Kakutani equivalence of unipotent flows Adam Kanigowski Montreal, - - PowerPoint PPT Presentation

kakutani equivalence of unipotent flows
SMART_READER_LITE
LIVE PREVIEW

Kakutani equivalence of unipotent flows Adam Kanigowski Montreal, - - PowerPoint PPT Presentation

Kakutani equivalence of unipotent flows Adam Kanigowski Montreal, 07.27.2018 (joint w. K. Vinhage and D. Wei) 1 / 11 General setting ( X , B , ) probability standard Borel space; ( T t ) : ( X , B , ) ( X , B , )


slide-1
SLIDE 1

Kakutani equivalence of unipotent flows

Adam Kanigowski Montreal, 07.27.2018 (joint w. K. Vinhage and D. Wei)

1 / 11

slide-2
SLIDE 2

General setting (X, B, µ) – probability standard Borel space; (Tt) : (X, B, µ) → (X, B, µ) – measure-preserving, ergodic flow. Isomorphism (Tt) : (X, B, µ) → (X, B, µ), (St) : (Y , C, ν) → (Y , C, ν) are isomorphic, if R ◦ Tt = St ◦ R for t ∈ R, where R : (X, B, µ) → (Y , C, ν) is invertible and ν(C) = µ(R−1C), C ∈ C. Classification up to isomorphism NOT possible in general (M. Foreman, D. Rudolph, B. Weiss, 2011)

2 / 11

slide-3
SLIDE 3

General setting (X, B, µ) – probability standard Borel space; (Tt) : (X, B, µ) → (X, B, µ) – measure-preserving, ergodic flow. Isomorphism (Tt) : (X, B, µ) → (X, B, µ), (St) : (Y , C, ν) → (Y , C, ν) are isomorphic, if R ◦ Tt = St ◦ R for t ∈ R, where R : (X, B, µ) → (Y , C, ν) is invertible and ν(C) = µ(R−1C), C ∈ C. Classification up to isomorphism NOT possible in general (M. Foreman, D. Rudolph, B. Weiss, 2011)

2 / 11

slide-4
SLIDE 4

General setting (X, B, µ) – probability standard Borel space; (Tt) : (X, B, µ) → (X, B, µ) – measure-preserving, ergodic flow. Isomorphism (Tt) : (X, B, µ) → (X, B, µ), (St) : (Y , C, ν) → (Y , C, ν) are isomorphic, if R ◦ Tt = St ◦ R for t ∈ R, where R : (X, B, µ) → (Y , C, ν) is invertible and ν(C) = µ(R−1C), C ∈ C. Classification up to isomorphism NOT possible in general (M. Foreman, D. Rudolph, B. Weiss, 2011)

2 / 11

slide-5
SLIDE 5

General setting (X, B, µ) – probability standard Borel space; (Tt) : (X, B, µ) → (X, B, µ) – measure-preserving, ergodic flow. Isomorphism (Tt) : (X, B, µ) → (X, B, µ), (St) : (Y , C, ν) → (Y , C, ν) are isomorphic, if R ◦ Tt = St ◦ R for t ∈ R, where R : (X, B, µ) → (Y , C, ν) is invertible and ν(C) = µ(R−1C), C ∈ C. Classification up to isomorphism NOT possible in general (M. Foreman, D. Rudolph, B. Weiss, 2011)

2 / 11

slide-6
SLIDE 6

Orbit and Kakutani equivalence

Orbit equivalence (Tt) and (St) are orbit equivalent, if there exists a invertible transformation that maps orbits of (Tt) to orbits of (St). (Dye’s theorem, 1959) Every two ergodic flows are orbit equivalent. Kakutani equivalence, 1943 (Tt) : (X, B, µ) → (X, B, µ) and (St) : (Y , C, ν) → (Y , C, ν) are Kakutani equivalent (denoted (Tt)

K

∼ (St)), if they have isomorphic sections, i.e. can be represented as special flows over the same transformation. Standard flows Let (Rα

t ) : T2 → T2, Rα t (x, y) = (x + t, y + tα). Then, for every

α, β / ∈ Q, (Rα

t ) K

∼ (Rβ

t ) (Katok, 1976; Ornstein, Rudolph, Weiss,

1982); (Tt) is standard, if (Tt)

K

∼ (Rα

t ) for some α /

∈ Q.

