Non-uniform Self-Moduli Peter Gerdes Group in Logic University of - - PowerPoint PPT Presentation

non uniform self moduli
SMART_READER_LITE
LIVE PREVIEW

Non-uniform Self-Moduli Peter Gerdes Group in Logic University of - - PowerPoint PPT Presentation

Background Uniform Moduli Nonuniform Moduli Non-uniform Self-Moduli Peter Gerdes Group in Logic University of California, Berkeley 2007-08 ASL Winter Meeting Peter Gerdes Non-uniform Self-Moduli Background Uniform Moduli Nonuniform


slide-1
SLIDE 1

Background Uniform Moduli Nonuniform Moduli

Non-uniform Self-Moduli

Peter Gerdes

Group in Logic University of California, Berkeley

2007-08 ASL Winter Meeting

Peter Gerdes Non-uniform Self-Moduli

slide-2
SLIDE 2

Background Uniform Moduli Nonuniform Moduli

Definitions and Notation

Notation Say σ ≻ τ if σ(n) ≥ τ(n) everywhere they are both defined. So if f, g ∈ ωω then f ≻ g ↔ (∀n)[f(n) ≥ g(n)] Definitions Let f ∈ ωω and X ⊂ ω. f is a modulus (of computation) for X if for all g ∈ ωω if g ≻ f = ⇒ g ≥T X. f is a uniform modulus for X if there is a recursive functional Φ such that g ≻ f = ⇒ Φ (g) = X. f is a self-modulus if f is a modulus for f

Peter Gerdes Non-uniform Self-Moduli

slide-3
SLIDE 3

Background Uniform Moduli Nonuniform Moduli

Definitions and Notation

Notation Say σ ≻ τ if σ(n) ≥ τ(n) everywhere they are both defined. So if f, g ∈ ωω then f ≻ g ↔ (∀n)[f(n) ≥ g(n)] Definitions Let f ∈ ωω and X ⊂ ω. f is a modulus (of computation) for X if for all g ∈ ωω if g ≻ f = ⇒ g ≥T X. f is a uniform modulus for X if there is a recursive functional Φ such that g ≻ f = ⇒ Φ (g) = X. f is a self-modulus if f is a modulus for f

Peter Gerdes Non-uniform Self-Moduli

slide-4
SLIDE 4

Background Uniform Moduli Nonuniform Moduli

Definitions and Notation

Notation Say σ ≻ τ if σ(n) ≥ τ(n) everywhere they are both defined. So if f, g ∈ ωω then f ≻ g ↔ (∀n)[f(n) ≥ g(n)] Definitions Let f ∈ ωω and X ⊂ ω. f is a modulus (of computation) for X if for all g ∈ ωω if g ≻ f = ⇒ g ≥T X. f is a uniform modulus for X if there is a recursive functional Φ such that g ≻ f = ⇒ Φ (g) = X. f is a self-modulus if f is a modulus for f

Peter Gerdes Non-uniform Self-Moduli

slide-5
SLIDE 5

Background Uniform Moduli Nonuniform Moduli

Definitions and Notation

Notation Say σ ≻ τ if σ(n) ≥ τ(n) everywhere they are both defined. So if f, g ∈ ωω then f ≻ g ↔ (∀n)[f(n) ≥ g(n)] Definitions Let f ∈ ωω and X ⊂ ω. f is a modulus (of computation) for X if for all g ∈ ωω if g ≻ f = ⇒ g ≥T X. f is a uniform modulus for X if there is a recursive functional Φ such that g ≻ f = ⇒ Φ (g) = X. f is a self-modulus if f is a modulus for f

Peter Gerdes Non-uniform Self-Moduli

slide-6
SLIDE 6

Background Uniform Moduli Nonuniform Moduli

Moduli of Computation

f Let f be a modulus for X. Then g1 ≻ f = ⇒ g1 ≥T X Same with g2 f is a uniform modulus if the same reduction works for all g ≻ f. Suppose h is faster growing than f. Then h computes X.

Peter Gerdes Non-uniform Self-Moduli

slide-7
SLIDE 7

Background Uniform Moduli Nonuniform Moduli

Moduli of Computation

f g1 Let f be a modulus for X. Then g1 ≻ f = ⇒ g1 ≥T X Same with g2 f is a uniform modulus if the same reduction works for all g ≻ f. Suppose h is faster growing than f. Then h computes X.

