Admissible rules of propositional dependence logic Fan Yang Utrecht - - PowerPoint PPT Presentation

admissible rules of propositional dependence logic
SMART_READER_LITE
LIVE PREVIEW

Admissible rules of propositional dependence logic Fan Yang Utrecht - - PowerPoint PPT Presentation

Admissible rules of propositional dependence logic Fan Yang Utrecht University Les Diablerets Jan 30 - Feb 2, 2015 Joint work with Rosalie Iemhoff 1/19 Outline propositional dependence logic


slide-1
SLIDE 1

Admissible rules of propositional dependence logic

Fan Yang

Utrecht University

Les Diablerets Jan 30 - Feb 2, 2015 ———————————————– Joint work with Rosalie Iemhoff

1/19

slide-2
SLIDE 2

Outline

1

propositional dependence logic

2

flat formulas and projective formulas

3

structural completeness

2/19

slide-3
SLIDE 3

Dependence between first-order variables

First Order Quantifiers: ∀x1∃y1∀x2∃y2φ

3/19

slide-4
SLIDE 4

Dependence between first-order variables

First Order Quantifiers: ∀x1∃y1∀x2∃y2φ

3/19

slide-5
SLIDE 5

Dependence between first-order variables

First Order Quantifiers: ∀x1∃y1∀x2∃y2φ

3/19

slide-6
SLIDE 6

Dependence between first-order variables

First Order Quantifiers: ∀x1∃y1∀x2∃y2φ Henkin Quantifiers (Henkin, 1961): ∀x1 ∃y1 ∀x2 ∃y2

  • φ

3/19

slide-7
SLIDE 7

Dependence between first-order variables

First Order Quantifiers: ∀x1∃y1∀x2∃y2φ Henkin Quantifiers (Henkin, 1961): ∀x1 ∃y1 ∀x2 ∃y2

  • φ

Independence Friendly Logic (Hintikka, Sandu, 1989): ∀x1∃y1∀x2∃y2/{x1}φ

3/19

slide-8
SLIDE 8

Dependence between first-order variables

First Order Quantifiers: ∀x1∃y1∀x2∃y2φ Henkin Quantifiers (Henkin, 1961): ∀x1 ∃y1 ∀x2 ∃y2

  • φ

Independence Friendly Logic (Hintikka, Sandu, 1989): ∀x1∃y1∀x2∃y2/{x1}φ First-order dependence Logic (Väänänen 2007): ∀x1∃y1∀x2∃y2( =(x2, y2) ∧ φ)

3/19

slide-9
SLIDE 9

Dependence between first-order variables

First Order Quantifiers: ∀x1∃y1∀x2∃y2φ Henkin Quantifiers (Henkin, 1961): ∀x1 ∃y1 ∀x2 ∃y2

  • φ

Independence Friendly Logic (Hintikka, Sandu, 1989): ∀x1∃y1∀x2∃y2/{x1}φ First-order dependence Logic (Väänänen 2007): ∀x1∃y1∀x2∃y2( =(x2, y2) ∧ φ) Theorem (Enderton, Walkoe). Over sentences, all of the above extensions of first-order logic are equivalent to Σ1

1.

3/19

slide-10
SLIDE 10

Propositional dependence logic = propositional logic + =( p, q)

I will be absent depending on whether he shows up or not. Whether it rains depends completely on whether it is summer or not. summer

rainy windy v1 1

4/19

slide-11
SLIDE 11

Propositional dependence logic = propositional logic + =( p, q)

I will be absent depending on whether he shows up or not. Whether it rains depends completely on whether it is summer or not. summer

rainy windy v1 1 =(p, q)

4/19

slide-12
SLIDE 12

Propositional dependence logic = propositional logic + =( p, q)

I will be absent depending on whether he shows up or not. Whether it rains depends completely on whether it is summer or not. summer

rainy windy v1 1 =(p, q)

4/19

slide-13
SLIDE 13

Propositional dependence logic = propositional logic + =( p, q)

I will be absent depending on whether he shows up or not. Whether it rains depends completely on whether it is summer or not. summer

rainy windy v1 1 =(p, q)

4/19

slide-14
SLIDE 14

Propositional dependence logic = propositional logic + =( p, q)

I will be absent depending on whether he shows up or not. Whether it rains depends completely on whether it is summer or not.

Team semantics (Hodges 1997)

summer

rainy windy v1 1 =(p, q)

4/19

slide-15
SLIDE 15

Propositional dependence logic = propositional logic + =( p, q)

I will be absent depending on whether he shows up or not. Whether it rains depends completely on whether it is summer or not.

Team semantics (Hodges 1997)

summer

rainy windy v1 1 v1 | = =(p, q)?

4/19

slide-16
SLIDE 16

Propositional dependence logic = propositional logic + =( p, q)

I will be absent depending on whether he shows up or not. Whether it rains depends completely on whether it is summer or not.

Team semantics (Hodges 1997)

summer

rainy windy v1 1 v2 1 1 v3 1 1 v4 1 v1 | = =(p, q)?

4/19

slide-17
SLIDE 17

Propositional dependence logic = propositional logic + =( p, q)

I will be absent depending on whether he shows up or not. Whether it rains depends completely on whether it is summer or not.

Team semantics (Hodges 1997)

X{

summer

rainy windy v1 1 v2 1 1 v3 1 1 v4 1 =(p, q)

4/19

slide-18
SLIDE 18

Propositional dependence logic = propositional logic + =( p, q)

I will be absent depending on whether he shows up or not. Whether it rains depends completely on whether it is summer or not.

Team semantics (Hodges 1997)

X A team {

summer

rainy windy v1 1 v2 1 1 v3 1 1 v4 1 X | = =(p, q)

4/19

slide-19
SLIDE 19

Propositional dependence logic = propositional logic + =( p, q)

I will be absent depending on whether he shows up or not. Whether it rains depends completely on whether it is summer or not.

Team semantics (Hodges 1997)

X A team {

summer

rainy windy v1 1 v2 1 1 v3 1 1 v4 1 X | = =(p, q)

4/19

slide-20
SLIDE 20

Propositional dependence logic = propositional logic + =( p, q)

I will be absent depending on whether he shows up or not. Whether it rains depends completely on whether it is summer or not.

Team semantics (Hodges 1997)

X A team {

summer

rainy windy v1 1 v2 1 1 v3 1 1 v4 1 X | = =(p, q)

4/19

slide-21
SLIDE 21

Propositional dependence logic = propositional logic + =( p, q)

I will be absent depending on whether he shows up or not. Whether it rains depends completely on whether it is summer or not.

Team semantics (Hodges 1997)

X A team {

summer

rainy windy v1 1 v2 1 1 v3 1 1 v4 1 X | = =(p, q)

4/19

slide-22
SLIDE 22

Propositional dependence logic = propositional logic + =( p, q)

I will be absent depending on whether he shows up or not. Whether it rains depends completely on whether it is summer or not.

Team semantics (Hodges 1997)

X A team {

summer

rainy windy v1 1 v2 1 1 v3 1 1 v4 1 X | = =(p, q)

4/19

slide-23
SLIDE 23

Propositional dependence logic = propositional logic + =( p, q)

I will be absent depending on whether he shows up or not. Whether it rains depends completely on whether it is summer or not.