3 / 11

slide-7
SLIDE 7

Orbit and Kakutani equivalence

Orbit equivalence (Tt) and (St) are orbit equivalent, if there exists a invertible transformation that maps orbits of (Tt) to orbits of (St). (Dye’s theorem, 1959) Every two ergodic flows are orbit equivalent. Kakutani equivalence, 1943 (Tt) : (X, B, µ) → (X, B, µ) and (St) : (Y , C, ν) → (Y , C, ν) are Kakutani equivalent (denoted (Tt)

K

∼ (St)), if they have isomorphic sections, i.e. can be represented as special flows over the same transformation. Standard flows Let (Rα

t ) : T2 → T2, Rα t (x, y) = (x + t, y + tα). Then, for every

α, β / ∈ Q, (Rα

t ) K

∼ (Rβ

t ) (Katok, 1976; Ornstein, Rudolph, Weiss,

1982); (Tt) is standard, if (Tt)

K

∼ (Rα

t ) for some α /

∈ Q.

3 / 11

slide-8
SLIDE 8

Orbit and Kakutani equivalence

Orbit equivalence (Tt) and (St) are orbit equivalent, if there exists a invertible transformation that maps orbits of (Tt) to orbits of (St). (Dye’s theorem, 1959) Every two ergodic flows are orbit equivalent. Kakutani equivalence, 1943 (Tt) : (X, B, µ) → (X, B, µ) and (St) : (Y , C, ν) → (Y , C, ν) are Kakutani equivalent (denoted (Tt)

K

∼ (St)), if they have isomorphic sections, i.e. can be represented as special flows over the same transformation. Standard flows Let (Rα

t ) : T2 → T2, Rα t (x, y) = (x + t, y + tα). Then, for every

α, β / ∈ Q, (Rα

t ) K

∼ (Rβ

t ) (Katok, 1976; Ornstein, Rudolph, Weiss,

1982); (Tt) is standard, if (Tt)

K

∼ (Rα

t ) for some α /

∈ Q.

3 / 11

slide-9
SLIDE 9

Orbit and Kakutani equivalence

Orbit equivalence (Tt) and (St) are orbit equivalent, if there exists a invertible transformation that maps orbits of (Tt) to orbits of (St). (Dye’s theorem, 1959) Every two ergodic flows are orbit equivalent. Kakutani equivalence, 1943 (Tt) : (X, B, µ) → (X, B, µ) and (St) : (Y , C, ν) → (Y , C, ν) are Kakutani equivalent (denoted (Tt)

K

∼ (St)), if they have isomorphic sections, i.e. can be represented as special flows over the same transformation. Standard flows Let (Rα

t ) : T2 → T2, Rα t (x, y) = (x + t, y + tα). Then, for every

α, β / ∈ Q, (Rα

t ) K

∼ (Rβ

t ) (Katok, 1976; Ornstein, Rudolph, Weiss,

1982); (Tt) is standard, if (Tt)

K

∼ (Rα

t ) for some α /

∈ Q.

3 / 11

slide-10
SLIDE 10

Orbit and Kakutani equivalence

Orbit equivalence (Tt) and (St) are orbit equivalent, if there exists a invertible transformation that maps orbits of (Tt) to orbits of (St). (Dye’s theorem, 1959) Every two ergodic flows are orbit equivalent. Kakutani equivalence, 1943 (Tt) : (X, B, µ) → (X, B, µ) and (St) : (Y , C, ν) → (Y , C, ν) are Kakutani equivalent (denoted (Tt)

K

∼ (St)), if they have isomorphic sections, i.e. can be represented as special flows over the same transformation. Standard flows Let (Rα

t ) : T2 → T2, Rα t (x, y) = (x + t, y + tα). Then, for every

α, β / ∈ Q, (Rα

t ) K

∼ (Rβ

t ) (Katok, 1976; Ornstein, Rudolph, Weiss,

1982); (Tt) is standard, if (Tt)

K

∼ (Rα

t ) for some α /

∈ Q.