Peter Gerdes Non-uniform Self-Moduli

slide-8
SLIDE 8

Background Uniform Moduli Nonuniform Moduli

Moduli of Computation

f g1 g2 Let f be a modulus for X. Then g1 ≻ f = ⇒ g1 ≥T X Same with g2 f is a uniform modulus if the same reduction works for all g ≻ f. Suppose h is faster growing than f. Then h computes X.

Peter Gerdes Non-uniform Self-Moduli

slide-9
SLIDE 9

Background Uniform Moduli Nonuniform Moduli

Moduli of Computation

f g1 g2 Let f be a modulus for X. Then g1 ≻ f = ⇒ g1 ≥T X Same with g2 f is a uniform modulus if the same reduction works for all g ≻ f. Suppose h is faster growing than f. Then h computes X.

Peter Gerdes Non-uniform Self-Moduli

slide-10
SLIDE 10

Background Uniform Moduli Nonuniform Moduli

Moduli of Computation

f g1 g2 h Let f be a modulus for X. Then g1 ≻ f = ⇒ g1 ≥T X Same with g2 f is a uniform modulus if the same reduction works for all g ≻ f. Suppose h is faster growing than f. Then h computes X.

Peter Gerdes Non-uniform Self-Moduli

slide-11
SLIDE 11

Background Uniform Moduli Nonuniform Moduli

Moduli of Computation

f g1 g2 h Let f be a modulus for X. Then g1 ≻ f = ⇒ g1 ≥T X Same with g2 f is a uniform modulus if the same reduction works for all g ≻ f. Suppose h is faster growing than f. Then h computes X.

Peter Gerdes Non-uniform Self-Moduli

slide-12
SLIDE 12

Background Uniform Moduli Nonuniform Moduli

Basic Facts

Observation Every α-REA degree has a uniform self-modulus. Observation Every ∆0

2 degree has a uniform self-modulus.

Modify proof that ∆0

2 degrees are hyperimmune.

Theorem (Slaman and Groszek) There is a uniform self-modulus that computes no non-recursive ∆0

2-set.

Theorem For every α there is a uniform self-modulus that computes no non-recursive ∆0

α-set. Peter Gerdes Non-uniform Self-Moduli

slide-13
SLIDE 13

Background Uniform Moduli Nonuniform Moduli

Basic Facts

Observation Every α-REA degree has a uniform self-modulus. Observation Every ∆0

2 degree has a uniform self-modulus.

Modify proof that ∆0

2 degrees are hyperimmune.

Theorem (Slaman and Groszek) There is a uniform self-modulus that computes no non-recursive ∆0

2-set.

Theorem For every α there is a uniform self-modulus that computes no non-recursive ∆0

α-set. Peter Gerdes Non-uniform Self-Moduli

slide-14
SLIDE 14

Background Uniform Moduli Nonuniform Moduli

Basic Facts

Observation Every α-REA degree has a uniform self-modulus. Observation Every ∆0

2 degree has a uniform self-modulus.

Modify proof that ∆0

2 degrees are hyperimmune.

Theorem (Slaman and Groszek) There is a uniform self-modulus that computes no non-recursive ∆0

2-set.

Theorem For every α there is a uniform self-modulus that computes no non-recursive ∆0

α-set. Peter Gerdes Non-uniform Self-Moduli

slide-15
SLIDE 15

Background Uniform Moduli Nonuniform Moduli

Basic Facts

Observation Every α-REA degree has a uniform self-modulus. Observation Every ∆0

2 degree has a uniform self-modulus.

Modify proof that ∆0

2 degrees are hyperimmune.

Theorem (Slaman and Groszek) There is a uniform self-modulus that computes no non-recursive ∆0

2-set.

Theorem For every α there is a uniform self-modulus that computes no non-recursive ∆0

α-set. Peter Gerdes Non-uniform Self-Moduli

slide-16
SLIDE 16

Background Uniform Moduli Nonuniform Moduli

Basic Facts

Observation Every α-REA degree has a uniform self-modulus. Observation Every ∆0

2 degree has a uniform self-modulus.

Modify proof that ∆0

2 degrees are hyperimmune.

Theorem (Slaman and Groszek) There is a uniform self-modulus that computes no non-recursive ∆0

2-set.

Theorem For every α there is a uniform self-modulus that computes no non-recursive ∆0

α-set. Peter Gerdes Non-uniform Self-Moduli

slide-17
SLIDE 17

Background Uniform Moduli Nonuniform Moduli

What Degrees Have Moduli?