Team semantics (Hodges 1997)

X A team {

summer

rainy windy v1 1 v2 1 1 v3 1 1 v4 1 v5 1 1 1

Y

{

X | = =(p, q), Y | = =(p, q)

4/19

slide-24
SLIDE 24

Propositional dependence logic and its variants

Well-formed formulas of propositional dependence logic (PD) are given by the following grammar φ ::= pi | ¬pi | =(pi1, . . . , pin−1, pin) | φ ∧ φ | φ ⊗ φ Well-formed formulas of propositional intuitionistic dependence logic (PID) are given by the following grammar: φ ::= pi | ⊥ | =(pi1, . . . , pin−1, pin) | φ ∧ φ | φ ∨ φ | φ → φ (¬φ := φ → ⊥) PD∨ is the logic extended from PD by adding the connective ∨.

p0 p1 p2 . . . v1 1 . . .

5/19

slide-25
SLIDE 25

Propositional dependence logic and its variants

Well-formed formulas of propositional dependence logic (PD) are given by the following grammar φ ::= pi | ¬pi | =(pi1, . . . , pin−1, pin) | φ ∧ φ | φ ⊗ φ Well-formed formulas of propositional intuitionistic dependence logic (PID) are given by the following grammar: φ ::= pi | ⊥ | =(pi1, . . . , pin−1, pin) | φ ∧ φ | φ ∨ φ | φ → φ (¬φ := φ → ⊥) PD∨ is the logic extended from PD by adding the connective ∨.

p0 p1 p2 . . . v1 1 . . .

5/19

slide-26
SLIDE 26

Propositional dependence logic and its variants

Well-formed formulas of propositional dependence logic (PD) are given by the following grammar φ ::= pi | ¬pi | =(pi1, . . . , pin−1, pin) | φ ∧ φ | φ ⊗ φ Well-formed formulas of propositional intuitionistic dependence logic (PID) are given by the following grammar: φ ::= pi | ⊥ | =(pi1, . . . , pin−1, pin) | φ ∧ φ | φ ∨ φ | φ → φ (¬φ := φ → ⊥) PD∨ is the logic extended from PD by adding the connective ∨.

p0 p1 p2 . . . v1 1 . . .

5/19

slide-27
SLIDE 27

Propositional dependence logic and its variants

Well-formed formulas of propositional dependence logic (PD) are given by the following grammar φ ::= pi | ¬pi | =(pi1, . . . , pin−1, pin) | φ ∧ φ | φ ⊗ φ Well-formed formulas of propositional intuitionistic dependence logic (PID) are given by the following grammar: φ ::= pi | ⊥ | =(pi1, . . . , pin−1, pin) | φ ∧ φ | φ ∨ φ | φ → φ (¬φ := φ → ⊥) PD∨ is the logic extended from PD by adding the connective ∨.

p0 p1 p2 . . . v1 1 . . .

5/19

slide-28
SLIDE 28

Propositional dependence logic and its variants

Well-formed formulas of propositional dependence logic (PD) are given by the following grammar φ ::= pi | ¬pi | =(pi1, . . . , pin−1, pin) | φ ∧ φ | φ ⊗ φ Well-formed formulas of propositional intuitionistic dependence logic (PID) are given by the following grammar: φ ::= pi | ⊥ | =(pi1, . . . , pin−1, pin) | φ ∧ φ | φ ∨ φ | φ → φ (¬φ := φ → ⊥) PD∨ is the logic extended from PD by adding the connective ∨. Team semantics: A valuation is a function v : N → {0, 1}.

p0 p1 p2 . . . v1 1 . . .

5/19

slide-29
SLIDE 29

Propositional dependence logic and its variants

Well-formed formulas of propositional dependence logic (PD) are given by the following grammar φ ::= pi | ¬pi | =(pi1, . . . , pin−1, pin) | φ ∧ φ | φ ⊗ φ Well-formed formulas of propositional intuitionistic dependence logic (PID) are given by the following grammar: φ ::= pi | ⊥ | =(pi1, . . . , pin−1, pin) | φ ∧ φ | φ ∨ φ | φ → φ (¬φ := φ → ⊥) PD∨ is the logic extended from PD by adding the connective ∨. Team semantics: A valuation is a function v : N → {0, 1}. A team is a set of valuations.

p0 p1 p2 . . . v1 1 . . . v2 1 1 . . . v4 1 . . . . . . . . . . . . . . . . . .

5/19

slide-30
SLIDE 30

Team Semantics

Let X be a team. X | = pi iff for all v ∈ X, v(i) = 1; X | = ¬pi iff for all v ∈ X, v(i) = 0; X | = ⊥ iff X = ∅; X | = =(pi1, . . . , pin) iff for all v, v′ ∈ X

  • v(i1) = v′(i1), . . . , v(in−1) = v′(in−1)
  • =

⇒ v(in) = v′(in); X | = φ ∧ ψ iff X | = φ and X | = ψ; X | = φ⊗ψ iff there exist teams Y, Z ⊆ X with X = Y ∪Z such that Y | = φ and Z | = ψ; p0 p1 p2 v1 1 v2 1 1 v3 1 v4 1 1 X | = φ ∨ ψ iff X | = φ or X | = ψ; X | = φ → ψ iff for any team Y ⊆ X,

6/19

slide-31
SLIDE 31

Team Semantics

Let X be a team. X | = pi iff for all v ∈ X, v(i) = 1; X | = ¬pi iff for all v ∈ X, v(i) = 0; X | = ⊥ iff X = ∅; X | = =(pi1, . . . , pin) iff for all v, v′ ∈ X

  • v(i1) = v′(i1), . . . , v(in−1) = v′(in−1)
  • =

⇒ v(in) = v′(in); X | = φ ∧ ψ iff X | = φ and X | = ψ; X | = φ⊗ψ iff there exist teams Y, Z ⊆ X with X = Y ∪Z such that Y | = φ and Z | = ψ; X { X | = p0 p0 p1 p2 v1 1 v2 1 1 v3 1 v4 1 1 X | = φ ∨ ψ iff X | = φ or X | = ψ; X | = φ → ψ iff for any team Y ⊆ X,

6/19

slide-32
SLIDE 32

Team Semantics

Let X be a team. X | = pi iff for all v ∈ X, v(i) = 1; X | = ¬pi iff for all v ∈ X, v(i) = 0; X | = ⊥ iff X = ∅; X | = =(pi1, . . . , pin) iff for all v, v′ ∈ X

  • v(i1) = v′(i1), . . . , v(in−1) = v′(in−1)
  • =

⇒ v(in) = v′(in); X | = φ ∧ ψ iff X | = φ and X | = ψ; X | = φ⊗ψ iff there exist teams Y, Z ⊆ X with X = Y ∪Z such that Y | = φ and Z | = ψ; X { X | = p0 Y | = ¬p0 Y { p0 p1 p2 v1 1 v2 1 1 v3 1 v4 1 1 X | = φ ∨ ψ iff X | = φ or X | = ψ; X | = φ → ψ iff for any team Y ⊆ X,

6/19

slide-33
SLIDE 33

Team Semantics

Let X be a team. X | = pi iff for all v ∈ X, v(i) = 1; X | = ¬pi iff for all v ∈ X, v(i) = 0; X | = ⊥ iff X = ∅; X | = =(pi1, . . . , pin) iff for all v, v′ ∈ X

  • v(i1) = v′(i1), . . . , v(in−1) = v′(in−1)
  • =

⇒ v(in) = v′(in); X | = φ ∧ ψ iff X | = φ and X | = ψ; X | = φ⊗ψ iff there exist teams Y, Z ⊆ X with X = Y ∪Z such that Y | = φ and Z | = ψ; X { X | = p0 Y | = ¬p0 X ∪ Y | = p0 X ∪ Y | = ¬p0 Y { p0 p1 p2 v1 1 v2 1 1 v3 1 v4 1 1 X | = φ ∨ ψ iff X | = φ or X | = ψ; X | = φ → ψ iff for any team Y ⊆ X,

