIDE 93

CELL-CENTRED LATTICES

Important property: Decomposable into congruent sectors

Kreweras lattice Triangular Lattice Square Lattice King Lattice

Counting lattice walks by winding angle Andrew Elvey Price

slide-94
SLIDE 94

CELL-CENTRED LATTICES

Important property: Decomposable into congruent sectors

Kreweras lattice Triangular Lattice Square Lattice King Lattice

Counting lattice walks by winding angle Andrew Elvey Price

slide-95
SLIDE 95

VERTEX-CENTRED LATTICES

Decompose into rotationally congruent sectors

Kreweras lattice Triangular Lattice Square Lattice King Lattice

Counting lattice walks by winding angle Andrew Elvey Price

slide-96
SLIDE 96

VERTEX-CENTRED LATTICES

Decompose into rotationally congruent sectors

Kreweras lattice Triangular Lattice Square Lattice King Lattice

Counting lattice walks by winding angle Andrew Elvey Price

slide-97
SLIDE 97

RECALL: KREWERAS ALMOST-EXCURSIONS

Define Tk(u, q) =

  • n=0

(−1)n(2n + 1)kqn(n+1)/2(un+1 − (−1)ku−n) = (u ± 1) − 3kq(u2 ± u−1) + 5kq3(u3 ± u−2) + O(q6). Let q(t) ≡ q = t3 + 15t6 + 279t9 + · · · satisfy t = q1/3 T1(1, q3) 4T0(q, q3) + 6T1(q, q3). The gf for cell-centred Kreweras-lattice almost-excursions is: E(t, s) = s (1 − s3)t

  • s − q−1/3 T1(q2, q3)

T1(1, q3) − q−1/3 T0(q, q3)T1(sq−2/3, q) T1(1, q3)T0(sq−2/3, q)

  • .

The gf for vertex-centred Kreweras-lattice almost-excursions is:

˜ E(t, s) = s(1 − s)q− 2

3

t(1 − s3) T0(q, q3)2 T1(1, q3)2 T1(q, q3)2 T0(q, q3)2 − T2(q, q3) T0(q, q3) − T2(s, q) 2T0(s, q) + T3(1, q) 6T1(1, q) + T3(1, q3) 3T1(1, q3)

  • .

Counting lattice walks by winding angle Andrew Elvey Price

slide-98
SLIDE 98

SQUARE LATTICE ALMOST-EXCURSIONS

Define Tk(u, q) =

  • n=0

(−1)n(2n + 1)kqn(n+1)/2(un+1 − (−1)ku−n) = (u ± 1) − 3kq(u2 ± u−1) + 5kq3(u3 ± u−2) + O(q6). Let q(t) ≡ q = t + 4t3 + 34t5 + 360t7 + · · · satisfy t = qT0(q2, q8)T1(1, q8) 2T0(q4, q8)(T0(q2, q8) + 2T1(q2, q8)). The gf for cell-centred Square-lattice almost-excursions is: s2 (1 − s4)t

  • s − s−1 + T0(q4, q8)

qT1(1, q8) − T0(q4, q8)T1(s−1q, q2) qT1(1, q8)T0(s−1q, q2)

  • .

The gf for vertex-centred Square-lattice almost-excursions is: sT0(q4, q8) qt(1 + s2)T1(1, q8)

  • 1 + 2T1(q2, q8)

T0(q2, q8) + (1 − s)T1(s−1, q2) (1 + s)T0(s−1, q2)

  • .

Counting lattice walks by winding angle Andrew Elvey Price

slide-99
SLIDE 99

Part 6: Final comments

Counting lattice walks by winding angle Andrew Elvey Price

slide-100
SLIDE 100

JACOBI THETA FUNCTION/ WEIERSTRASS FUNCTION

PARAMETERISATION COMBINATORIAL FUNCTIONAL EQUATION SOLUTION METHOD

Counting lattice walks by winding angle Andrew Elvey Price

slide-101
SLIDE 101

JACOBI THETA FUNCTION/ WEIERSTRASS FUNCTION

PARAMETERISATION COMBINATORIAL FUNCTIONAL EQUATION SOLUTION METHOD

This method... Sometimes works on equations with two catalytic variables Successful on

Various 2 dimensional lattice walk models [Bernardi, Bousquet-Mélou, E.P., Fayolle, Kurkova, Raschel, Trotignon] Some planar map models [Bousquet Mélou, E.P., Kostov, Zinn-Justin].