Theorem (Slaman and Groszek) X has a modulus if and only if X is ∆1

1.

Proof. ⇐ 0

  • (α) has a uniform self-modulus. Call it θα

⇒ If X has a modulus f then it must also have a uniform modulus ˆ f.

Try to build g ≻ f, g T X with Hechler conditions. This must fail producing a uniform modulus (and uniform reduction).

A uniform reduction provides a ∆1

1 definition for X

The uniform modulus produced may be very complex relative to the modulus.

Peter Gerdes Non-uniform Self-Moduli

slide-18
SLIDE 18

Background Uniform Moduli Nonuniform Moduli

What Degrees Have Moduli?

Theorem (Slaman and Groszek) X has a modulus if and only if X is ∆1

1.

Proof. ⇐ 0

  • (α) has a uniform self-modulus. Call it θα

⇒ If X has a modulus f then it must also have a uniform modulus ˆ f.

Try to build g ≻ f, g T X with Hechler conditions. This must fail producing a uniform modulus (and uniform reduction).

A uniform reduction provides a ∆1

1 definition for X

The uniform modulus produced may be very complex relative to the modulus.

Peter Gerdes Non-uniform Self-Moduli

slide-19
SLIDE 19

Background Uniform Moduli Nonuniform Moduli

What Degrees Have Moduli?

Theorem (Slaman and Groszek) X has a modulus if and only if X is ∆1

1.

Proof. ⇐ 0

  • (α) has a uniform self-modulus. Call it θα

⇒ If X has a modulus f then it must also have a uniform modulus ˆ f.

Try to build g ≻ f, g T X with Hechler conditions. This must fail producing a uniform modulus (and uniform reduction).

A uniform reduction provides a ∆1

1 definition for X

The uniform modulus produced may be very complex relative to the modulus.

Peter Gerdes Non-uniform Self-Moduli

slide-20
SLIDE 20

Background Uniform Moduli Nonuniform Moduli

What Degrees Have Moduli?

Theorem (Slaman and Groszek) X has a modulus if and only if X is ∆1

1.

Proof. ⇐ 0

  • (α) has a uniform self-modulus. Call it θα

⇒ If X has a modulus f then it must also have a uniform modulus ˆ f.

Try to build g ≻ f, g T X with Hechler conditions. This must fail producing a uniform modulus (and uniform reduction).

A uniform reduction provides a ∆1

1 definition for X

The uniform modulus produced may be very complex relative to the modulus.

Peter Gerdes Non-uniform Self-Moduli

slide-21
SLIDE 21

Background Uniform Moduli Nonuniform Moduli

What Degrees Have Moduli?

Theorem (Slaman and Groszek) X has a modulus if and only if X is ∆1

1.

Proof. ⇐ 0

  • (α) has a uniform self-modulus. Call it θα

⇒ If X has a modulus f then it must also have a uniform modulus ˆ f.

Try to build g ≻ f, g T X with Hechler conditions. This must fail producing a uniform modulus (and uniform reduction).

A uniform reduction provides a ∆1

1 definition for X

The uniform modulus produced may be very complex relative to the modulus.

Peter Gerdes Non-uniform Self-Moduli

slide-22
SLIDE 22

Background Uniform Moduli Nonuniform Moduli

What Degrees Have Moduli?

Theorem (Slaman and Groszek) X has a modulus if and only if X is ∆1

1.

Proof. ⇐ 0

  • (α) has a uniform self-modulus. Call it θα

⇒ If X has a modulus f then it must also have a uniform modulus ˆ f.

Try to build g ≻ f, g T X with Hechler conditions. This must fail producing a uniform modulus (and uniform reduction).

A uniform reduction provides a ∆1

1 definition for X

The uniform modulus produced may be very complex relative to the modulus.

Peter Gerdes Non-uniform Self-Moduli

slide-23
SLIDE 23

Background Uniform Moduli Nonuniform Moduli

What Degrees Have Moduli?

Theorem (Slaman and Groszek) X has a modulus if and only if X is ∆1

1.

Proof. ⇐ 0

  • (α) has a uniform self-modulus. Call it θα

⇒ If X has a modulus f then it must also have a uniform modulus ˆ f.

Try to build g ≻ f, g T X with Hechler conditions. This must fail producing a uniform modulus (and uniform reduction).

A uniform reduction provides a ∆1

1 definition for X

The uniform modulus produced may be very complex relative to the modulus.

