An old new class of meromorphic functions. Norbert Steinmetz. June 13, 2015

Similar documents
Local and global finite branching of solutions of ODEs in the complex plane

On the singularities of non-linear ODEs

Boutroux s Method vs. Re-scaling Lower estimates for the orders of growth of the second and fourth Painlevé transcendents

Solutions to Complex Analysis Prelims Ben Strasser

On rational approximation of algebraic functions. Julius Borcea. Rikard Bøgvad & Boris Shapiro

Math Final Exam.

Global Properties Of The Painlevé Transcendents

A UNIFIED APPROACH TO THE PAINLEVÉ TRANSCENDENTS

MATH FINAL SOLUTION

MATH COMPLEX ANALYSIS. Contents

Non-real zeroes of real entire functions and their derivatives

7 Asymptotics for Meromorphic Functions

Part IB. Further Analysis. Year

Qualifying Exam Complex Analysis (Math 530) January 2019

8 8 THE RIEMANN MAPPING THEOREM. 8.1 Simply Connected Surfaces

Part II. Riemann Surfaces. Year

Explicit Examples of Strebel Differentials

Integrability, Nonintegrability and the Poly-Painlevé Test

FINAL EXAM MATH 220A, UCSD, AUTUMN 14. You have three hours.

. Then g is holomorphic and bounded in U. So z 0 is a removable singularity of g. Since f(z) = w 0 + 1

Theorem [Mean Value Theorem for Harmonic Functions] Let u be harmonic on D(z 0, R). Then for any r (0, R), u(z 0 ) = 1 z z 0 r

1. The COMPLEX PLANE AND ELEMENTARY FUNCTIONS: Complex numbers; stereographic projection; simple and multiple connectivity, elementary functions.

ON FACTORIZATIONS OF ENTIRE FUNCTIONS OF BOUNDED TYPE

Solutions for Problem Set #5 due October 17, 2003 Dustin Cartwright and Dylan Thurston

COMPLEX ANALYSIS Spring 2014

The Canonical Sheaf. Stefano Filipazzi. September 14, 2015

. OKAMOTO S SPACE FOR THE FIRST PAINLEVÉ EQUATION IN BOUTROUX COORDINATES J.J. DUISTERMAAT AND N. JOSHI

Complex Analysis Important Concepts

Math 185 Fall 2015, Sample Final Exam Solutions

Bernstein s analytic continuation of complex powers

INTRODUCTION TO PADÉ APPROXIMANTS. E. B. Saff Center for Constructive Approximation

III. Consequences of Cauchy s Theorem

Problems for MATH-6300 Complex Analysis

Considering our result for the sum and product of analytic functions, this means that for (a 0, a 1,..., a N ) C N+1, the polynomial.

5.3 The Upper Half Plane

Lecture 3. Frits Beukers. Arithmetic of values of E- and G-function. Lecture 3 E- and G-functions 1 / 20

SOLUTIONS OF NONHOMOGENEOUS LINEAR DIFFERENTIAL EQUATIONS WITH EXCEPTIONALLY FEW ZEROS

Let X be a topological space. We want it to look locally like C. So we make the following definition.

3. 4. Uniformly normal families and generalisations

COMPLEX ANALYSIS Spring 2014

MEROMORPHIC FUNCTIONS AND ALSO THEIR FIRST TWO DERIVATIVES HAVE THE SAME ZEROS

Ramanujan s first letter to Hardy: 5 + = 1 + e 2π 1 + e 4π 1 +

BANK-LAINE FUNCTIONS WITH SPARSE ZEROS

Homework 7 solutions M328K by Mark Lindberg/Marie-Amelie Lawn

CONSEQUENCES OF POWER SERIES REPRESENTATION

Limits in differential fields of holomorphic germs

ELLIPTIC FUNCTIONS AND THETA FUNCTIONS

MEROMORPHIC FUNCTIONS WHOSE JULIA SETS CONTAIN A FREE JORDAN ARC

Part IB. Complex Analysis. Year

Growth of meromorphic solutions of delay differential equations

Complex Analysis Qual Sheet

Final Exam - MATH 630: Solutions

MATH 566 LECTURE NOTES 6: NORMAL FAMILIES AND THE THEOREMS OF PICARD

MA30056: Complex Analysis. Exercise Sheet 7: Applications and Sequences of Complex Functions

Integral Bases. 1. Resultants

Riemann Surfaces and Algebraic Curves

Evaluation of integrals

Complex Analysis Topic: Singularities

Math 220A - Fall Final Exam Solutions

From alternating sign matrices to Painlevé VI

MATH 452. SAMPLE 3 SOLUTIONS May 3, (10 pts) Let f(x + iy) = u(x, y) + iv(x, y) be an analytic function. Show that u(x, y) is harmonic.