Questions for the audience: Does anyone have a nice equation to try? Can anyone suggest a better name for the method?

Counting lattice walks by winding angle Andrew Elvey Price

slide-102
SLIDE 102

JACOBI THETA FUNCTION/ WEIERSTRASS FUNCTION

PARAMETERISATION COMBINATORIAL FUNCTIONAL EQUATION SOLUTION METHOD

This method... Sometimes works on equations with two catalytic variables Successful on

Various 2 dimensional lattice walk models [Bernardi, Bousquet-Mélou, E.P., Fayolle, Kurkova, Raschel, Trotignon] Some planar map models [Bousquet Mélou, E.P., Kostov, Zinn-Justin].

Questions for the audience: Does anyone have a nice equation to try? Can anyone suggest a better name for the method?

Thank you!

Counting lattice walks by winding angle Andrew Elvey Price

slide-103
SLIDE 103

Thank you!

Counting lattice walks by winding angle Andrew Elvey Price

slide-104
SLIDE 104

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

Write K(x, y) = A(x)y2 + B(x)y + C(x), then Y(x) = −B(x) ±

  • B(x)2 − 4A(x)C(x)

2A(x) parameterizes K(x, Y(x)) = 0. Typically, Y+(x) is meromorphic on:

Counting lattice walks by winding angle Andrew Elvey Price

slide-105
SLIDE 105

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

Write K(x, y) = A(x)y2 + B(x)y + C(x), then Y(x) = −B(x) ±

  • B(x)2 − 4A(x)C(x)

2A(x) parameterizes K(x, Y(x)) = 0. Typically, Y+(x) is meromorphic on:

x3 x1

πτ 2

π π + πτ

2 X(z)

x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-106
SLIDE 106

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

Write K(x, y) = A(x)y2 + B(x)y + C(x), then Y(x) = −B(x) ±

  • B(x)2 − 4A(x)C(x)

2A(x) parameterizes K(x, Y(x)) = 0. Typically, Y+(x) is meromorphic on:

x3 x1

πτ 2

π π + πτ

2 X(z)

x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-107
SLIDE 107

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

x3 x1

πτ 2

π π + πτ

2

x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-108
SLIDE 108

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x3 x1 x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-109
SLIDE 109

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x3 x1 x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-110
SLIDE 110

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x3 x1 x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-111
SLIDE 111

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x3 x1 x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-112
SLIDE 112

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x3 x1 x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-113
SLIDE 113

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x3 x1 x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-114
SLIDE 114

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x3 x1 x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-115
SLIDE 115

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x3 x1 x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-116
SLIDE 116

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x3 x1 x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-117
SLIDE 117

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x3 x1 x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-118
SLIDE 118

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x3 x1 x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-119
SLIDE 119

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x3 x1 x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-120
SLIDE 120

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x3 x1 x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-121
SLIDE 121

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x3 x1 x2 x4

Counting lattice walks by winding angle Andrew Elvey Price

slide-122
SLIDE 122

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x1 x4 x3 x2

Counting lattice walks by winding angle Andrew Elvey Price

slide-123
SLIDE 123

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x4 x3

Counting lattice walks by winding angle Andrew Elvey Price

slide-124
SLIDE 124

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x4 x3

Counting lattice walks by winding angle Andrew Elvey Price

slide-125
SLIDE 125

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

π + πτ

2

x4 x3

Counting lattice walks by winding angle Andrew Elvey Price

slide-126
SLIDE 126

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

πτ 2

π π + πτ

2 X(z)

Counting lattice walks by winding angle Andrew Elvey Price

slide-127
SLIDE 127

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

X π+πτ

2

  • X(0)