6/19

slide-34
SLIDE 34

Team Semantics

Let X be a team. X | = pi iff for all v ∈ X, v(i) = 1; X | = ¬pi iff for all v ∈ X, v(i) = 0; X | = ⊥ iff X = ∅; X | = =(pi1, . . . , pin) iff for all v, v′ ∈ X

  • v(i1) = v′(i1), . . . , v(in−1) = v′(in−1)
  • =

⇒ v(in) = v′(in); X | = φ ∧ ψ iff X | = φ and X | = ψ; X | = φ⊗ψ iff there exist teams Y, Z ⊆ X with X = Y ∪Z such that Y | = φ and Z | = ψ; p0 p1 p2 v1 1 v2 1 1 v3 1 v4 1 1 X | = φ ∨ ψ iff X | = φ or X | = ψ; X | = φ → ψ iff for any team Y ⊆ X,

6/19

slide-35
SLIDE 35

Team Semantics

Let X be a team. X | = pi iff for all v ∈ X, v(i) = 1; X | = ¬pi iff for all v ∈ X, v(i) = 0; X | = ⊥ iff X = ∅; X | = =(pi1, . . . , pin) iff for all v, v′ ∈ X

  • v(i1) = v′(i1), . . . , v(in−1) = v′(in−1)
  • =

⇒ v(in) = v′(in); X | = φ ∧ ψ iff X | = φ and X | = ψ; X | = φ⊗ψ iff there exist teams Y, Z ⊆ X with X = Y ∪Z such that Y | = φ and Z | = ψ; X | = =(p0, p1) p0 p1 p2 v1 1 v2 1 1 v3 1 v4 1 1 X | = φ ∨ ψ iff X | = φ or X | = ψ; X | = φ → ψ iff for any team Y ⊆ X,

6/19

slide-36
SLIDE 36

Team Semantics

Let X be a team. X | = pi iff for all v ∈ X, v(i) = 1; X | = ¬pi iff for all v ∈ X, v(i) = 0; X | = ⊥ iff X = ∅; X | = =(pi1, . . . , pin) iff for all v, v′ ∈ X

  • v(i1) = v′(i1), . . . , v(in−1) = v′(in−1)
  • =

⇒ v(in) = v′(in); X | = φ ∧ ψ iff X | = φ and X | = ψ; X | = φ⊗ψ iff there exist teams Y, Z ⊆ X with X = Y ∪Z such that Y | = φ and Z | = ψ; X | = =(p0, p1) p0 p1 p2 v1 1 v2 1 1 v3 1 v4 1 1 X | = φ ∨ ψ iff X | = φ or X | = ψ; X | = φ → ψ iff for any team Y ⊆ X,

6/19

slide-37
SLIDE 37

Team Semantics

Let X be a team. X | = pi iff for all v ∈ X, v(i) = 1; X | = ¬pi iff for all v ∈ X, v(i) = 0; X | = ⊥ iff X = ∅; X | = =(pi1, . . . , pin) iff for all v, v′ ∈ X

  • v(i1) = v′(i1), . . . , v(in−1) = v′(in−1)
  • =

⇒ v(in) = v′(in); X | = φ ∧ ψ iff X | = φ and X | = ψ; X | = φ⊗ψ iff there exist teams Y, Z ⊆ X with X = Y ∪Z such that Y | = φ and Z | = ψ; X | = φ ∨ ψ iff X | = φ or X | = ψ; X | = φ → ψ iff for any team Y ⊆ X, Y | = φ = ⇒ Y | = ψ.

6/19

slide-38
SLIDE 38

Team Semantics

Let X be a team. X | = pi iff for all v ∈ X, v(i) = 1; X | = ¬pi iff for all v ∈ X, v(i) = 0; X | = ⊥ iff X = ∅; X | = =(pi1, . . . , pin) iff for all v, v′ ∈ X

  • v(i1) = v′(i1), . . . , v(in−1) = v′(in−1)
  • =

⇒ v(in) = v′(in); X | = φ ∧ ψ iff X | = φ and X | = ψ; X | = φ⊗ψ iff there exist teams Y, Z ⊆ X with X = Y ∪Z such that Y | = φ and Z | = ψ; X | = φ ∨ ψ iff X | = φ or X | = ψ; X | = φ → ψ iff for any team Y ⊆ X, Y | = φ = ⇒ Y | = ψ.

6/19

slide-39
SLIDE 39

Team Semantics

Let X be a team. X | = pi iff for all v ∈ X, v(i) = 1; X | = ¬pi iff for all v ∈ X, v(i) = 0; X | = ⊥ iff X = ∅; X | = =(pi1, . . . , pin) iff for all v, v′ ∈ X

  • v(i1) = v′(i1), . . . , v(in−1) = v′(in−1)
  • =

⇒ v(in) = v′(in); X | = φ ∧ ψ iff X | = φ and X | = ψ; X | = φ⊗ψ iff there exist teams Y, Z ⊆ X with X = Y ∪Z such that Y | = φ and Z | = ψ; X | = φ ∨ ψ iff X | = φ or X | = ψ; X | = φ → ψ iff for any team Y ⊆ X, Y | = φ = ⇒ Y | = ψ.

6/19

slide-40
SLIDE 40

Team Semantics

Let X be a team. X | = pi iff for all v ∈ X, v(i) = 1; X | = ¬pi iff for all v ∈ X, v(i) = 0; X | = ⊥ iff X = ∅; X | = =(pi1, . . . , pin) iff for all v, v′ ∈ X

  • v(i1) = v′(i1), . . . , v(in−1) = v′(in−1)
  • =

⇒ v(in) = v′(in); X | = φ ∧ ψ iff X | = φ and X | = ψ; X | = φ⊗ψ iff there exist teams Y, Z ⊆ X with X = Y ∪Z such that Y | = φ and Z | = ψ; X | = φ ∨ ψ iff X | = φ or X | = ψ; X | = φ → ψ iff for any team Y ⊆ X, Y | = φ = ⇒ Y | = ψ. The logics PD, PD∨, PID are downwards closed, that is, X | = φ and Y ⊆ X = ⇒ Y | = φ; and have the empty team property, that is, ∅ | = φ, for all φ.

6/19

slide-41
SLIDE 41

Team Semantics

Let X be a team. X | = pi iff for all v ∈ X, v(i) = 1; X | = ¬pi iff for all v ∈ X, v(i) = 0; X | = ⊥ iff X = ∅; X | = =(pi1, . . . , pin) iff for all v, v′ ∈ X

  • v(i1) = v′(i1), . . . , v(in−1) = v′(in−1)
  • =

⇒ v(in) = v′(in); X | = φ ∧ ψ iff X | = φ and X | = ψ; X | = φ⊗ψ iff there exist teams Y, Z ⊆ X with X = Y ∪Z such that Y | = φ and Z | = ψ; X | = φ ∨ ψ iff X | = φ or X | = ψ; X | = φ → ψ iff for any team Y ⊆ X, Y | = φ = ⇒ Y | = ψ. The logics PD, PD∨, PID are downwards closed, that is, X | = φ and Y ⊆ X = ⇒ Y | = φ; and have the empty team property, that is, ∅ | = φ, for all φ.