Peter Gerdes Non-uniform Self-Moduli

slide-24
SLIDE 24

Background Uniform Moduli Nonuniform Moduli

Uniformity

Question Can we bound the complexity of a uniform modulus for X relative to a modulus for X? Sufficent to examine self-moduli. Particularly interesting since there is a nice characterization of degrees with uniform self-moduli but not (yet?) for degrees with self-moduli. Theorem d

  • contains a uniform self-modulus iff d
  • contains a Π0

2 singleton.

Peter Gerdes Non-uniform Self-Moduli

slide-25
SLIDE 25

Background Uniform Moduli Nonuniform Moduli

Uniformity

Question Can we bound the complexity of a uniform modulus for X relative to a modulus for X? Sufficent to examine self-moduli. Particularly interesting since there is a nice characterization of degrees with uniform self-moduli but not (yet?) for degrees with self-moduli. Theorem d

  • contains a uniform self-modulus iff d
  • contains a Π0

2 singleton.

Peter Gerdes Non-uniform Self-Moduli

slide-26
SLIDE 26

Background Uniform Moduli Nonuniform Moduli

Uniformity

Question Can we bound the complexity of a uniform modulus for X relative to a modulus for X? Sufficent to examine self-moduli. Particularly interesting since there is a nice characterization of degrees with uniform self-moduli but not (yet?) for degrees with self-moduli. Theorem d

  • contains a uniform self-modulus iff d
  • contains a Π0

2 singleton.

Peter Gerdes Non-uniform Self-Moduli

slide-27
SLIDE 27

Background Uniform Moduli Nonuniform Moduli

Partial Answer

Theorem For all n ∈ ω there is a self-modulus f so that no h ≤T f (n) is a uniform modulus for f. Remark Going past ω is deceptively hard. Plan

1

Find a simple property guaranteeing no h ≤T f (n) is a uniform modulus for f.

2

Build a self-modulus satisfying this property.

Peter Gerdes Non-uniform Self-Moduli

slide-28
SLIDE 28

Background Uniform Moduli Nonuniform Moduli

Partial Answer

Theorem For all n ∈ ω there is a self-modulus f so that no h ≤T f (n) is a uniform modulus for f. Remark Going past ω is deceptively hard. Plan

1

Find a simple property guaranteeing no h ≤T f (n) is a uniform modulus for f.

2

Build a self-modulus satisfying this property.

Peter Gerdes Non-uniform Self-Moduli

slide-29
SLIDE 29

Background Uniform Moduli Nonuniform Moduli

Partial Answer

Theorem For all n ∈ ω there is a self-modulus f so that no h ≤T f (n) is a uniform modulus for f. Remark Going past ω is deceptively hard. Plan

1

Find a simple property guaranteeing no h ≤T f (n) is a uniform modulus for f.

2

Build a self-modulus satisfying this property.

Peter Gerdes Non-uniform Self-Moduli

slide-30
SLIDE 30

Background Uniform Moduli Nonuniform Moduli

Partial Answer

Theorem For all n ∈ ω there is a self-modulus f so that no h ≤T f (n) is a uniform modulus for f. Remark Going past ω is deceptively hard. Plan

1

Find a simple property guaranteeing no h ≤T f (n) is a uniform modulus for f.

2

Build a self-modulus satisfying this property.

Peter Gerdes Non-uniform Self-Moduli

slide-31
SLIDE 31

Background Uniform Moduli Nonuniform Moduli

Avoiding Uniformity

Lemma If f is n + 2 locally generic on a perfect tree T no h ≤T f (n) is a uniform modulus for f. Proof. Suppose Φ witnesses h = ϕi(f (n)) violates the lemma. Pick k so f ↾k forces both that:

1

h = ϕi(f (n)) is total.

2

If σ ∈ ω<ω and σ ≻ h then Φ (σ) ⊂ f

Let ˆ f ⊃ f ↾k be a distinct n + 2 generic path through T. h and ˆ h must be total so pick g ≻ h, ˆ h. But Φ (g) ⊂ f and Φ (g) ⊂ ˆ f so it can’t be total.

Peter Gerdes Non-uniform Self-Moduli

slide-32
SLIDE 32

Background Uniform Moduli Nonuniform Moduli

Avoiding Uniformity

Lemma If f is n + 2 locally generic on a perfect tree T no h ≤T f (n) is a uniform modulus for f. Proof. Suppose Φ witnesses h = ϕi(f (n)) violates the lemma. Pick k so f ↾k forces both that:

1

h = ϕi(f (n)) is total.