πτ 2

π π + πτ

2 X(z) X π

2

  • X

πτ

2

  • π+πτ

2 π 2

Counting lattice walks by winding angle Andrew Elvey Price

slide-128
SLIDE 128

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

X π+πτ

2

  • X(0)

πτ 2

π π + πτ

2 X(z) X π

2

  • X

πτ

2

  • π+πτ

2 π 2

By symmetry, for r ∈ R: X(r) = X(π − r) = X(−r) X( πτ

2 + r) = X( πτ 2 − r)

Counting lattice walks by winding angle Andrew Elvey Price

slide-129
SLIDE 129

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

X π+πτ

2

  • X(0)

πτ 2

π π + πτ

2 X(z) X π

2

  • X

πτ

2

  • π+πτ

2 π 2

For z ∈ C: X(z) = X(π − z) = X(−z) X(z) = X(πτ − z)

Counting lattice walks by winding angle Andrew Elvey Price

slide-130
SLIDE 130

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

X π+πτ

2

  • X(0)

πτ 2

π π + πτ

2 X(z) X π

2

  • X

πτ

2

  • π+πτ

2 π 2

For z ∈ C: X(z) = X(π − z) = X(−z) = X(πτ + z) X(z) = X(πτ − z)

Counting lattice walks by winding angle Andrew Elvey Price

slide-131
SLIDE 131

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

X π+πτ

2

  • X(0)

πτ 2

π π + πτ

2 X(z) X π

2

  • X

πτ

2

  • π+πτ

2 π 2

For z ∈ C: X(z) = X(π − z) = X(−z) = X(πτ + z) X(z) = cϑ(z − α)ϑ(z + α) ϑ(z − β)ϑ(z + β)

Counting lattice walks by winding angle Andrew Elvey Price

slide-132
SLIDE 132

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

X π+πτ

2

  • X(0)

πτ 2

π π + πτ

2 X(z) X π

2

  • X

πτ

2

  • π+πτ

2 π 2

Recall: y(x) = −B(x) ±

  • B(x)2 − 4A(x)C(x)

2A(x) . Consider Y(z) = y(X(z)). By symmetry, for r ∈ R: X(r) = X(−r), so Y(r) + Y(−r) = − B(X(r))

A(X(r)).

Similarly, Y πτ 2 + r

  • + Y

πτ 2 − r

  • = −B
  • X

πτ

2 + r

  • A
  • X

πτ

2 + r

.

Counting lattice walks by winding angle Andrew Elvey Price

slide-133
SLIDE 133

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

X π+πτ

2

  • X(0)

πτ 2

π π + πτ

2 X(z) X π

2

  • X

πτ

2

  • π+πτ

2 π 2

Recall: y(x) = −B(x) ±

  • B(x)2 − 4A(x)C(x)

2A(x) . Consider Y(z) = y(X(z)). For z ∈ C: Y(z) + Y(−z) = − B(X(z))

A(X(z)).

Y (z) + Y (πτ − z) = −B (X (z)) A (X (z)).

Counting lattice walks by winding angle Andrew Elvey Price

slide-134
SLIDE 134

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

X π+πτ

2

  • X(0)

πτ 2

π π + πτ

2 X(z) X π

2

  • X

πτ

2

  • π+πτ

2 π 2

For z ∈ C: Y(z) + Y(−z) = − B(X(z))

A(X(z)).

Y (z) + Y (πτ − z) = −B (X (z)) A (X (z)). So Y(z) = Y(z + πτ) = Y(z + π) ⇒ Y(z) = c ϑ(z − γ)ϑ(z − δ) ϑ(z − ǫ)ϑ(z − γ − δ + ǫ).

Counting lattice walks by winding angle Andrew Elvey Price

slide-135
SLIDE 135

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

Equation characterising Q(x, y) ≡ Q(t, x, y) for quadrant walks: K(x, y)Q(x, y) + R(x, y) = 0. K(x, y) = 0 is parameterised by X(z) = c1 ϑ(z − α1)ϑ(z − β1) ϑ(z − γ1)ϑ(z − δ1) and Y(z) = c2 ϑ(z − α2)ϑ(z − β2) ϑ(z − γ2)ϑ(z − δ2) , where the constants satisfy αj + βj = γj + δj for j = 1, 2. So, R(X(z), Y(z)) = 0.