6/19

slide-42
SLIDE 42

Team Semantics

Let X be a team. X | = pi iff for all v ∈ X, v(i) = 1; X | = ¬pi iff for all v ∈ X, v(i) = 0; X | = ⊥ iff X = ∅; X | = =(pi1, . . . , pin) iff for all v, v′ ∈ X

  • v(i1) = v′(i1), . . . , v(in−1) = v′(in−1)
  • =

⇒ v(in) = v′(in); X | = φ ∧ ψ iff X | = φ and X | = ψ; X | = φ⊗ψ iff there exist teams Y, Z ⊆ X with X = Y ∪Z such that Y | = φ and Z | = ψ; X | = φ ∨ ψ iff X | = φ or X | = ψ; X | = φ → ψ iff for any team Y ⊆ X, Y | = φ = ⇒ Y | = ψ. Intuitionistic disjunction ∨ has the disjunction property: | = φ ∨ ψ = ⇒ | = φ or | = ψ.

7/19

slide-43
SLIDE 43

Dependence atoms are definable in the fragment of PID and PD∨ without dependence atoms: =(p0, p1) ≡

  • p0 ∧
  • p1 ∨ ¬p1
  • ¬p0 ∧
  • p1 ∨ ¬p1
  • ≡ (p0 ∨ ¬p0) → (p1 ∨ ¬p1)

p0 p1 p2 v1 1 v2 1 1 v3 1 v4 1 1

8/19

slide-44
SLIDE 44

Dependence atoms are definable in the fragment of PID and PD∨ without dependence atoms: =(p0, p1) ≡

  • p0 ∧
  • p1 ∨ ¬p1
  • ¬p0 ∧
  • p1 ∨ ¬p1
  • ≡ (p0 ∨ ¬p0) → (p1 ∨ ¬p1)

p0 p1 p2 v1 1 v2 1 1 v3 1 v4 1 1 Observation (de Jongh, Litak) PID− is equivalent to inquisitive logic (Ciardelli, Roelofsen 2011).

8/19

slide-45
SLIDE 45

Fix N = {i1, . . . , in} ⊆ N. An n-valuation on N is a function s : N → {0, 1}. An n-team on N is a set of n-valuations on N.

01 00 11 10

There are in total 2n distinct n-valuations, and 22n n-teams, among which there exists a biggest team (denoted by 2n) consisting of all n-valuations on N.

9/19

slide-46
SLIDE 46

Fix N = {i1, . . . , in} ⊆ N. An n-valuation on N is a function s : N → {0, 1}. An n-team on N is a set of n-valuations on N.

01 00 11 10

There are in total 2n distinct n-valuations, and 22n n-teams, among which there exists a biggest team (denoted by 2n) consisting of all n-valuations on N.

9/19

slide-47
SLIDE 47

Fix N = {i1, . . . , in} ⊆ N. An n-valuation on N is a function s : N → {0, 1}. An n-team on N is a set of n-valuations on N.

01 00 11 10

There are in total 2n distinct n-valuations, and 22n n-teams, among which there exists a biggest team (denoted by 2n) consisting of all n-valuations on N.

9/19

slide-48
SLIDE 48

Fix N = {i1, . . . , in} ⊆ N. An n-valuation on N is a function s : N → {0, 1}. An n-team on N is a set of n-valuations on N.

01 00 11 10

There are in total 2n distinct n-valuations, and 22n n-teams, among which there exists a biggest team (denoted by 2n) consisting of all n-valuations on N.

9/19

slide-49
SLIDE 49

Fix N = {i1, . . . , in} ⊆ N. An n-valuation on N is a function s : N → {0, 1}. An n-team on N is a set of n-valuations on N.

01 00 11 10

There are in total 2n distinct n-valuations, and 22n n-teams, among which there exists a biggest team (denoted by 2n) consisting of all n-valuations on N.

9/19

slide-50
SLIDE 50

Fix N = {i1, . . . , in} ⊆ N. An n-valuation on N is a function s : N → {0, 1}. An n-team on N is a set of n-valuations on N.

01 00 11 10

There are in total 2n distinct n-valuations, and 22n n-teams, among which there exists a biggest team (denoted by 2n) consisting of all n-valuations on N.

9/19

slide-51
SLIDE 51

Fix N = {i1, . . . , in} ⊆ N. An n-valuation on N is a function s : N → {0, 1}. An n-team on N is a set of n-valuations on N.

01 00 11 10

There are in total 2n distinct n-valuations, and 22n n-teams, among which there exists a biggest team (denoted by 2n) consisting of all n-valuations on N.

9/19

slide-52
SLIDE 52

Fix N = {i1, . . . , in} ⊆ N. An n-valuation on N is a function s : N → {0, 1}. An n-team on N is a set of n-valuations on N.

01 00 11 10

There are in total 2n distinct n-valuations, and 22n n-teams, among which there exists a biggest team (denoted by 2n) consisting of all n-valuations on N.

9/19

slide-53
SLIDE 53

Fix N = {i1, . . . , in} ⊆ N. An n-valuation on N is a function s : N → {0, 1}. An n-team on N is a set of n-valuations on N.

01 00 11 10

There are in total 2n distinct n-valuations, and 22n n-teams, among which there exists a biggest team (denoted by 2n) consisting of all n-valuations on N.

9/19

slide-54
SLIDE 54

Fix N = {i1, . . . , in} ⊆ N. An n-valuation on N is a function s : N → {0, 1}. An n-team on N is a set of n-valuations on N.

01 00 11 10

There are in total 2n distinct n-valuations, and 22n n-teams, among which there exists a biggest team (denoted by 2n) consisting of all n-valuations on N.

9/19

slide-55
SLIDE 55

Fix N = {i1, . . . , in} ⊆ N. An n-valuation on N is a function s : N → {0, 1}. An n-team on N is a set of n-valuations on N.

01 00 11 10

There are in total 2n distinct n-valuations, and 22n n-teams, among which there exists a biggest team (denoted by 2n) consisting of all n-valuations on N.

9/19

slide-56
SLIDE 56

Fix N = {i1, . . . , in} ⊆ N. An n-valuation on N is a function s : N → {0, 1}. An n-team on N is a set of n-valuations on N.

01 00 11 10

There are in total 2n distinct n-valuations, and 22n n-teams, among which there exists a biggest team (denoted by 2n) consisting of all n-valuations on N.

9/19

slide-57
SLIDE 57

℘(2n)

00 01 10 11 00 10 11 01 10 11 00 01 11 00 01 10 01 10 01 11 10 11 11 00 10 00 01 00 10 11 01 00

10/19

slide-58
SLIDE 58

(℘(2n), ⊇)

00 01 10 11 00 10 11 01 10 11 00 01 11 00 01 10 01 10 01 11 10 11 11 00 10 00 01 00 10 11 01 00

10/19

slide-59
SLIDE 59

(℘(2n) \ {∅}, ⊇)

00 01 10 11 00 10 11 01 10 11 00 01 11 00 01 10 01 10 01 11 10 11 11 00 10 00 01 00 10 11 01 00

10/19

slide-60
SLIDE 60

A Medvedev frame: (℘(2n) \ {∅}, ⊇)

00 01 10 11 00 10 11 01 10 11 00 01 11 00 01 10 01 10 01 11 10 11 11 00 10 00 01 00 10 11 01 00

10/19

slide-61
SLIDE 61

A Medvedev frame: (℘(2n) \ {∅}, ⊇)

00 01 10 11 00 10 11 01 10 11 00 01 11 00 01 10 01 10 01 11 10 11 11 00 10 00 01 00 10 11 01 00

p, q

10/19

slide-62
SLIDE 62

A Medvedev frame: (℘(2n) \ {∅}, ⊇)

00 01 10 11 00 10 11 01 10 11 00 01 11 00 01 10 01 10 01 11 10 11 11 00 10 00 01 00 10 11 01 00

p → q p, q

10/19

slide-63
SLIDE 63

A Medvedev frame: (℘(2n) \ {∅}, ⊇)