1 Structures 2. 2 Framework of Riemann surfaces Basic configuration Holomorphic functions... 3

1. Partial Fraction Expansion All the polynomials in this note are assumed to be complex polynomials.

WEIERSTRASS THEOREMS AND RINGS OF HOLOMORPHIC FUNCTIONS

On Bank-Laine functions

Randomized Algorithms III Min Cut

Complex Analysis Qualifying Exam Solutions

Partition function of one matrix-models in the multi-cut case and Painlevé equations. Tamara Grava and Christian Klein

18.04 Practice problems exam 2, Spring 2018 Solutions

The Geometry of Cubic Maps

Chapter 4: Open mapping theorem, removable singularities

Conformal maps. Lent 2019 COMPLEX METHODS G. Taylor. A star means optional and not necessarily harder.

The Dynamics of Two and Three Circle Inversion Daniel M. Look Indiana University of Pennsylvania

Solutions to practice problems for the final

MTH 3102 Complex Variables Final Exam May 1, :30pm-5:30pm, Skurla Hall, Room 106

MTH 3102 Complex Variables Final Exam May 1, :30pm-5:30pm, Skurla Hall, Room 106

Complex Analysis. Travis Dirle. December 4, 2016

Department of Mathematics, University of California, Berkeley. GRADUATE PRELIMINARY EXAMINATION, Part A Spring Semester 2015

Conformal Mappings. Chapter Schwarz Lemma

Congruent Numbers, Elliptic Curves, and Elliptic Functions

Algebraic function fields

INTRODUCTION TO COMPLEX ANALYSIS W W L CHEN

Elliptic Curves and Elliptic Functions

PROPER HOLOMORPHIC IMMERSIONS IN HOMOTOPY CLASSES OF MAPS FROM FINITELY CONNECTED PLANAR DOMAINS INTO C C

f (n) (z 0 ) Theorem [Morera s Theorem] Suppose f is continuous on a domain U, and satisfies that for any closed curve γ in U, γ

Complex Analysis Math 185A, Winter 2010 Final: Solutions

Here are brief notes about topics covered in class on complex numbers, focusing on what is not covered in the textbook.

20 The modular equation

ALGEBRAIC GEOMETRY AND RIEMANN SURFACES

Siegel Discs in Complex Dynamics

Properties of Entire Functions

On values of Modular Forms at Algebraic Points

13 Maximum Modulus Principle

Poles and α-points of Meromorphic Solutions of the First Painlevé Hierarchy

GROWTH OF SOLUTIONS TO HIGHER ORDER LINEAR HOMOGENEOUS DIFFERENTIAL EQUATIONS IN ANGULAR DOMAINS

Chapter 4. The Lamé equation. 4.1 The Lamé equation

Math Homework 2

Complex Analysis, Stein and Shakarchi Meromorphic Functions and the Logarithm

Part IB Complex Analysis

Taylor Expansions with Finite Radius of Convergence Work of Robert Jentzsch, 1917

Transcription:

Norbert Steinmetz June 13, 2015

NEW & OLD NEW: Hamiltonian system p = H q = q 2 zp α q = H p = p 2 + zq + β with Hamiltonian H(z, p, q) = 1 3 (p3 + q 3 ) + zpq + αq + βp OLD: Almost Painlevé IV 2ww = w 2 w 4 4zw 3 (2ωα+2 ωβ+3z 2 )w 2 (ωα ωβ+1) 2. OLD & NEW: w = ωp + ωq z (ω 3 = 1).

Three reasons to consider p = q 2 zp α q = p 2 + zq + β The solutions are meromorphic in the plane, while any known and essentially different system with Hamiltonian H(z, p, q) = 1 m pm + 1 n qn + nµ+mν<mn a µν (z)p µ q ν (m, n 3) has movable algebraic poles or more complicated singularities (Kecker). The solutions have extremely nice properties New insight in Painlevé IV There exists a method

Re-scaling (also re-normalisation) Set p h (z) = h 1 p(h + h 1 z), q h (z) = q(h + h 1 z). p = q 2 zp α q = p 2 + zq + β p = lim h p h q = lim h q h (formal) p = q 2 p q = p 2 + q 1 3 (p3 + q 3 ) + pq = c?! properties of (p, q) (???) properties of (p, q) W 2 The problem (???) disappears if one can show that the families (p h ) h >1 and (q h ) h >1 are normal in the plane.