3 / 11

slide-11
SLIDE 11

Orbit and Kakutani equivalence

Orbit equivalence (Tt) and (St) are orbit equivalent, if there exists a invertible transformation that maps orbits of (Tt) to orbits of (St). (Dye’s theorem, 1959) Every two ergodic flows are orbit equivalent. Kakutani equivalence, 1943 (Tt) : (X, B, µ) → (X, B, µ) and (St) : (Y , C, ν) → (Y , C, ν) are Kakutani equivalent (denoted (Tt)

K

∼ (St)), if they have isomorphic sections, i.e. can be represented as special flows over the same transformation. Standard flows Let (Rα

t ) : T2 → T2, Rα t (x, y) = (x + t, y + tα). Then, for every

α, β / ∈ Q, (Rα

t ) K

∼ (Rβ

t ) (Katok, 1976; Ornstein, Rudolph, Weiss,

1982); (Tt) is standard, if (Tt)

K

∼ (Rα

t ) for some α /

∈ Q.

3 / 11

slide-12
SLIDE 12

Some results on Kakutani equivalence

Standard systems finite rank systems (Ornstein, Rudolph, Weiss, 1982); closed under factors, inverse limits, compact extensions (Katok, 1976; Ornstein, Rudolph, Weiss, 1982) horocycle flows (Ratner, 1978) Non-standard systems first example due to Feldman, 1975; uncountably many pairwise non Kakutani equivalent systems (Ornstein, Rudolph, Weiss, 1982); Let (ht) denote the horocycle flow on SL(2, R)/Γ. Then (ht)k K ≁ (ht)l for k = l (Ratner, 1980).

4 / 11

slide-13
SLIDE 13

Some results on Kakutani equivalence

Standard systems finite rank systems (Ornstein, Rudolph, Weiss, 1982); closed under factors, inverse limits, compact extensions (Katok, 1976; Ornstein, Rudolph, Weiss, 1982) horocycle flows (Ratner, 1978) Non-standard systems first example due to Feldman, 1975; uncountably many pairwise non Kakutani equivalent systems (Ornstein, Rudolph, Weiss, 1982); Let (ht) denote the horocycle flow on SL(2, R)/Γ. Then (ht)k K ≁ (ht)l for k = l (Ratner, 1980).

4 / 11

slide-14
SLIDE 14

Some results on Kakutani equivalence

Standard systems finite rank systems (Ornstein, Rudolph, Weiss, 1982); closed under factors, inverse limits, compact extensions (Katok, 1976; Ornstein, Rudolph, Weiss, 1982) horocycle flows (Ratner, 1978) Non-standard systems first example due to Feldman, 1975; uncountably many pairwise non Kakutani equivalent systems (Ornstein, Rudolph, Weiss, 1982); Let (ht) denote the horocycle flow on SL(2, R)/Γ. Then (ht)k K ≁ (ht)l for k = l (Ratner, 1980).

4 / 11

slide-15
SLIDE 15

Some results on Kakutani equivalence

Standard systems finite rank systems (Ornstein, Rudolph, Weiss, 1982); closed under factors, inverse limits, compact extensions (Katok, 1976; Ornstein, Rudolph, Weiss, 1982) horocycle flows (Ratner, 1978) Non-standard systems first example due to Feldman, 1975; uncountably many pairwise non Kakutani equivalent systems (Ornstein, Rudolph, Weiss, 1982); Let (ht) denote the horocycle flow on SL(2, R)/Γ. Then (ht)k K ≁ (ht)l for k = l (Ratner, 1980).

4 / 11

slide-16
SLIDE 16

Some results on Kakutani equivalence

Standard systems finite rank systems (Ornstein, Rudolph, Weiss, 1982); closed under factors, inverse limits, compact extensions (Katok, 1976; Ornstein, Rudolph, Weiss, 1982) horocycle flows (Ratner, 1978) Non-standard systems first example due to Feldman, 1975; uncountably many pairwise non Kakutani equivalent systems (Ornstein, Rudolph, Weiss, 1982); Let (ht) denote the horocycle flow on SL(2, R)/Γ. Then (ht)k K ≁ (ht)l for k = l (Ratner, 1980).