00 01 10 11 00 10 11 01 10 11 00 01 11 00 01 10 01 10 01 11 10 11 11 00 10 00 01 00 10 11 01 00

p → q ¬¬p → p p, q

10/19

slide-64
SLIDE 64

A Medvedev frame: (℘(2n) \ {∅}, ⊇)

00 01 10 11 00 10 11 01 10 11 00 01 11 00 01 10 01 10 01 11 10 11 11 00 10 00 01 00 10 11 01 00

p → q ¬¬p → p p, q

[Ciardelli, Roelofsen 2011]: PID− = ML¬ = {φ | τ(φ) ∈ ML, where τ(p) = ¬p}

10/19

slide-65
SLIDE 65

A Medvedev frame: (℘(2n) \ {∅}, ⊇)

00 01 10 11 00 10 11 01 10 11 00 01 11 00 01 10 01 10 01 11 10 11 11 00 10 00 01 00 10 11 01 00

p → q ¬¬p → p p, q

[Ciardelli, Roelofsen 2011]: PID− = ML¬ = {φ | τ(φ) ∈ ML, where τ(p) = ¬p} = KP¬ = KP ⊕ ¬¬p → p

10/19

slide-66
SLIDE 66

Fix N = {i1, . . . , in}. Put φ(pi1, . . . , pin) := {X ⊆ 2n | X | = φ}, ∇N := {K ⊆ 22n | ∅ ∈ K,

  • X ∈ K, Y ⊆ X =

⇒ Y ∈ K

  • }.

For each K ∈ ∇N, consider

X∈K ΘX. For any n-team Y on N,

Y | =

  • ΘX ⇐

⇒ ∃X ∈ K(Y ⊆ X) ⇐ ⇒ Y ∈ K.

11/19

slide-67
SLIDE 67

Fix N = {i1, . . . , in}. Put φ(pi1, . . . , pin) := {X ⊆ 2n | X | = φ}, ∇N := {K ⊆ 22n | ∅ ∈ K,

  • X ∈ K, Y ⊆ X =

⇒ Y ∈ K

  • }.

For each K ∈ ∇N, consider

X∈K ΘX. For any n-team Y on N,

Y | =

  • ΘX ⇐

⇒ ∃X ∈ K(Y ⊆ X) ⇐ ⇒ Y ∈ K.

11/19

slide-68
SLIDE 68

Fix N = {i1, . . . , in}. Put φ(pi1, . . . , pin) := {X ⊆ 2n | X | = φ}, ∇N := {K ⊆ 22n | ∅ ∈ K,

  • X ∈ K, Y ⊆ X =

⇒ Y ∈ K

  • }.

Theorem (Ciardelli, Huuskonen, Y.) PID, PD∨, PD are maximal downwards closed logics, i.e., for L ∈ {PID, PD∨, PD}, ∇N = {φ | φ(pi1, . . . , pin) is an n-formula of L}. In particular, PID ≡ PD∨ ≡ PD. For each K ∈ ∇N, consider

X∈K ΘX. For any n-team Y on N,

Y | =

  • ΘX ⇐

⇒ ∃X ∈ K(Y ⊆ X) ⇐ ⇒ Y ∈ K.

11/19

slide-69
SLIDE 69

Fix N = {i1, . . . , in}. Put φ(pi1, . . . , pin) := {X ⊆ 2n | X | = φ}, ∇N := {K ⊆ 22n | ∅ ∈ K,

  • X ∈ K, Y ⊆ X =

⇒ Y ∈ K

  • }.

Theorem (Ciardelli, Huuskonen, Y.) PID, PD∨, PD are maximal downwards closed logics, i.e., for L ∈ {PID, PD∨, PD}, ∇N = {φ | φ(pi1, . . . , pin) is an n-formula of L}. In particular, PID ≡ PD∨ ≡ PD.

Theorem (Y.)

Every instance of ∨ and → is definable in PD, but ∨ and → are not uniformly definable in PD.

11/19

slide-70
SLIDE 70

Fix N = {i1, . . . , in}. Put φ(pi1, . . . , pin) := {X ⊆ 2n | X | = φ}, ∇N := {K ⊆ 22n | ∅ ∈ K,

  • X ∈ K, Y ⊆ X =

⇒ Y ∈ K

  • }.

Theorem (Ciardelli, Huuskonen, Y.) PID, PD∨, PD are maximal downwards closed logics, i.e., for L ∈ {PID, PD∨, PD}, ∇N = {φ | φ(pi1, . . . , pin) is an n-formula of L}. In particular, PID ≡ PD∨ ≡ PD.

  • Proof. We only treat PID and PD∨.

For each K ∈ ∇N, consider

X∈K ΘX. For any n-team Y on N,

Y | =

  • ΘX ⇐

⇒ ∃X ∈ K(Y ⊆ X) ⇐ ⇒ Y ∈ K.

11/19

slide-71
SLIDE 71

Fix N = {i1, . . . , in}. Put φ(pi1, . . . , pin) := {X ⊆ 2n | X | = φ}, ∇N := {K ⊆ 22n | ∅ ∈ K,

  • X ∈ K, Y ⊆ X =

⇒ Y ∈ K

  • }.

Theorem (Ciardelli, Huuskonen, Y.) PID, PD∨, PD are maximal downwards closed logics, i.e., for L ∈ {PID, PD∨, PD}, ∇N = {φ | φ(pi1, . . . , pin) is an n-formula of L}. In particular, PID ≡ PD∨ ≡ PD.

  • Proof. We only treat PID and PD∨. First, consider an n-team:

X{ p1 p2 v1 1 1 v2 1 v3 1

For each K ∈ ∇N, consider

X∈K ΘX. For any n-team Y on N,

Y | =

  • ΘX ⇐

⇒ ∃X ∈ K(Y ⊆ X) ⇐ ⇒ Y ∈ K.

11/19

slide-72
SLIDE 72

Fix N = {i1, . . . , in}. Put φ(pi1, . . . , pin) := {X ⊆ 2n | X | = φ}, ∇N := {K ⊆ 22n | ∅ ∈ K,

  • X ∈ K, Y ⊆ X =

⇒ Y ∈ K

  • }.

Theorem (Ciardelli, Huuskonen, Y.) PID, PD∨, PD are maximal downwards closed logics, i.e., for L ∈ {PID, PD∨, PD}, ∇N = {φ | φ(pi1, . . . , pin) is an n-formula of L}. In particular, PID ≡ PD∨ ≡ PD.

  • Proof. We only treat PID and PD∨. First, consider an n-team:

X{ p1 p2 v1 1 1 v2 1 v3 1 Let ΘX :=        for PD; for PID. Then Y | = ΘX ⇐ ⇒ Y ⊆ X, for any n-team Y.

For each K ∈ ∇N, consider

X∈K ΘX. For any n-team Y on N,

Y | =

  • ΘX ⇐

⇒ ∃X ∈ K(Y ⊆ X) ⇐ ⇒ Y ∈ K.

11/19

slide-73
SLIDE 73

Fix N = {i1, . . . , in}. Put φ(pi1, . . . , pin) := {X ⊆ 2n | X | = φ}, ∇N := {K ⊆ 22n | ∅ ∈ K,

  • X ∈ K, Y ⊆ X =

⇒ Y ∈ K

  • }.

Theorem (Ciardelli, Huuskonen, Y.) PID, PD∨, PD are maximal downwards closed logics, i.e., for L ∈ {PID, PD∨, PD}, ∇N = {φ | φ(pi1, . . . , pin) is an n-formula of L}. In particular, PID ≡ PD∨ ≡ PD.

