Distribution of PCF cubic polynomials Charles Favre - - PowerPoint PPT Presentation

distribution of pcf cubic polynomials
SMART_READER_LITE
LIVE PREVIEW

Distribution of PCF cubic polynomials Charles Favre - - PowerPoint PPT Presentation

Distribution of PCF cubic polynomials Charles Favre charles.favre@polytechnique.edu June 28th, 2016 The parameter space of polynomials Poly d : degree d polynomials modulo conjugacy by affine transformations. { P = a d z d + + a 0


slide-1
SLIDE 1

Distribution of PCF cubic polynomials

Charles Favre charles.favre@polytechnique.edu June 28th, 2016

slide-2
SLIDE 2

The parameter space of polynomials

Polyd: degree d polynomials modulo conjugacy by affine transformations.

◮ {P = adzd + · · · + a0, ad = 0}/{P ∼ φ ◦ P ◦ φ−1, φ(z) =

az + b}

◮ C∗ × Cd/ Aff(2, C).

Complex affine variety of dimension d − 1: finite quotient singularities.

slide-3
SLIDE 3

Exploring the geometry of Poly2

Interested in the geometry of the locus of polynomials with special dynamics in Polyd.

slide-4
SLIDE 4

Exploring the geometry of Poly2

Interested in the geometry of the locus of polynomials with special dynamics in Polyd. Work with a suitable ramified cover of Polyd by Cd−1.

slide-5
SLIDE 5

Exploring the geometry of Poly2

Interested in the geometry of the locus of polynomials with special dynamics in Polyd. Work with a suitable ramified cover of Polyd by Cd−1.

◮ d = 2: parameterization Pc(z) = z2 + c, c ∈ C. Critical

point: 0.

◮ PCF maps: 0 is pre-periodic.

slide-6
SLIDE 6

Exploring the geometry of Poly2

Interested in the geometry of the locus of polynomials with special dynamics in Polyd. Work with a suitable ramified cover of Polyd by Cd−1.

◮ d = 2: parameterization Pc(z) = z2 + c, c ∈ C. Critical

point: 0.

◮ PCF maps: 0 is pre-periodic.

PCF maps are defined over Q: c = 0, c2 + c = 0, (c2 + c)2 + c = 0, . . .

slide-7
SLIDE 7

Distribution of hyperbolic quadratic PCF polynomials

slide-8
SLIDE 8

Distribution of hyperbolic quadratic PCF polynomials

slide-9
SLIDE 9

Distribution of quadratic PCF polynomials

PCF(n) = {Pc, the orbit of 0 has cardinality ≤ n}.

Theorem (Levin, Baker-H’sia, F .-Rivera-Letelier, ...)

The probability measures µn equidistributed on PCF(n) converge weakly towards the harmonic measure of the Mandelbrot set.

slide-10
SLIDE 10

Distribution of quadratic PCF polynomials

PCF(n) = {Pc, the orbit of 0 has cardinality ≤ n}.

Theorem (Levin, Baker-H’sia, F .-Rivera-Letelier, ...)

The probability measures µn equidistributed on PCF(n) converge weakly towards the harmonic measure of the Mandelbrot set.

◮ Levin: potential theoretic arguments ◮ Baker-H’sia: exploit adelic arguments, and work over all

completions of Q both Archimedean and non-Archimedean: C, Cp for p prime.

slide-11
SLIDE 11

The cubic case Poly3

◮ Parameterization: Pc,a(z) = 1 3z3 − c 2z2 + a3, a, c ∈ C. ◮ Crit(Pc,a) = {0, c} ◮ Pc,a(0) = a3, Pc,a(c) = a3 − c3 6 .

PCF(n, m) = {orbit of 0 has cardinality ≤ n} & {orbit of c has cardinality ≤ m}

slide-12
SLIDE 12

Finiteness of PCF maps

Pc,a(z) = 1

3z3 − c 2z2 + a3

Proposition (Branner-Hubbard)

The set PCF(n, m) is finite and defined over Q.

slide-13
SLIDE 13

Finiteness of PCF maps

Pc,a(z) = 1

3z3 − c 2z2 + a3

Proposition (Branner-Hubbard)

The set PCF(n, m) is finite and defined over Q.

◮ PCF(n, m) is bounded in C2 by Koebe distortion estimates;

slide-14
SLIDE 14

Finiteness of PCF maps

Pc,a(z) = 1

3z3 − c 2z2 + a3

Proposition (Branner-Hubbard)

The set PCF(n, m) is finite and defined over Q.

◮ in Cp with p ≥ 5:

slide-15
SLIDE 15

Finiteness of PCF maps

Pc,a(z) = 1

3z3 − c 2z2 + a3

Proposition (Branner-Hubbard)

The set PCF(n, m) is finite and defined over Q.

◮ in Cp with p ≥ 5:

◮ if |a| > max{1, |c|}, then |Pc,a(0)| = |a|3, |P2

c,a(0)| = |a|9

and, |Pn

c,a(0)| = |a|3n → ∞;

slide-16
SLIDE 16

Finiteness of PCF maps

Pc,a(z) = 1

3z3 − c 2z2 + a3

Proposition (Branner-Hubbard)

The set PCF(n, m) is finite and defined over Q.