4 / 11

slide-17
SLIDE 17

Some results on Kakutani equivalence

Standard systems finite rank systems (Ornstein, Rudolph, Weiss, 1982); closed under factors, inverse limits, compact extensions (Katok, 1976; Ornstein, Rudolph, Weiss, 1982) horocycle flows (Ratner, 1978) Non-standard systems first example due to Feldman, 1975; uncountably many pairwise non Kakutani equivalent systems (Ornstein, Rudolph, Weiss, 1982); Let (ht) denote the horocycle flow on SL(2, R)/Γ. Then (ht)k K ≁ (ht)l for k = l (Ratner, 1980).

4 / 11

slide-18
SLIDE 18

Some results on Kakutani equivalence

Standard systems finite rank systems (Ornstein, Rudolph, Weiss, 1982); closed under factors, inverse limits, compact extensions (Katok, 1976; Ornstein, Rudolph, Weiss, 1982) horocycle flows (Ratner, 1978) Non-standard systems first example due to Feldman, 1975; uncountably many pairwise non Kakutani equivalent systems (Ornstein, Rudolph, Weiss, 1982); Let (ht) denote the horocycle flow on SL(2, R)/Γ. Then (ht)k K ≁ (ht)l for k = l (Ratner, 1980).

4 / 11

slide-19
SLIDE 19

Kakutani invariant of M. Ratner

Kakutani invariant (Ratner, 1980) (Tt) → e((Tt)) ∈ [0, +∞]; If (Tt)

K

∼ (St), then e((Tt)) = e((St)). ¯ f -metric Fix a finite partition P and ǫ > 0. For N > 0, x, y ∈ X are (ǫ, P)- matchable for time N if there exists a set A ⊂ [0, N], |A| > (1 − ǫ)N and an increasing, measure preserving map h : A → h(A) such that Ttx and Th(t)y are in one atom of P for t ∈ A.

5 / 11

slide-20
SLIDE 20

Kakutani invariant of M. Ratner

Kakutani invariant (Ratner, 1980) (Tt) → e((Tt)) ∈ [0, +∞]; If (Tt)

K

∼ (St), then e((Tt)) = e((St)). ¯ f -metric Fix a finite partition P and ǫ > 0. For N > 0, x, y ∈ X are (ǫ, P)- matchable for time N if there exists a set A ⊂ [0, N], |A| > (1 − ǫ)N and an increasing, measure preserving map h : A → h(A) such that Ttx and Th(t)y are in one atom of P for t ∈ A.

5 / 11

slide-21
SLIDE 21

Kakutani invariant of M. Ratner

Kakutani invariant (Ratner, 1980) (Tt) → e((Tt)) ∈ [0, +∞]; If (Tt)

K

∼ (St), then e((Tt)) = e((St)). ¯ f -metric Fix a finite partition P and ǫ > 0. For N > 0, x, y ∈ X are (ǫ, P)- matchable for time N if there exists a set A ⊂ [0, N], |A| > (1 − ǫ)N and an increasing, measure preserving map h : A → h(A) such that Ttx and Th(t)y are in one atom of P for t ∈ A.

5 / 11

slide-22
SLIDE 22

Kakutani invariant of M. Ratner

Kakutani invariant (Ratner, 1980) (Tt) → e((Tt)) ∈ [0, +∞]; If (Tt)

K

∼ (St), then e((Tt)) = e((St)). ¯ f -metric Fix a finite partition P and ǫ > 0. For N > 0, x, y ∈ X are (ǫ, P)- matchable for time N if there exists a set A ⊂ [0, N], |A| > (1 − ǫ)N and an increasing, measure preserving map h : A → h(A) such that Ttx and Th(t)y are in one atom of P for t ∈ A.

5 / 11

slide-23
SLIDE 23

Kakutani equivalence of unipotent flows

Ratner’s problem, 1994 What can be said about Kakutani equivalence for unipotent flows

  • n quotients of semisimple Lie groups?