  • Proof. We only treat PID and PD∨. First, consider an n-team:

X{ p1 p2 v1 1 1 v2 1 v3 1 Let ΘX :=       

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PD; for PID. Then Y | = ΘX ⇐ ⇒ Y ⊆ X, for any n-team Y.

For each K ∈ ∇N, consider

X∈K ΘX. For any n-team Y on N,

Y | =

  • ΘX ⇐

⇒ ∃X ∈ K(Y ⊆ X) ⇐ ⇒ Y ∈ K.

11/19

slide-74
SLIDE 74

Fix N = {i1, . . . , in}. Put φ(pi1, . . . , pin) := {X ⊆ 2n | X | = φ}, ∇N := {K ⊆ 22n | ∅ ∈ K,

  • X ∈ K, Y ⊆ X =

⇒ Y ∈ K

  • }.

Theorem (Ciardelli, Huuskonen, Y.) PID, PD∨, PD are maximal downwards closed logics, i.e., for L ∈ {PID, PD∨, PD}, ∇N = {φ | φ(pi1, . . . , pin) is an n-formula of L}. In particular, PID ≡ PD∨ ≡ PD.

  • Proof. We only treat PID and PD∨. First, consider an n-team:

X{ p1 p2 v1 1 1 v2 1 v3 1 Let ΘX :=       

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PD; ¬¬

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PID. Then Y | = ΘX ⇐ ⇒ Y ⊆ X, for any n-team Y.

For each K ∈ ∇N, consider

X∈K ΘX. For any n-team Y on N,

Y | =

  • ΘX ⇐

⇒ ∃X ∈ K(Y ⊆ X) ⇐ ⇒ Y ∈ K.

11/19

slide-75
SLIDE 75

Fix N = {i1, . . . , in}. Put φ(pi1, . . . , pin) := {X ⊆ 2n | X | = φ}, ∇N := {K ⊆ 22n | ∅ ∈ K,

  • X ∈ K, Y ⊆ X =

⇒ Y ∈ K

  • }.

Theorem (Ciardelli, Huuskonen, Y.) PID, PD∨, PD are maximal downwards closed logics, i.e., for L ∈ {PID, PD∨, PD}, ∇N = {φ | φ(pi1, . . . , pin) is an n-formula of L}. In particular, PID ≡ PD∨ ≡ PD.

  • Proof. We only treat PID and PD∨. First, consider an n-team:

X{ p1 p2 v1 1 1 v2 1 v3 1 Let ΘX :=       

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PD; ¬¬

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PID. Then Y | = ΘX ⇐ ⇒ Y ⊆ X, for any n-team Y.

For each K ∈ ∇N, consider

X∈K ΘX. For any n-team Y on N,

Y | =

  • ΘX ⇐

⇒ ∃X ∈ K(Y ⊆ X) ⇐ ⇒ Y ∈ K.

11/19

slide-76
SLIDE 76

Theorem (Ciardelli, Huuskonen, Y.) PID, PD∨, PD are maximal downwards closed logics, i.e., for L ∈ {PID, PD∨, PD}, ∇N = {φ | φ(pi1, . . . , pin) is an n-formula of L}. In particular, PID ≡ PD∨ ≡ PD.

  • Proof. We only treat PID and PD∨. First, consider an n-team:

X{ p1 p2 v1 1 1 v2 1 v3 1 Let ΘX :=       

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PD; ¬¬

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PID. Then Y | = ΘX ⇐ ⇒ Y ⊆ X, for any n-team Y.

For each K ∈ ∇N, consider

X∈K ΘX. For any n-team Y on N,

Y | =

  • X∈K

ΘX ⇐ ⇒ ∃X ∈ K(Y ⊆ X) ⇐ ⇒ Y ∈ K. Hence

X∈K ΘX = K.

11/19

slide-77
SLIDE 77
  • Theorem. PID, PD∨, PD are sound and complete w.r.t. their deductive

systems.

  • Corrolary. For every formula φ of PID and PD∨, φ ⊣⊢

i∈I ΘXi.

A Hilbert Style deductive system for PID− (Ciardelli, Roelofsen 2011) Axioms: all substitution instances of IPC axioms all substitution instances of (KP) (¬pi → (pj ∨ pk)) → ((¬pi → pj) ∨ (¬pi → pk)). ¬¬pi → pi for all propositional variables pi Rules: Modus Ponens Natural deduction systems for PD and PD∨ (Väänänen, Y.) For L ∈ {PD, PD∨}, if φ does not contain any ∨ or dependence atoms, then ⊢CPC φ ⇐ ⇒ ⊢L φ.

12/19

slide-78
SLIDE 78
  • Theorem. PID, PD∨, PD are sound and complete w.r.t. their deductive

systems.

  • Corrolary. For every formula φ of PID and PD∨, φ ⊣⊢

i∈I ΘXi.

A Hilbert Style deductive system for PID− (Ciardelli, Roelofsen 2011) Axioms: all substitution instances of IPC axioms all substitution instances of (KP) (¬pi → (pj ∨ pk)) → ((¬pi → pj) ∨ (¬pi → pk)). ¬¬pi → pi for all propositional variables pi Rules: Modus Ponens Natural deduction systems for PD and PD∨ (Väänänen, Y.) For L ∈ {PD, PD∨}, if φ does not contain any ∨ or dependence atoms, then ⊢CPC φ ⇐ ⇒ ⊢L φ.

12/19

slide-79
SLIDE 79
  • Theorem. PID, PD∨, PD are sound and complete w.r.t. their deductive

systems.

  • Corrolary. For every formula φ of PID and PD∨, φ ⊣⊢

i∈I ΘXi.

A Hilbert Style deductive system for PID− (Ciardelli, Roelofsen 2011) Axioms: all substitution instances of IPC axioms all substitution instances of (KP) (¬pi → (pj ∨ pk)) → ((¬pi → pj) ∨ (¬pi → pk)). ¬¬pi → pi for all propositional variables pi Rules: Modus Ponens Natural deduction systems for PD and PD∨ (Väänänen, Y.) For L ∈ {PD, PD∨}, if φ does not contain any ∨ or dependence atoms, then ⊢CPC φ ⇐ ⇒ ⊢L φ.

12/19

slide-80
SLIDE 80

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ.

13/19

slide-81
SLIDE 81

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ.

00 01 11 01 11 11 00 01 00 11 01 00

13/19

slide-82
SLIDE 82

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ.

00 01 11 01 11 11 00 01 00 11 01 00

φ

13/19

slide-83
SLIDE 83

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ.

00 01 11 01 11 11 00 01 00 11 01 00

φ φ φ φ

13/19

slide-84
SLIDE 84

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ.

00 01 11 01 11 11 00 01 00 11 01 00

φ φ φ φ

Fact: Formulas with no occurrences of dependence atoms or intuitionistic disjunction ∨ are flat.

13/19

slide-85
SLIDE 85

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ.

00 01 11 01 11 11 00 01 00 11 01 00

φ φ φ φ

Fact: Formulas with no occurrences of dependence atoms or intuitionistic disjunction ∨ are flat. E.g. ΘX =       

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PD; ¬¬

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PID. is flat

13/19

slide-86
SLIDE 86

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ.

00 01 11 01 11 11 00 01 00 11 01 00

φ φ φ φ

Fact: Formulas with no occurrences of dependence atoms or intuitionistic disjunction ∨ are flat.

  • Lemma. In PID, a formula φ is flat if ⊢ φ ↔ ¬¬φ.

E.g. ΘX =       

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PD; ¬¬

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PID. is flat

13/19

slide-87
SLIDE 87

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ.