Counting lattice walks by winding angle Andrew Elvey Price

slide-136
SLIDE 136

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

In general: K(x, y) = 0 is parameterised by X(z) = c1 ϑ(z − α1)ϑ(z − β1) ϑ(z − γ1)ϑ(z − δ1) and Y(z) = c2 ϑ(z − α2)ϑ(z − β2) ϑ(z − γ2)ϑ(z − δ2) , with αj + βj = γj + δj for j = 1, 2.

Counting lattice walks by winding angle Andrew Elvey Price

slide-137
SLIDE 137

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

For Kreweras paths: Q(x, y) = 1+xytQ(x, y)+ t x (Q(x, y) − Q(0, y))+ t y (Q(x, y) − Q(x, 0)) . Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = c1 ϑ(z − α1)ϑ(z − β1) ϑ(z − γ1)ϑ(z − δ1) and Y(z) = c2 ϑ(z − α2)ϑ(z − β2) ϑ(z − γ2)ϑ(z − δ2) , with αj + βj = γj + δj for j = 1, 2.

Counting lattice walks by winding angle Andrew Elvey Price

slide-138
SLIDE 138

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

For Kreweras paths: Q(x, y) = 1+xytQ(x, y)+ t x (Q(x, y) − Q(0, y))+ t y (Q(x, y) − Q(x, 0)) . Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = c1 ϑ(z − α1)ϑ(z − β1) ϑ(z − γ1)ϑ(z − δ1) and Y(z) = c2 ϑ(z − α2)ϑ(z − β2) ϑ(z − γ2)ϑ(z − δ2) , with αj + βj = γj + δj for j = 1, 2. K(0, 0) = 0, so WLOG α1 = α2 = 0.

Counting lattice walks by winding angle Andrew Elvey Price

slide-139
SLIDE 139

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

For Kreweras paths: Q(x, y) = 1+xytQ(x, y)+ t x (Q(x, y) − Q(0, y))+ t y (Q(x, y) − Q(x, 0)) . Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = c1 ϑ(z)ϑ(z − β1) ϑ(z − γ1)ϑ(z − δ1) and Y(z) = c2 ϑ(z)ϑ(z − β2) ϑ(z − γ2)ϑ(z − δ2), with αj + βj = γj + δj for j = 1, 2. K(0, 0) = 0, so WLOG α1 = α2 = 0.

Counting lattice walks by winding angle Andrew Elvey Price

slide-140
SLIDE 140

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

For Kreweras paths: Q(x, y) = 1+xytQ(x, y)+ t x (Q(x, y) − Q(0, y))+ t y (Q(x, y) − Q(x, 0)) . Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = c1 ϑ(z)ϑ(z − β1) ϑ(z − γ1)ϑ(z − δ1) and Y(z) = c2 ϑ(z)ϑ(z − β2) ϑ(z − γ2)ϑ(z − δ2), with αj + βj = γj + δj for j = 1, 2. K(0, 0) = 0, so WLOG α1 = α2 = 0. as x → 0, we have y(x) ∼ −x or y(x) ∼ − 1

x2 , so Y(z) has a

double pole at z = β1.

Counting lattice walks by winding angle Andrew Elvey Price

slide-141
SLIDE 141

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

For Kreweras paths: Q(x, y) = 1+xytQ(x, y)+ t x (Q(x, y) − Q(0, y))+ t y (Q(x, y) − Q(x, 0)) . Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = c1 ϑ(z)ϑ(z − β1) ϑ(z − γ1)ϑ(z − δ1) and Y(z) = c2 ϑ(z)ϑ(z − β2) ϑ(z − β1)2 , with αj + βj = γj + δj for j = 1, 2. K(0, 0) = 0, so WLOG α1 = α2 = 0. as x → 0, we have y(x) ∼ −x or y(x) ∼ − 1

x2 , so Y(z) has a

double pole at z = β1.