Setting G is a semisimple matrix Lie group with Lie algebra Lie(G); U ∈ Lie(G) is such that adU is nilpotent, where adU : Lie(G) → Lie(G), adU(V ) = [U, V ]; Γ is a uniform lattice in G; φU

t : G/Γ → G/Γ, φU t (xΓ) = exp(tU)xΓ.

6 / 11

slide-24
SLIDE 24

Kakutani equivalence of unipotent flows

Ratner’s problem, 1994 What can be said about Kakutani equivalence for unipotent flows

  • n quotients of semisimple Lie groups?

Setting G is a semisimple matrix Lie group with Lie algebra Lie(G); U ∈ Lie(G) is such that adU is nilpotent, where adU : Lie(G) → Lie(G), adU(V ) = [U, V ]; Γ is a uniform lattice in G; φU

t : G/Γ → G/Γ, φU t (xΓ) = exp(tU)xΓ.

6 / 11

slide-25
SLIDE 25

Kakutani equivalence of unipotent flows

Ratner’s problem, 1994 What can be said about Kakutani equivalence for unipotent flows

  • n quotients of semisimple Lie groups?

Setting G is a semisimple matrix Lie group with Lie algebra Lie(G); U ∈ Lie(G) is such that adU is nilpotent, where adU : Lie(G) → Lie(G), adU(V ) = [U, V ]; Γ is a uniform lattice in G; φU

t : G/Γ → G/Γ, φU t (xΓ) = exp(tU)xΓ.

6 / 11

slide-26
SLIDE 26

Kakutani equivalence of unipotent flows

Ratner’s problem, 1994 What can be said about Kakutani equivalence for unipotent flows

  • n quotients of semisimple Lie groups?

Setting G is a semisimple matrix Lie group with Lie algebra Lie(G); U ∈ Lie(G) is such that adU is nilpotent, where adU : Lie(G) → Lie(G), adU(V ) = [U, V ]; Γ is a uniform lattice in G; φU

t : G/Γ → G/Γ, φU t (xΓ) = exp(tU)xΓ.

6 / 11

slide-27
SLIDE 27

Kakutani equivalence of unipotent flows

Ratner’s problem, 1994 What can be said about Kakutani equivalence for unipotent flows

  • n quotients of semisimple Lie groups?

Setting G is a semisimple matrix Lie group with Lie algebra Lie(G); U ∈ Lie(G) is such that adU is nilpotent, where adU : Lie(G) → Lie(G), adU(V ) = [U, V ]; Γ is a uniform lattice in G; φU

t : G/Γ → G/Γ, φU t (xΓ) = exp(tU)xΓ.

6 / 11

slide-28
SLIDE 28

Chain basis

Chain basis for unipotent elements U ∈ Lie(G) is a unipotent element. There exists a basis (X i

j )1≤j≤mi,1≤i≤K, of Lie(G) such that

X i

mi adU

→ X i

mi−1 adU

→ . . .

adU

→ X i

1 adU

→ 0,

for every 1 ≤ i ≤ K. In particular, X i

1 ∈ C(U) for 1 ≤ i ≤ K.

Growth number of U Let GR(U) := 1 2

K

  • i=1

mi(mi − 1).

7 / 11

slide-29
SLIDE 29

Chain basis

Chain basis for unipotent elements U ∈ Lie(G) is a unipotent element. There exists a basis (X i

j )1≤j≤mi,1≤i≤K, of Lie(G) such that

X i

mi adU

→ X i

mi−1 adU

→ . . .

adU

→ X i

1 adU

→ 0,

for every 1 ≤ i ≤ K. In particular, X i

1 ∈ C(U) for 1 ≤ i ≤ K.

Growth number of U Let GR(U) := 1 2

K

  • i=1

mi(mi − 1).

7 / 11

slide-30
SLIDE 30

Main theorem

Main theorem (K., Vinhage, Wei, 2018) Let (φU

t ) be a unipotent flow on G/Γ. Then

e((φU

t )) = GR(U) − 3.