◮ in Cp with p ≥ 5:

◮ if |a| > max{1, |c|}, then |Pc,a(0)| = |a|3, |P2

c,a(0)| = |a|9

and, |Pn

c,a(0)| = |a|3n → ∞;

◮ PCF(n, m) ⊂ {|c|, |a| ≤ 1}.

slide-17
SLIDE 17

Finiteness of PCF maps

Pc,a(z) = 1

3z3 − c 2z2 + a3

Proposition (Branner-Hubbard)

The set PCF(n, m) is finite and defined over Q.

Proposition (Ingram)

For any finite extension K/Q the set PCF ∩K 2 is finite.

slide-18
SLIDE 18

Distribution of PCF maps

Pc,a(z) = 1

3z3 − c 2z2 + a3

Theorem (F.-Gauthier)

The probability measures µn,m equidistributed on PCF(n, m) converge weakly towards the equilibrium measure µ of the connectedness locus C as n, m → ∞.

slide-19
SLIDE 19

Distribution of PCF maps

Pc,a(z) = 1

3z3 − c 2z2 + a3

Theorem (F.-Gauthier)

The probability measures µn,m equidistributed on PCF(n, m) converge weakly towards the equilibrium measure µ of the connectedness locus C as n, m → ∞.

◮ C = {(c, a) , Julia set of Pc,a is connected} =

{(c, a) , 0 and c have bounded orbit};

slide-20
SLIDE 20

Distribution of PCF maps

Pc,a(z) = 1

3z3 − c 2z2 + a3

Theorem (F.-Gauthier)

The probability measures µn,m equidistributed on PCF(n, m) converge weakly towards the equilibrium measure µ of the connectedness locus C as n, m → ∞.

◮ C = {(c, a) , Julia set of Pc,a is connected} =

{(c, a) , 0 and c have bounded orbit};

◮ Green function GC: C0 psh ≥ 0 function, C = {GC = 0},

G(c, a) = log max{1, |c|, |a|} + O(1);

slide-21
SLIDE 21

Distribution of PCF maps

Pc,a(z) = 1

3z3 − c 2z2 + a3

Theorem (F.-Gauthier)

The probability measures µn,m equidistributed on PCF(n, m) converge weakly towards the equilibrium measure µ of the connectedness locus C as n, m → ∞.

◮ C = {(c, a) , Julia set of Pc,a is connected} =

{(c, a) , 0 and c have bounded orbit};

◮ Green function GC: C0 psh ≥ 0 function, C = {GC = 0},

G(c, a) = log max{1, |c|, |a|} + O(1); µ = Monge-Ampère(GC) = (ddc)2GC.

slide-22
SLIDE 22

The adelic approach

Interpretation of the Green function in the parameter space: GC = max{Gc,a(c), Gc,a(0)}

◮ Dynamical Green function:

Gc,a = lim 1 3n log max{1, |Pn

c,a|}

slide-23
SLIDE 23

The adelic approach

Interpretation of the Green function in the parameter space: GC = max{Gc,a(c), Gc,a(0)}

◮ Dynamical Green function:

Gc,a = lim 1 3n log max{1, |Pn

c,a|} ◮ Same construction for any norm | · |p on Q: GC,p

slide-24
SLIDE 24

The adelic approach

Interpretation of the Green function in the parameter space: GC = max{Gc,a(c), Gc,a(0)}

◮ Dynamical Green function:

Gc,a = lim 1 3n log max{1, |Pn

c,a|} ◮ Same construction for any norm | · |p on Q: GC,p

Key observation: Pc,a ∈ PCF iff Height(c, a) := 1 deg(c, a)

  • p
  • c′,a′

GC,p(c′, a′) = 0 − → Apply Yuan’s theorem!

slide-25
SLIDE 25

Special curves

Problem (Baker-DeMarco)

Describe all irreducible algebraic curves C in Poly3 such that PCF ∩C is infinite.

◮ motivated by the André-Oort conjecture in arithmetic

geometry

slide-26
SLIDE 26

Special curves

Problem (Baker-DeMarco)

Describe all irreducible algebraic curves C in Poly3 such that PCF ∩C is infinite.