Counting lattice walks by winding angle Andrew Elvey Price

slide-142
SLIDE 142

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

For Kreweras paths: Q(x, y) = 1+xytQ(x, y)+ t x (Q(x, y) − Q(0, y))+ t y (Q(x, y) − Q(x, 0)) . Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = c1 ϑ(z)ϑ(z − β1) ϑ(z − γ1)ϑ(z − δ1) and Y(z) = c2 ϑ(z)ϑ(z − 2β1) ϑ(z − β1)2 , with αj + βj = γj + δj for j = 1, 2. K(0, 0) = 0, so WLOG α1 = α2 = 0. as x → 0, we have y(x) ∼ −x or y(x) ∼ − 1

x2 , so Y(z) has a

double pole at z = β1.

Counting lattice walks by winding angle Andrew Elvey Price

slide-143
SLIDE 143

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

For Kreweras paths: Q(x, y) = 1+xytQ(x, y)+ t x (Q(x, y) − Q(0, y))+ t y (Q(x, y) − Q(x, 0)) . Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = c1 ϑ(z)ϑ(z − β1) ϑ(z − γ1)ϑ(z − δ1) and Y(z) = c2 ϑ(z)ϑ(z − 2β1) ϑ(z − β1)2 , with αj + βj = γj + δj for j = 1, 2. K(0, 0) = 0, so WLOG α1 = α2 = 0. as x → 0, we have y(x) ∼ −x or y(x) ∼ − 1

x2 , so Y(z) has a

double pole at z = β1. Similarly: X(z) has a double pole at z = β2 = 2β1.

Counting lattice walks by winding angle Andrew Elvey Price

slide-144
SLIDE 144

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

For Kreweras paths: Q(x, y) = 1+xytQ(x, y)+ t x (Q(x, y) − Q(0, y))+ t y (Q(x, y) − Q(x, 0)) . Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = c1 ϑ(z)ϑ(z − β1) ϑ(z + β1)ϑ(z − 2β1) and Y(z) = c2 ϑ(z)ϑ(z − 2β1) ϑ(z − β1)2 , with αj + βj = γj + δj for j = 1, 2. K(0, 0) = 0, so WLOG α1 = α2 = 0. as x → 0, we have y(x) ∼ −x or y(x) ∼ − 1

x2 , so Y(z) has a

double pole at z = β1. Similarly: X(z) has a double pole at z = β2 = 2β1.

Counting lattice walks by winding angle Andrew Elvey Price

slide-145
SLIDE 145

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

For Kreweras paths: Q(x, y) = 1+xytQ(x, y)+ t x (Q(x, y) − Q(0, y))+ t y (Q(x, y) − Q(x, 0)) . Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = c1 ϑ(z)ϑ(z − β1) ϑ(z + β1)ϑ(z − 2β1) and Y(z) = c2 ϑ(z)ϑ(z − 2β1) ϑ(z − β1)2 , with αj + βj = γj + δj for j = 1, 2. K(0, 0) = 0, so WLOG α1 = α2 = 0. as x → 0, we have y(x) ∼ −x or y(x) ∼ − 1

x2 , so Y(z) has a

double pole at z = β1. Similarly: X(z) has a double pole at z = β2 = 2β1. So 3β1 = πτ.

Counting lattice walks by winding angle Andrew Elvey Price

slide-146
SLIDE 146

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

For Kreweras paths: Q(x, y) = 1+xytQ(x, y)+ t x (Q(x, y) − Q(0, y))+ t y (Q(x, y) − Q(x, 0)) . Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = c1 ϑ(z)ϑ

  • z − πτ

3

  • ϑ
  • z + πτ

3

  • ϑ
  • z − 2πτ

3

  • and Y(z) = c2

ϑ(z)ϑ

  • z − 2πτ

3

  • ϑ
  • z − πτ

3

2 , with αj + βj = γj + δj for j = 1, 2. K(0, 0) = 0, so WLOG α1 = α2 = 0. as x → 0, we have y(x) ∼ −x or y(x) ∼ − 1

x2 , so Y(z) has a

double pole at z = β1. Similarly: X(z) has a double pole at z = β2 = 2β1. So 3β1 = πτ.