2

If σ ∈ ω<ω and σ ≻ h then Φ (σ) ⊂ f

Let ˆ f ⊃ f ↾k be a distinct n + 2 generic path through T. h and ˆ h must be total so pick g ≻ h, ˆ h. But Φ (g) ⊂ f and Φ (g) ⊂ ˆ f so it can’t be total.

Peter Gerdes Non-uniform Self-Moduli

slide-33
SLIDE 33

Background Uniform Moduli Nonuniform Moduli

Avoiding Uniformity

Lemma If f is n + 2 locally generic on a perfect tree T no h ≤T f (n) is a uniform modulus for f. Proof. Suppose Φ witnesses h = ϕi(f (n)) violates the lemma. Pick k so f ↾k forces both that:

1

h = ϕi(f (n)) is total.

2

If σ ∈ ω<ω and σ ≻ h then Φ (σ) ⊂ f

Let ˆ f ⊃ f ↾k be a distinct n + 2 generic path through T. h and ˆ h must be total so pick g ≻ h, ˆ h. But Φ (g) ⊂ f and Φ (g) ⊂ ˆ f so it can’t be total.

Peter Gerdes Non-uniform Self-Moduli

slide-34
SLIDE 34

Background Uniform Moduli Nonuniform Moduli

Avoiding Uniformity

Lemma If f is n + 2 locally generic on a perfect tree T no h ≤T f (n) is a uniform modulus for f. Proof. Suppose Φ witnesses h = ϕi(f (n)) violates the lemma. Pick k so f ↾k forces both that:

1

h = ϕi(f (n)) is total.

2

If σ ∈ ω<ω and σ ≻ h then Φ (σ) ⊂ f

Let ˆ f ⊃ f ↾k be a distinct n + 2 generic path through T. h and ˆ h must be total so pick g ≻ h, ˆ h. But Φ (g) ⊂ f and Φ (g) ⊂ ˆ f so it can’t be total.

Peter Gerdes Non-uniform Self-Moduli

slide-35
SLIDE 35

Background Uniform Moduli Nonuniform Moduli

Avoiding Uniformity

Lemma If f is n + 2 locally generic on a perfect tree T no h ≤T f (n) is a uniform modulus for f. Proof. Suppose Φ witnesses h = ϕi(f (n)) violates the lemma. Pick k so f ↾k forces both that:

1

h = ϕi(f (n)) is total.

2

If σ ∈ ω<ω and σ ≻ h then Φ (σ) ⊂ f

Let ˆ f ⊃ f ↾k be a distinct n + 2 generic path through T. h and ˆ h must be total so pick g ≻ h, ˆ h. But Φ (g) ⊂ f and Φ (g) ⊂ ˆ f so it can’t be total.

Peter Gerdes Non-uniform Self-Moduli

slide-36
SLIDE 36

Background Uniform Moduli Nonuniform Moduli

Avoiding Uniformity

Lemma If f is n + 2 locally generic on a perfect tree T no h ≤T f (n) is a uniform modulus for f. Proof. Suppose Φ witnesses h = ϕi(f (n)) violates the lemma. Pick k so f ↾k forces both that:

1

h = ϕi(f (n)) is total.

2

If σ ∈ ω<ω and σ ≻ h then Φ (σ) ⊂ f

Let ˆ f ⊃ f ↾k be a distinct n + 2 generic path through T. h and ˆ h must be total so pick g ≻ h, ˆ h. But Φ (g) ⊂ f and Φ (g) ⊂ ˆ f so it can’t be total.

Peter Gerdes Non-uniform Self-Moduli

slide-37
SLIDE 37

Background Uniform Moduli Nonuniform Moduli

Guaranteeing Reductions

Uniform Reductions Build f computably in 0

  • (n+2)

If g ≻ θn+2 then (uniformly) g ≥T f How can we guarantee every ‘small’ g ≻ f computes f? Non-uniformity requires our procedure fails for ‘large’ g Idea! Use smallness of g to recover f. For each k < n + 2 encode f into locations f dips below θk. Since g ≻ f we can recover infinitely many of these locations.

Peter Gerdes Non-uniform Self-Moduli

slide-38
SLIDE 38

Background Uniform Moduli Nonuniform Moduli

Guaranteeing Reductions

Uniform Reductions Build f computably in 0

  • (n+2)

If g ≻ θn+2 then (uniformly) g ≥T f How can we guarantee every ‘small’ g ≻ f computes f? Non-uniformity requires our procedure fails for ‘large’ g Idea! Use smallness of g to recover f. For each k < n + 2 encode f into locations f dips below θk. Since g ≻ f we can recover infinitely many of these locations.