Lemma

A formula φ is flat iff φ ⊣⊢ ΘX for some X.

14/19

slide-88
SLIDE 88

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ.

Lemma

A formula φ is flat iff φ ⊣⊢ ΘX for some X.

  • Proof. “=

⇒”: Suppose φ is flat. Note that φ ⊣⊢ ΘX1 ∨ · · · ∨ ΘXk,

14/19

slide-89
SLIDE 89

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ. Fact: Y | = ΘX ⇐ ⇒ Y ⊆ X, for any n-team Y.

Lemma

A formula φ is flat iff φ ⊣⊢ ΘX for some X.

  • Proof. “=

⇒”: Suppose φ is flat. Note that φ ⊣⊢ ΘX1 ∨ · · · ∨ ΘXk, where w.l.o.g. we assume that Xi’s are maximal.

X1 X2 X3 Xk · · · · · ·

14/19

slide-90
SLIDE 90

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ. Fact: Y | = ΘX ⇐ ⇒ Y ⊆ X, for any n-team Y.

Lemma

A formula φ is flat iff φ ⊣⊢ ΘX for some X.

  • Proof. “=

⇒”: Suppose φ is flat. Note that φ ⊣⊢ ΘX1 ∨ · · · ∨ ΘXk, where w.l.o.g. we assume that Xi’s are maximal. If k > 1,

X1 X2 X3 Xk · · · · · ·

14/19

slide-91
SLIDE 91

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ. Fact: Y | = ΘX ⇐ ⇒ Y ⊆ X, for any n-team Y.

Lemma

A formula φ is flat iff φ ⊣⊢ ΘX for some X.

  • Proof. “=

⇒”: Suppose φ is flat. Note that φ ⊣⊢ ΘX1 ∨ · · · ∨ ΘXk, where w.l.o.g. we assume that Xi’s are maximal. If k > 1, then pick

X1 X2 X3 Xk · · · · · · v1

14/19

slide-92
SLIDE 92

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ. Fact: Y | = ΘX ⇐ ⇒ Y ⊆ X, for any n-team Y.

Lemma

A formula φ is flat iff φ ⊣⊢ ΘX for some X.

  • Proof. “=

⇒”: Suppose φ is flat. Note that φ ⊣⊢ ΘX1 ∨ · · · ∨ ΘXk, where w.l.o.g. we assume that Xi’s are maximal. If k > 1, then pick

X1 X2 X3 Xk · · · · · · v1 v2

14/19

slide-93
SLIDE 93

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ. Fact: Y | = ΘX ⇐ ⇒ Y ⊆ X, for any n-team Y.

Lemma

A formula φ is flat iff φ ⊣⊢ ΘX for some X.

  • Proof. “=

⇒”: Suppose φ is flat. Note that φ ⊣⊢ ΘX1 ∨ · · · ∨ ΘXk, where w.l.o.g. we assume that Xi’s are maximal. If k > 1, then pick

X1 X2 X3 Xk · · · · · · v1 v2 vk

14/19

slide-94
SLIDE 94

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ. Fact: Y | = ΘX ⇐ ⇒ Y ⊆ X, for any n-team Y.

Lemma

A formula φ is flat iff φ ⊣⊢ ΘX for some X.

  • Proof. “=

⇒”: Suppose φ is flat. Note that φ ⊣⊢ ΘX1 ∨ · · · ∨ ΘXk, where w.l.o.g. we assume that Xi’s are maximal. If k > 1, then pick

X1 X2 X3 Xk · · · · · · v1 v2 vk

{vi} ⊆ Xi and {v1, . . . , vk} Xi for all 1 ≤ i ≤ k = ⇒ {vi} | = ΘXi and {v1, . . . , vk} | = ΘXi for all 1 ≤ i ≤ k = ⇒ {vi} | = φ for all 1 ≤ i ≤ k whereas {v1, . . . , vk} | = φ

14/19

slide-95
SLIDE 95

Definition

A formula φ is said to be flat iff for all teams X, X | = φ ⇐ ⇒ ∀v ∈ X, {v} | = φ. Fact: Y | = ΘX ⇐ ⇒ Y ⊆ X, for any n-team Y.

Lemma

A formula φ is flat iff φ ⊣⊢ ΘX for some X.

  • Proof. “=

⇒”: Suppose φ is flat. Note that φ ⊣⊢ ΘX1 ∨ · · · ∨ ΘXk, where w.l.o.g. we assume that Xi’s are maximal. If k > 1, then pick

X1 X2 X3 Xk · · · · · · v1 v2 vk

{vi} ⊆ Xi and {v1, . . . , vk} Xi for all 1 ≤ i ≤ k = ⇒ {vi} | = ΘXi and {v1, . . . , vk} | = ΘXi for all 1 ≤ i ≤ k = ⇒ {vi} | = φ for all 1 ≤ i ≤ k whereas {v1, . . . , vk} | = φ = ⇒ k = 1 and φ ⊣⊢ ΘX1.

14/19

slide-96
SLIDE 96

We say that a substitution σ of a logic L is well-behaved, if σ is well-defined: σ(φ) is a well-formed formula of L ⊢L is closed under σ: φ ⊢L ψ implies σ(φ) ⊢L σ(ψ)

15/19

slide-97
SLIDE 97

We say that a substitution σ of a logic L is well-behaved, if σ is well-defined: σ(φ) is a well-formed formula of L ⊢L is closed under σ: φ ⊢L ψ implies σ(φ) ⊢L σ(ψ)

Recall: Well-formed formulas of PD are built from the following grammar: φ ::= pi | ¬pi | =(pi1, . . . , pik ) | φ ∧ φ | φ ⊗ φ where pi, pi1, . . . , pik are propositional variables.

15/19

slide-98
SLIDE 98

We say that a substitution σ of a logic L is well-behaved, if σ is well-defined: σ(φ) is a well-formed formula of L ⊢L is closed under σ: φ ⊢L ψ implies σ(φ) ⊢L σ(ψ) For PID: ⊢ ¬¬p → p, but ⊢ ¬¬(p ∨ ¬p) → (p ∨ ¬p)

15/19

slide-99
SLIDE 99

We say that a substitution σ of a logic L is well-behaved, if σ is well-defined: σ(φ) is a well-formed formula of L ⊢L is closed under σ: φ ⊢L ψ implies σ(φ) ⊢L σ(ψ) For PID: ⊢ ¬¬p → p, but ⊢ ¬¬(p ∨ ¬p) → (p ∨ ¬p)

Definition

A substitution σ is called a flat substitution if σ(p) is flat for all p.

Lemma

Flat substitutions are well-behaved in PID and PD.

  • Proof. For PID, it follows from (Ciardelli). For PD, nontrivial.

15/19

slide-100
SLIDE 100

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

16/19

slide-101
SLIDE 101

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-projective iff φ ⊣⊢ ΘX for some n-team X.

16/19

slide-102
SLIDE 102

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-projective iff φ ⊣⊢ ΘX for some n-team X.
  • Proof. “=

⇒”: Note that φ ≡

i∈I ΘXi.

16/19

slide-103
SLIDE 103

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-projective iff φ ⊣⊢ ΘX for some n-team X.
  • Proof. “=

⇒”: Note that φ ≡

i∈I ΘXi. Let σ be the F-projective unifier

  • f φ. By (1), |

= σ(φ) = ⇒ | = σ(ΘXi) for some i ∈ I (by DP) = ⇒ φ | = ΘXi (by (2)).