◮ motivated by the André-Oort conjecture in arithmetic

geometry

Theorem (Baker-DeMarco, Ghioca-Ye, F .-Gauthier)

Suppose C is an irreducible algebraic curve in Poly3 such that PCF ∩C is infinite. Then there exists a persistent critical relation.

slide-27
SLIDE 27

The curves Pern(λ): DeMarco’s conjecture

Pern(λ) = {Pc,a admitting a periodic point

  • f period n and multiplier λ}
slide-28
SLIDE 28

The curves Pern(λ): DeMarco’s conjecture

Pern(λ) = {Pc,a admitting a periodic point

  • f period n and multiplier λ}

◮ Geometry of Pern(0): Milnor, DeMarco-Schiff, irreducibility

by Arfeux-Kiwi (n prime);

slide-29
SLIDE 29

The curves Pern(λ): DeMarco’s conjecture

Pern(λ) = {Pc,a admitting a periodic point

  • f period n and multiplier λ}

◮ Geometry of Pern(0): Milnor, DeMarco-Schiff, irreducibility

by Arfeux-Kiwi (n prime);

◮ Distribution of Pern(λ) when n → ∞ described by

Bassaneli-Berteloot.

slide-30
SLIDE 30

The curves Pern(λ): DeMarco’s conjecture

Pern(λ) = {Pc,a admitting a periodic point

  • f period n and multiplier λ}

◮ Geometry of Pern(0): Milnor, DeMarco-Schiff, irreducibility

by Arfeux-Kiwi;

◮ Distribution of Pern(λ) when n → ∞ described by

Bassaneli-Berteloot.

Theorem (F.-Gauthier)

The set PCF ∩ Pern(λ) is infinite iff λ = 0. n = 1 by Baker-DeMarco

slide-31
SLIDE 31

Scheme of proof

C irreducible component of Pern(λ) containing infinitely many PCF: λ ∈ ¯ Q.

  • 1. One of the critical point is persistently preperiodic on C.
  • 2. C contains a unicritical PCF polynomial =

⇒ |λ|3 < 1.

  • 3. There exists a quadratic PCF polynomial having λ as a

multiplier = ⇒ |λ|3 = 1.

slide-32
SLIDE 32

Step 2

Pc,a(z) = 1

3z3 − c 2z2 + a3

C contains a unicritical PCF polynomial, and |λ|3 < 1. a) Existence of the PCF unicritical polynomial by Bezout and Step 1.

slide-33
SLIDE 33

Step 2

Pc,a(z) = 1

3z3 − c 2z2 + a3

C contains a unicritical PCF polynomial, and |λ|3 < 1. a) Existence of the PCF unicritical polynomial by Bezout and Step 1. b) Suppose P(z) = z3 + b is PCF

◮ |b|3 ≤ 1 ◮ periodic orbits are included in |z|3 ≤ 1 ◮ |P′(z)|3 = |3z2|3 < 1 on the unit ball hence |λ|3 < 1

slide-34
SLIDE 34

Step 1

One of the critical point is persistently preperiodic on C.

slide-35
SLIDE 35

Step 1

One of the critical point is persistently preperiodic on C.

  • 1. Apply the previous theorem: there exists a persistent

critical relation.

slide-36
SLIDE 36

Step 1

One of the critical point is persistently preperiodic on C.

  • 1. Apply the previous theorem: there exists a persistent

critical relation.

  • 2. Suppose Pm′

c,a(0) = Pm c,a(c) for all c, a ∈ C.

◮ a branch at infinity c of C induces a cubic polynomial

P = Pc(t),a(t) ∈ C((t))[z]

◮ both points 0 and c tend to ∞ on c ◮ Kiwi & Bezivin =

⇒ all multipliers of P are repelling

◮ contradiction with the existence of a periodic point with

constant multiplier λ

slide-37
SLIDE 37

Construction of the quadratic renormalization

Use more from Kiwi on the dynamics of P over the non-Archimedean field C((t))!

slide-38
SLIDE 38

Construction of the quadratic renormalization

Use more from Kiwi on the dynamics of P over the non-Archimedean field C((t))!

◮ Step 1 implies 0 is pre-periodic whereas c tends to ∞. ◮ If 0 is in the Julia set, then all cycles are repelling (Kiwi).

slide-39
SLIDE 39

Construction of the quadratic renormalization

Use more from Kiwi on the dynamics of P over the non-Archimedean field C((t))!

◮ Step 1 implies 0 is pre-periodic whereas c tends to ∞. ◮ If 0 is in the Julia set, then all cycles are repelling (Kiwi). ◮ Non-wandering theorem of Kiwi-Trucco: 0 belongs to a

periodic (closed) ball: Pn(B) = B.

slide-40
SLIDE 40

Construction of the quadratic renormalization

Use more from Kiwi on the dynamics of P over the non-Archimedean field C((t))!

◮ Step 1 implies 0 is pre-periodic whereas c tends to ∞. ◮ If 0 is in the Julia set, then all cycles are repelling (Kiwi). ◮ Non-wandering theorem of Kiwi-Trucco: 0 belongs to a

periodic (closed) ball: Pm(B) = B.

Pm(z) = a0(t) + a1(t)z + . . . + ak(t)zk Pm{|z| ≤ 1} = {|z| ≤ 1} implies ai(t) ∈ C[[t]] The renormalization is a0(0) + a1(0)z + . . . + ak(0)zk.

slide-41
SLIDE 41

Dynamics of the quadratic renormalization

◮ The renormalization is a quadratic PCF polynomial Q. ◮ One multiplier of Q is λ: non-repelling orbits pass through

B (Kiwi)

◮ All multipliers of Q have | · |3 = 1.

slide-42
SLIDE 42

Happy Birthday Sebastian!