Peter Gerdes Non-uniform Self-Moduli

slide-39
SLIDE 39

Background Uniform Moduli Nonuniform Moduli

Guaranteeing Reductions

Uniform Reductions Build f computably in 0

  • (n+2)

If g ≻ θn+2 then (uniformly) g ≥T f How can we guarantee every ‘small’ g ≻ f computes f? Non-uniformity requires our procedure fails for ‘large’ g Idea! Use smallness of g to recover f. For each k < n + 2 encode f into locations f dips below θk. Since g ≻ f we can recover infinitely many of these locations.

Peter Gerdes Non-uniform Self-Moduli

slide-40
SLIDE 40

Background Uniform Moduli Nonuniform Moduli

First Attempt

Naive Strategy Create sequence of trees Tk ⊂ Tk−1 for 1 ≤ k ≤ n + 2 with Tk+1 representing our attempts to meet Σ0

k+1 sets on Tk.

Prune Tk to ensure at most one σ ∈ Tk+1 of length x − 1 satisfies σ(x) < θk+1(x) Let k be least such that g T 0

  • (k+1).

Infinitely often g must dip below θk+1. g can enumerate the set of x with g(x) < θk+1(x). f ↾x is unique σ ∈ Tk+1 with σ(x) ≤ g(x).

Peter Gerdes Non-uniform Self-Moduli

slide-41
SLIDE 41

Background Uniform Moduli Nonuniform Moduli

First Attempt

Naive Strategy Create sequence of trees Tk ⊂ Tk−1 for 1 ≤ k ≤ n + 2 with Tk+1 representing our attempts to meet Σ0

k+1 sets on Tk.

Prune Tk to ensure at most one σ ∈ Tk+1 of length x − 1 satisfies σ(x) < θk+1(x) Let k be least such that g T 0

  • (k+1).

Infinitely often g must dip below θk+1. g can enumerate the set of x with g(x) < θk+1(x). f ↾x is unique σ ∈ Tk+1 with σ(x) ≤ g(x).

Peter Gerdes Non-uniform Self-Moduli

slide-42
SLIDE 42

Background Uniform Moduli Nonuniform Moduli

First Attempt

Naive Strategy Create sequence of trees Tk ⊂ Tk−1 for 1 ≤ k ≤ n + 2 with Tk+1 representing our attempts to meet Σ0

k+1 sets on Tk.

Prune Tk to ensure at most one σ ∈ Tk+1 of length x − 1 satisfies σ(x) < θk+1(x) Let k be least such that g T 0

  • (k+1).

Infinitely often g must dip below θk+1. g can enumerate the set of x with g(x) < θk+1(x). f ↾x is unique σ ∈ Tk+1 with σ(x) ≤ g(x).

Peter Gerdes Non-uniform Self-Moduli

slide-43
SLIDE 43

Background Uniform Moduli Nonuniform Moduli

First Attempt

Naive Strategy Create sequence of trees Tk ⊂ Tk−1 for 1 ≤ k ≤ n + 2 with Tk+1 representing our attempts to meet Σ0

k+1 sets on Tk.

Prune Tk to ensure at most one σ ∈ Tk+1 of length x − 1 satisfies σ(x) < θk+1(x) Let k be least such that g T 0

  • (k+1).

Infinitely often g must dip below θk+1. g can enumerate the set of x with g(x) < θk+1(x). f ↾x is unique σ ∈ Tk+1 with σ(x) ≤ g(x).

Peter Gerdes Non-uniform Self-Moduli

slide-44
SLIDE 44

Background Uniform Moduli Nonuniform Moduli

First Attempt

Naive Strategy Create sequence of trees Tk ⊂ Tk−1 for 1 ≤ k ≤ n + 2 with Tk+1 representing our attempts to meet Σ0

k+1 sets on Tk.

Prune Tk to ensure at most one σ ∈ Tk+1 of length x − 1 satisfies σ(x) < θk+1(x) Let k be least such that g T 0

  • (k+1).

Infinitely often g must dip below θk+1. g can enumerate the set of x with g(x) < θk+1(x). f ↾x is unique σ ∈ Tk+1 with σ(x) ≤ g(x).