Re-scaling constant solutions to ( ) p = q 2 p q = p 2 + q give rise to asymptotic expansions on Stokes sectors for the solutions to p = q 2 zp α q = p 2 + zq + β non-constant (periodic or elliptic) solutions to ( ) yield information on the distribution of poles of (p, q).

Poles Trivial and non-trivial properties p and q have simple poles λ Λ with residues ω and ω, resp. (ω 3 = 1). H has simple poles with residue 1: H(z) = (z λ) 1 + [2h + 1 3 λ3 + 1 (α + β)λ] + 2 (h = h(λ) is unknown and may be prescribed once!) The discs δ (λ) = {z : z λ < δ λ 1 } (λ 0, δ > 0 sufficiently small) are mutually disjoint. p + q = O( z ) and p + q = O( z 2 ) outside the exceptional set Λ δ = λ 0 δ(λ).

Poles H(z) = O( z 3 ) outside Λ δ. The order of growth is at most 4: n(r, Λ) = 1 H(z) dz = O(r 4 ) 2πi Γ r (Γ r = approximately the circle z = r). The residues are equi-distributed: n(r, Λ ω ) = 1n(r, Λ) + O(r 2 ) (ω 3 = 1) follows from 3 ωn(r, Λ ω ) = 1 p(z) dz = O(r 2 ). ω 3 =1 2πi Γ r n(r, Λ) = O(r 2 ) if at least one residue ω is missing.

Generalised spherical derivative Lemma. p z 2 + p + q 2 z 2 + q = O(1). 2 Proof. Res z=λ p = ω: f = ωp + p 2 ωzp is regular on δ (λ) and satisfies f (z) = O( z 2 ) on δ (λ): Maximum principle. More examples: (R) w = z n w 2 : (P I ) w = z + 6w 2 : (P II ) w = α + zw + 2w 3 : (P IV ) 2ww = w 2 + 3w 4 + 8zw 3 + 4(z 2 α)w 2 + 2β : w z n + w = O(1) 2 w z + w = 2 O( z w z + w = O(1) 2 w z 2 + w 2 = O(1) 1 4 )

Generalised spherical derivative Lemma. Suppose f is meromorphic in the z-plane and satisfies f (z) z 2a + f (z) 2 = O( z b a ) (z ) for some a R, b > 1. Then the re-scaled family (f h ) h >1 f h (z) = h a f (h + h b z) is normal in the z-plane. (non-constant f = lim f h k h k function). Proof. Part of the afternoon exam. Yosida

Re-scaling The families (p h ) h >1 with p h (z) = h 1 p(h + h 1 z) and (q h ) h >1 are normal on the plane. The limit functions p = lim p h k and q = lim q h k satisfy h k h k p = q 2 p, q = p 2 + q with constant Hamiltonian H(p, q) = 1 3 (p3 + q 3 ) + pq c = lim h k h 3 k H(h k) (inf h k dist(h k, Λ) > 0) lim λ k 2λ 3 k h(λ k) + 1 ( lim h 3 k h k λ k = 0) k the constants c form the cluster set C(p, q); it is closed, bounded and connected.

The algebraic curve H(x, y) = 1 3 (x 3 + y 3 ) + xy c = 0 either is reducible (c = 1 3 ), H(x, y) = (y + x 1)(y + ωx ω)(y + ωx ω) or else has genus 0 (c = 0), and is parametrised by x = 3ωt2 3 ωt and y = t 3 + 1 t 3 + 1, or else has genus 1 (c 0, 1 ) and is parametrised by a 3 pair of elliptic functions: (x ) 3 + 9x(x ) 2 + 24x 2 x + (x 6 + (16 6c)x 3 + 9c 2 ) = 0 (y ) 3 9y(y ) 2 + 24y 2 y (y 6 (16 6c)y 3 + 9c 2 ) = 0

The hamiltonian system p = q 2 p q = p 2 + q has non-constant solutions q = 1 p, p + ω p + ω = 3z ei (ω = e 2πi/3 ), and similar expressions if q = ω ωp in the reducible case p = 3ωe2z 3 ωez and q = in case of genus 0, e 3z + 1 e 3z + 1 p(z) = ωζ(z z ω (c)), q(z) = ωζ(z z ω (c)) in ω 3 =1 ω 3 =1 case of genus 1 with lattice depending on c.