Counting lattice walks by winding angle Andrew Elvey Price

slide-147
SLIDE 147

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

For Kreweras paths: Q(x, y) = 1+xytQ(x, y)+ t x (Q(x, y) − Q(0, y))+ t y (Q(x, y) − Q(x, 0)) . Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = c1 ϑ(z)ϑ

  • z − πτ

3

  • ϑ
  • z + πτ

3

  • ϑ
  • z − 2πτ

3

and Y(z) = c2 ϑ(z)ϑ

  • z + πτ

3

  • ϑ
  • z − πτ

3

z + 2πτ

3

, with αj + βj = γj + δj for j = 1, 2. K(0, 0) = 0, so WLOG α1 = α2 = 0. as x → 0, we have y(x) ∼ −x or y(x) ∼ − 1

x2 , so Y(z) has a

double pole at z = β1. Similarly: X(z) has a double pole at z = β2 = 2β1. So 3β1 = πτ.

Counting lattice walks by winding angle Andrew Elvey Price

slide-148
SLIDE 148

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

For Kreweras paths: Q(x, y) = 1+xytQ(x, y)+ t x (Q(x, y) − Q(0, y))+ t y (Q(x, y) − Q(x, 0)) . Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = c1 ϑ(z)ϑ

  • z − πτ

3

  • ϑ
  • z + πτ

3

  • ϑ
  • z − 2πτ

3

and Y(z) = c2 ϑ(z)ϑ

  • z + πτ

3

  • ϑ
  • z − πτ

3

z + 2πτ

3

, with αj + βj = γj + δj for j = 1, 2. K(0, 0) = 0, so WLOG α1 = α2 = 0. as x → 0, we have y(x) ∼ −x or y(x) ∼ − 1

x2 , so Y(z) has a

double pole at z = β1. Similarly: X(z) has a double pole at z = β2 = 2β1. So 3β1 = πτ.

Counting lattice walks by winding angle Andrew Elvey Price

slide-149
SLIDE 149

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

For Kreweras paths: Q(x, y) = 1+xytQ(x, y)+ t x (Q(x, y) − Q(0, y))+ t y (Q(x, y) − Q(x, 0)) . Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = e− 4πτi

9 ϑ(z)ϑ

  • z − πτ

3

  • ϑ
  • z + πτ

3

  • ϑ
  • z − 2πτ

3

  • and Y(z) = e− 4πτi

9 ϑ(z)ϑ

  • z + πτ

3

  • ϑ
  • z − πτ

3

z + 2πτ

3

, with αj + βj = γj + δj for j = 1, 2. K(0, 0) = 0, so WLOG α1 = α2 = 0. as x → 0, we have y(x) ∼ −x or y(x) ∼ − 1

x2 , so Y(z) has a

double pole at z = β1. Similarly: X(z) has a double pole at z = β2 = 2β1. So 3β1 = πτ.

Counting lattice walks by winding angle Andrew Elvey Price

slide-150
SLIDE 150

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ) and Y(z) = X(z + πτ), where t = 1 X(z)Y(z) + X(z)−1 + Y(z)−1 .

Counting lattice walks by winding angle Andrew Elvey Price

slide-151
SLIDE 151

BONUS SLIDE: PARAMETERIZATION OF K(x, y) = 0

Then K(x, y) = xy − tx2y2 − tx − ty = 0 is parameterised by X(z) = e− 4πτi

3 ϑ(z, 3τ)ϑ (z − πτ, 3τ)

ϑ (z + πτ, 3τ) ϑ (z − 2πτ, 3τ) and Y(z) = X(z + πτ), where t = e− πτi

3

ϑ′(0, 3τ) 4iϑ(πτ, 3τ) + 6ϑ′(πτ, 3τ).

Counting lattice walks by winding angle Andrew Elvey Price