Peter Gerdes Non-uniform Self-Moduli

slide-45
SLIDE 45

Background Uniform Moduli Nonuniform Moduli

First Attempt

Naive Strategy Create sequence of trees Tk ⊂ Tk−1 for 1 ≤ k ≤ n + 2 with Tk+1 representing our attempts to meet Σ0

k+1 sets on Tk.

Prune Tk to ensure at most one σ ∈ Tk+1 of length x − 1 satisfies σ(x) < θk+1(x) Let k be least such that g T 0

  • (k+1).

Infinitely often g must dip below θk+1. g can enumerate the set of x with g(x) < θk+1(x). f ↾x is unique σ ∈ Tk+1 with σ(x) ≤ g(x).

Peter Gerdes Non-uniform Self-Moduli

slide-46
SLIDE 46

Background Uniform Moduli Nonuniform Moduli

Problems

1

Need to incorporate multiple strings from Tk in Tk+1 that aren’t above θk+1 Solution

τ ∈ Tk+1 must dip below θk+1 for a 0

  • (k)-long interval for

uniqueness. Achieved by ‘cancelling’ lower priority strings that dip in wrong places.

2

Tk+1 is a ∆0

k+2 set and g only computes 0

  • (k)

Solution Tk+1 = lims→∞ Tk+1[s] Use priority argument to ensure that g(x) is large enough to believe f ↾x∈ Tk+1 at true stages.

Peter Gerdes Non-uniform Self-Moduli

slide-47
SLIDE 47

Background Uniform Moduli Nonuniform Moduli

Problems

1

Need to incorporate multiple strings from Tk in Tk+1 that aren’t above θk+1 Solution

τ ∈ Tk+1 must dip below θk+1 for a 0

  • (k)-long interval for

uniqueness. Achieved by ‘cancelling’ lower priority strings that dip in wrong places.

2

Tk+1 is a ∆0

k+2 set and g only computes 0

  • (k)

Solution Tk+1 = lims→∞ Tk+1[s] Use priority argument to ensure that g(x) is large enough to believe f ↾x∈ Tk+1 at true stages.

Peter Gerdes Non-uniform Self-Moduli

slide-48
SLIDE 48

Background Uniform Moduli Nonuniform Moduli

Problems

1

Need to incorporate multiple strings from Tk in Tk+1 that aren’t above θk+1 Solution

τ ∈ Tk+1 must dip below θk+1 for a 0

  • (k)-long interval for

uniqueness. Achieved by ‘cancelling’ lower priority strings that dip in wrong places.

2

Tk+1 is a ∆0

k+2 set and g only computes 0

  • (k)

Solution Tk+1 = lims→∞ Tk+1[s] Use priority argument to ensure that g(x) is large enough to believe f ↾x∈ Tk+1 at true stages.

Peter Gerdes Non-uniform Self-Moduli

slide-49
SLIDE 49

Background Uniform Moduli Nonuniform Moduli

Problems

1

Need to incorporate multiple strings from Tk in Tk+1 that aren’t above θk+1 Solution

τ ∈ Tk+1 must dip below θk+1 for a 0

  • (k)-long interval for

uniqueness. Achieved by ‘cancelling’ lower priority strings that dip in wrong places.

2

Tk+1 is a ∆0

k+2 set and g only computes 0

  • (k)

Solution Tk+1 = lims→∞ Tk+1[s] Use priority argument to ensure that g(x) is large enough to believe f ↾x∈ Tk+1 at true stages.

Peter Gerdes Non-uniform Self-Moduli

slide-50
SLIDE 50

Background Uniform Moduli Nonuniform Moduli

Problems

1

Need to incorporate multiple strings from Tk in Tk+1 that aren’t above θk+1 Solution

τ ∈ Tk+1 must dip below θk+1 for a 0

  • (k)-long interval for

uniqueness. Achieved by ‘cancelling’ lower priority strings that dip in wrong places.

2

Tk+1 is a ∆0

k+2 set and g only computes 0

  • (k)

Solution Tk+1 = lims→∞ Tk+1[s] Use priority argument to ensure that g(x) is large enough to believe f ↾x∈ Tk+1 at true stages.

Peter Gerdes Non-uniform Self-Moduli

slide-51
SLIDE 51

Background Uniform Moduli Nonuniform Moduli

Picturing The Construction

g ≻ f searches for stage to commit to f using T1 ∆0

2 guess at T1 changes with stage.