Distribution of poles: reducible case p = q 2 p p = q 2 zp α q = p 2 + q q = p 2 + zq + β 1 3 (p3 + q 3 ) + pq = 1 3 0 z-plane ( z < 8) λ z-plane ( z λ < 8 λ 1 ) ( λ large) The discs have radii o( λ 1 ).

Distribution of poles: genus zero p = q 2 p p = q 2 zp α q = p 2 + q q = p 2 + zq + β 1 3 (p3 + q 3 ) + pq = 0 0 z-plane ( z < 7) z-plane ( z λ < 7 λ 1 ) The residues alternate λ

Distribution of poles: genus one p = q 2 p p = q 2 zp α q = p 2 + q q = p 2 + zq + β 1 3 (p3 + q 3 ) + pq = c 0, 1 3 z-plane z-plane

The cluster set and sub-normal solutions Transcendental solutions satisfy C 1 r 2 n(r, Λ) C 2 r 4. Solutions satisfying n(r, Λ) r 2 are called sub-normal. C(p, q) {0, 1/3} n(r k, Λ) r 4 k Guess: n(r, w) r 4. Hinkkanen and Laine: C(p, q) {0, 1/3} n(r, Λ) r 2

Strings of poles Any sequence (λ k ) satisfying some parabolic-like recursion λ k+1 = λ k + (c + o(1))λ s k (c 0, s > 1 rational) is called a (c, s)-string. λ k ((1 + s)c)k) 1/(1+s) (consider µ k = λ 1+s k ) n(r, {λ k }) r 1+s (1 + s) c arg λ k 1 2π arg c modulo 1+s 1+s Sub-normal solutions have finitely many (± 2π 3, 1)-strings of poles if C(p, q) = {1/3} (± 2πi, 1)-strings of poles if C(p, q) = {0} 3

Much ado about nothing: Sub-normal solutions with C(p, q) = {0} do not exist. Proof. There are finitely many strings of poles, each asymptotic to some ray arg z = (2ν + 1)π/4 p, q and H have asymptotic expansions on each sector (2ν 1)π/4 < arg z < (2ν + 1)π/4, in particular H(z) = αβ z α3 + β 3 + O( z 5 ) 3z 3 g(z) = exp( H(z) dz) is entire of order 4, with zeros at the poles of (p, q) Phragmén-Lindelöf argument on arg z 2ν+1 4 π < δ g is a polynomial (of degree αβ), hence p and q are rational functions.

C(p, q) = {1/3} Type-1 sub-normal solutions Transcendental solutions(p, q) having only poles with fixed pair of residues (ω, ω) are sub-normal with cluster set C(p, q) = {1/3}. In case of ω = 1: p + q z = 0 α β + 1 = 0 p = α z 2 + zp p 2, q = β + z 2 zq + q 2 One string of poles in four [two adjacent] Stokes directions p ϱz p ϱz p ϱz p ϱz p ϱz p ϱz Generic and degenerate distribution of poles and asymptotics; Res p = 1; ϱ = 1 2 ( 1 + i 3).

C(p, q) = {1/3} Type-2 sub-normal solutions Transcendental solutions having only poles with pairs of residues (ω 1, ω 1 ) and (ω 2, ω 2 ) are sub-normal with cluster set C(p, q) = {1/3}. In case of ω 1,2 = 1 2 ( 1 ± i 3): α = β + 2 1 p 2 + q 2 + z 2 pq + z(p + q) + 3β + 3 0 p 2 + A 2 (z, p)p + A 4 (z, p) = 0 (Fuchsian) Two strings of poles in 4 [2 adjacent] directions p ωz p ωz p ωz p ωz p ωz p ωz Res p = ω 1, Res p = ω 2 ; in the narrow domains between two strings, p(z) = z + o( z ) holds.