Moreover, if GR(U) = 3, then (φU

t ) is standard.

Corollaries The only standard unipotent flows are of the form id × 1 t 1

  • acting on (G × SL(2, R))/Γ, where Γ is

irreducible; If dim G > 3 and G is simple, then no unipotent flow on G/Γ is standard.

8 / 11

slide-31
SLIDE 31

Main theorem

Main theorem (K., Vinhage, Wei, 2018) Let (φU

t ) be a unipotent flow on G/Γ. Then

e((φU

t )) = GR(U) − 3.

Moreover, if GR(U) = 3, then (φU

t ) is standard.

Corollaries The only standard unipotent flows are of the form id × 1 t 1

  • acting on (G × SL(2, R))/Γ, where Γ is

irreducible; If dim G > 3 and G is simple, then no unipotent flow on G/Γ is standard.

8 / 11

slide-32
SLIDE 32

Main theorem

Main theorem (K., Vinhage, Wei, 2018) Let (φU

t ) be a unipotent flow on G/Γ. Then

e((φU

t )) = GR(U) − 3.

Moreover, if GR(U) = 3, then (φU

t ) is standard.

Corollaries The only standard unipotent flows are of the form id × 1 t 1

  • acting on (G × SL(2, R))/Γ, where Γ is

irreducible; If dim G > 3 and G is simple, then no unipotent flow on G/Γ is standard.

8 / 11

slide-33
SLIDE 33

Examples

Jakobson-Morozov theorem For every unipotent U ∈ Lie(G), there exists V , X ∈ Lie(G) such that [X, U] = 2U, [X, V ] = −2V , [U, V ] = X. So, V → X → −2U is always one of the chains and hence GR(U) ≥ 3. Examples e((ht)k) = 3k − 3; Let U =   1 1   ∈ Lie(SL(3, R)). Then e((φU

t )) = 10.

9 / 11

slide-34
SLIDE 34

Examples

Jakobson-Morozov theorem For every unipotent U ∈ Lie(G), there exists V , X ∈ Lie(G) such that [X, U] = 2U, [X, V ] = −2V , [U, V ] = X. So, V → X → −2U is always one of the chains and hence GR(U) ≥ 3. Examples e((ht)k) = 3k − 3; Let U =   1 1   ∈ Lie(SL(3, R)). Then e((φU

t )) = 10.

9 / 11

slide-35
SLIDE 35

Examples

Jakobson-Morozov theorem For every unipotent U ∈ Lie(G), there exists V , X ∈ Lie(G) such that [X, U] = 2U, [X, V ] = −2V , [U, V ] = X. So, V → X → −2U is always one of the chains and hence GR(U) ≥ 3. Examples e((ht)k) = 3k − 3; Let U =   1 1   ∈ Lie(SL(3, R)). Then e((φU

t )) = 10.

9 / 11

slide-36
SLIDE 36

Examples

Jakobson-Morozov theorem For every unipotent U ∈ Lie(G), there exists V , X ∈ Lie(G) such that [X, U] = 2U, [X, V ] = −2V , [U, V ] = X. So, V → X → −2U is always one of the chains and hence GR(U) ≥ 3. Examples e((ht)k) = 3k − 3; Let U =   1 1   ∈ Lie(SL(3, R)). Then e((φU

t )) = 10.

9 / 11

slide-37
SLIDE 37

Some questions

Questions Is the flow generated by U on G/Γ Kakutani equivalent to the flow generated by U on G/Γ′? Is the Kakutani invariant a full invariant in the class of unipotent flows?

10 / 11

slide-38
SLIDE 38

Some questions

Questions Is the flow generated by U on G/Γ Kakutani equivalent to the flow generated by U on G/Γ′? Is the Kakutani invariant a full invariant in the class of unipotent flows?

10 / 11

slide-39
SLIDE 39

Some questions

Questions Is the flow generated by U on G/Γ Kakutani equivalent to the flow generated by U on G/Γ′? Is the Kakutani invariant a full invariant in the class of unipotent flows?

10 / 11

slide-40
SLIDE 40

THANK YOU !

11 / 11