Approximation to θ1 increasing to θ1 When g(x) < θ1(x)[s] then T1[s] gives unique value for f ↾x. g Function g that wants to compute f

Peter Gerdes Non-uniform Self-Moduli

slide-52
SLIDE 52

Background Uniform Moduli Nonuniform Moduli

Picturing The Construction

g ≻ f searches for stage to commit to f using T1 ∆0

2 guess at T1 changes with stage.

Approximation to θ1 increasing to θ1 When g(x) < θ1(x)[s] then T1[s] gives unique value for f ↾x. θ1 Final fast growing function of degree 0

  • ′.

Peter Gerdes Non-uniform Self-Moduli

slide-53
SLIDE 53

Background Uniform Moduli Nonuniform Moduli

Picturing The Construction

g ≻ f searches for stage to commit to f using T1 ∆0

2 guess at T1 changes with stage.

Approximation to θ1 increasing to θ1 When g(x) < θ1(x)[s] then T1[s] gives unique value for f ↾x. Final tree T1 of possible paths for f

Peter Gerdes Non-uniform Self-Moduli

slide-54
SLIDE 54

Background Uniform Moduli Nonuniform Moduli

Picturing The Construction

g ≻ f searches for stage to commit to f using T1 ∆0

2 guess at T1 changes with stage.

Approximation to θ1 increasing to θ1 When g(x) < θ1(x)[s] then T1[s] gives unique value for f ↾x. g θ1 step: 2 Membership in T1 changes during computation steps.

Peter Gerdes Non-uniform Self-Moduli

slide-55
SLIDE 55

Background Uniform Moduli Nonuniform Moduli

Picturing The Construction

g ≻ f searches for stage to commit to f using T1 ∆0

2 guess at T1 changes with stage.

Approximation to θ1 increasing to θ1 When g(x) < θ1(x)[s] then T1[s] gives unique value for f ↾x. g θ1 step: 3 Membership in T1 changes during computation steps.

Peter Gerdes Non-uniform Self-Moduli

slide-56
SLIDE 56

Background Uniform Moduli Nonuniform Moduli

Picturing The Construction

g ≻ f searches for stage to commit to f using T1 ∆0

2 guess at T1 changes with stage.

Approximation to θ1 increasing to θ1 When g(x) < θ1(x)[s] then T1[s] gives unique value for f ↾x. g θ1 step: 4 Membership in T1 changes during computation steps.

Peter Gerdes Non-uniform Self-Moduli

slide-57
SLIDE 57

Background Uniform Moduli Nonuniform Moduli

Picturing The Construction

g ≻ f searches for stage to commit to f using T1 ∆0

2 guess at T1 changes with stage.

Approximation to θ1 increasing to θ1 When g(x) < θ1(x)[s] then T1[s] gives unique value for f ↾x. g θ1 step: 5 Membership in T1 changes during computation steps.

Peter Gerdes Non-uniform Self-Moduli

slide-58
SLIDE 58

Background Uniform Moduli Nonuniform Moduli

Picturing The Construction

g ≻ f searches for stage to commit to f using T1 ∆0

2 guess at T1 changes with stage.

Approximation to θ1 increasing to θ1 When g(x) < θ1(x)[s] then T1[s] gives unique value for f ↾x. g θ1 step: 6 At this step g notices a value at which it is small.

Peter Gerdes Non-uniform Self-Moduli

slide-59
SLIDE 59

Background Uniform Moduli Nonuniform Moduli

Picturing The Construction

g ≻ f searches for stage to commit to f using T1 ∆0

2 guess at T1 changes with stage.

Approximation to θ1 increasing to θ1 When g(x) < θ1(x)[s] then T1[s] gives unique value for f ↾x. g θ1 step: 6 Construction guarantees that no false path is below g

Peter Gerdes Non-uniform Self-Moduli

slide-60
SLIDE 60

Background Uniform Moduli Nonuniform Moduli

Picturing The Construction

g ≻ f searches for stage to commit to f using T1 ∆0

2 guess at T1 changes with stage.

Approximation to θ1 increasing to θ1 When g(x) < θ1(x)[s] then T1[s] gives unique value for f ↾x. g θ1 step: 6 g can commit to an initial segment of f

Peter Gerdes Non-uniform Self-Moduli

slide-61
SLIDE 61

Background Uniform Moduli Nonuniform Moduli

Thanks

In no particular order: My advisor Leo Harrington for taking the time to talk about these issues with me. Theodore Slaman for introducing me to moduli of computation. The conference organizers for setting this all up.

Peter Gerdes Non-uniform Self-Moduli