C(p, q) = {1/3} Type-3 sub-normal solutions

C(p, q) = {1/3} Type-n sub-normal solutions Theorem. Every sub-normal solutions has some type. Type-n solutions have 4 n [in exceptional cases only 2 n] strings of poles. They exist if n 0 modulo 3 (for known parameters), but never if n 0 modulo 3. p z p ωz p ωz p z p ωz p ωz p ωz p ωz n = 3k + 1 (ω = 1 2 ( 1 + i 3)) n = 3k + 2 4[2] ((k + 1) + k + k) strings 4[2] ((k + 1) + (k + 1) + k) strings

Asymptotics and distribution of residues p(z) ωz + p(z) = z + o( z ) p(z) = ωz + o( z ) p(z) = ωz + o( z ) p z + p(z) ωz + (1 st quadrant) p(z) = z + o( z ) p(z) = ωz + o( z ) p(z) = ωz + o( z ) p(z) = z + o( z ) p ωz + (4 th quadrant) Res = 1, Res = ω = 1 2 ( 1 + i 3), Res = ω

Bäcklund transformations (Kecker) The Bäcklund transformations (ω 3 = 1) M ω : (p, q) ( ωp, ωq) and ( ) ωα ωβ + 1 p(z) ω p B ω : ωp(z) + ωq(z) z q ωα ωβ + 1 q(z) + ω ωp(z) + ωq(z) z transform p = q 2 zp α, q = p 2 + zq + β into the same Hamiltonian system with resp. parameters ( ωα, ωβ) (ωβ ω, ωα + ω).

Main Result Theorem. To any Type-n (n > 1) sub-normal solution (p, q) there exists some Bäcklund transformation B 1 M ω which maps (p, q) onto some Type-(n 1) sub-normal solution if n 2 modulo 3; Type-(n 2) sub-normal solution if n 1 modulo 3. Corollary. Every sub-normal solution has its source in a Type-1 subnormal solution; p and q satisfy first order differential equations P(z, p, p ) = Q(z, q, q ) = 0 of Fuchsian type and are algebraically dependent, K(z, p, q) = 0 (genus zero).

Proof of the Main Result p and q have asymptotic expansions p(z) τ ν z +, q(z) τ ν z + (τ 3 ν = 1) on (ν 1)π/2 < arg z < νπ/2; shortly written τ 2 τ 1 τ 3 τ 4. Reason: p = q 2 p, q = p 2 + q, 1 3 (p3 + q 3 ) + pq = 1 3 has the constant solutions (p, q) = (τ ν, τ ν ) under B 1 M ω, old strings of poles may disappear or may change their residue, and new strings may be created n(r, Λ) = n(r, M ω Λ). 81 configurations τ 2 τ 1 τ 3 τ 4 may be reduced to 5.

Proof of the Main Result τ 1 τ 1 = τ 1 τ 1 is ruled out (Phragmén-Lindelöf). τ 3 τ 3 τ 3 τ 3 Set p ω = M ω p, Λ ω = M ω Λ and H ω = M ω H. n(r, Λ) n(r, B 1 Λ ω ) = n(r, Λ ω ) n(r, B 1 Λ ω ) = 1 (H ω (z) B 1 H ω (z)) dz 2πi Γ r (H ω B 1 H ω = p ω B 1 p ω ) = 1 (p ω (z) B 1 p ω (z)) dz 2πi Γ r 3r 2 (ω, {τ ν }) 4π (ω, {τ ν }) assumes integer values 1, = 0, and 1 if ω varies. Choose ω with (ω, {τ ν }) 1 to decrease the number of strings.

Symbolic illustration 5 4 2 1 5 4 2 1 ( display poles with different residues, not necessarily the same in different pictures)

Application to Painlevé IV 2ww = w 2 w 4 4zw 3 (2α + 2β + 3z 2 )w 2 (α β + 1) 2 is equivalent to Painlevé IV. Known: α β + 1 0 m(r, 1/w) = O(log r) α β + 1 = 0 m(r, 1/w) 1 T (r, w) + O(log r), 2 except when α = 1( 1 ± i 3), β = α + 1, and 2 w = i 3 (3zw + w 2 ) : N(r, 1/w) = 0. Theorem. δ(0, w) = 0, except when w = p + q z and (p, q) is a Type-(3k + 1) sub-normal solution with parameters α k = 1 2 ( 1 ± (2k + 1) 3i), β k = α k + 1, and δ(0, w) = 1 2k + 1. For k = 0, zero is a Picard value.

Application to Painlevé IV n = 3k + 1: w = p + q z has 4(k + 1) and 4k strings of simple poles and, respectively, and 4k strings of 2k double zeros : δ(0, w) = 1 k + 1 + k = 1 2k + 1. n = 3k + 2: α β + 1 = 0 is impossible.

Thank you for listening