16/19

slide-104
SLIDE 104

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-projective iff φ ⊣⊢ ΘX for some n-team X.
  • Proof. “=

⇒”: Note that φ ≡

i∈I ΘXi. Let σ be the F-projective unifier

  • f φ. By (1), |

= σ(φ) = ⇒ | = σ(ΘXi) for some i ∈ I (by DP) = ⇒ φ | = ΘXi (by (2)). “⇐ =”: Recall: ΘX =       

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PD; ¬¬

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PID.

16/19

slide-105
SLIDE 105

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-projective iff φ ⊣⊢ ΘX for some n-team X.
  • Proof. “=

⇒”: Note that φ ≡

i∈I ΘXi. Let σ be the F-projective unifier

  • f φ. By (1), |

= σ(φ) = ⇒ | = σ(ΘXi) for some i ∈ I (by DP) = ⇒ φ | = ΘXi (by (2)). “⇐ =”: Recall: ΘX =       

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PD; ¬¬

  • v∈X

(pv(i1)

i1

∧ · · · ∧ pv(in)

in

), for PID. We only give the proof for PD.

16/19

slide-106
SLIDE 106

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-projective iff φ ⊣⊢ ΘX for some n-team X.
  • Proof. (ctd.) View φ = ΘX =

v∈X(pv(i1) i1

∧ · · · ∧ pv(in)

in

) as a formula of CPC,

17/19

slide-107
SLIDE 107

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-projective iff φ ⊣⊢ ΘX for some n-team X.
  • Proof. (ctd.) View φ = ΘX =

v∈X(pv(i1) i1

∧ · · · ∧ pv(in)

in

) as a formula of CPC, then v(φi) = 1 for any v ∈ X.

17/19

slide-108
SLIDE 108

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-projective iff φ ⊣⊢ ΘX for some n-team X.
  • Proof. (ctd.) View φ = ΘX =

v∈X(pv(i1) i1

∧ · · · ∧ pv(in)

in

) as a formula of CPC, then v(φi) = 1 for any v ∈ X. Define σφ

v as follows:

σφ

v (p) =

  • φ ∧ p,

if v(p) = 0; φ → p, if v(p) = 1. (Prucnal’s trick)

17/19

slide-109
SLIDE 109

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-projective iff φ ⊣⊢ ΘX for some n-team X.
  • Proof. (ctd.) View φ = ΘX =

v∈X(pv(i1) i1

∧ · · · ∧ pv(in)

in

) as a formula of CPC, then v(φi) = 1 for any v ∈ X. Define σφ

v as follows:

σφ

v (p) =

  • φ ∧ p,

if v(p) = 0; φ → p, if v(p) = 1. (Prucnal’s trick)

17/19

slide-110
SLIDE 110

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-projective iff φ ⊣⊢ ΘX for some n-team X.
  • Proof. (ctd.) View φ = ΘX =

v∈X(pv(i1) i1

∧ · · · ∧ pv(in)

in

) as a formula of CPC, then v(φi) = 1 for any v ∈ X. Define σφ

v as follows:

σφ

v (p) =

  • φ ∧ p,

if v(p) = 0; φ∼ ⊗ p, if v(p) = 1. (Prucnal’s trick)

17/19

slide-111
SLIDE 111

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-projective iff φ ⊣⊢ ΘX for some n-team X.
  • Proof. (ctd.) View φ = ΘX =

v∈X(pv(i1) i1

∧ · · · ∧ pv(in)

in

) as a formula of CPC, then v(φi) = 1 for any v ∈ X. Define σφ

v as follows:

σφ

v (p) =

  • φ ∧ p,

if v(p) = 0; φ∼ ⊗ p, if v(p) = 1. (Prucnal’s trick) Put σ = σφ

v . Then ⊢CPC σ(φ) as v(φ) = 1.

17/19

slide-112
SLIDE 112

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-projective iff φ ⊣⊢ ΘX for some n-team X.
  • Proof. (ctd.) View φ = ΘX =

v∈X(pv(i1) i1

∧ · · · ∧ pv(in)

in

) as a formula of CPC, then v(φi) = 1 for any v ∈ X. Define σφ

v as follows:

σφ

v (p) =

  • φ ∧ p,

if v(p) = 0; φ∼ ⊗ p, if v(p) = 1. (Prucnal’s trick) Put σ = σφ

v . Then ⊢CPC σ(φ) as v(φ) = 1. Clearly, σ ∈ F

17/19

slide-113
SLIDE 113

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-projective iff φ ⊣⊢ ΘX for some n-team X.
  • Proof. (ctd.) View φ = ΘX =

v∈X(pv(i1) i1

∧ · · · ∧ pv(in)

in

) as a formula of CPC, then v(φi) = 1 for any v ∈ X. Define σφ

v as follows:

σφ

v (p) =

  • φ ∧ p,

if v(p) = 0; φ∼ ⊗ p, if v(p) = 1. (Prucnal’s trick) Put σ = σφ

v . Then ⊢CPC σ(φ) as v(φ) = 1. Clearly, σ ∈ F and σ(φ) is

classical, thus ⊢PD σ(φ).

17/19

slide-114
SLIDE 114

Definition (Projective formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-projective if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) φ, σ(ψ) ⊢L ψ and φ, ψ ⊢L σ(ψ) for all L-formulas ψ.

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-projective iff φ ⊣⊢ ΘX for some n-team X.
  • Proof. (ctd.) View φ = ΘX =

v∈X(pv(i1) i1

∧ · · · ∧ pv(in)

in

) as a formula of CPC, then v(φi) = 1 for any v ∈ X. Define σφ

v as follows:

σφ

v (p) =

  • φ ∧ p,

if v(p) = 0; φ∼ ⊗ p, if v(p) = 1. (Prucnal’s trick) Put σ = σφ

v . Then ⊢CPC σ(φ) as v(φ) = 1. Clearly, σ ∈ F and σ(φ) is

classical, thus ⊢PD σ(φ). Moreover, (2) is satisfied by the choice of σ.

17/19

slide-115
SLIDE 115

Definition (Exact formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-exact if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) ⊢L σ(ψ) = ⇒ φ ⊢L ψ for all L-formulas ψ

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-exact iff φ ⊣⊢ ΘX for some n-team X.

Corollary

For formulas φ of the logics PD and PID, TFAE: φ is flat; φ is F-projective; φ is F-exact; φ ⊣⊢ ΘX for some X

18/19

slide-116
SLIDE 116

Definition (Exact formula)

Let S be a set of well-behaved substitutions of a logic L. An L-formula φ is said to be S-exact if there exists σ ∈ S such that (1) ⊢L σ(φ) (2) ⊢L σ(ψ) = ⇒ φ ⊢L ψ for all L-formulas ψ

Lemma

Let L ∈ {PD, PID} and F the set of all flat substitutions. An n-formula φ

  • f L is F-exact iff φ ⊣⊢ ΘX for some n-team X.

Corollary

For formulas φ of the logics PD and PID, TFAE: φ is flat; φ is F-projective; φ is F-exact; φ ⊣⊢ ΘX for some X

18/19

slide-117
SLIDE 117

Definition

Let S be a set of well-behaved substitutions of a logic L. A rule φ/ψ of L is said to be a S-admissible rule, in symbols φ | ∼S

L ψ or simply φ |

∼S ψ, if for all σ ∈ S, ⊢L σ(φ) = ⇒ ⊢L σ(ψ). L is said to be S-structurally complete if every S-admissible rule of L is derivable, i.e., φ | ∼S

L ψ =

⇒ φ ⊢L ψ.

19/19

slide-118
SLIDE 118

Definition

Let S be a set of well-behaved substitutions of a logic L. A rule φ/ψ of L is said to be a S-admissible rule, in symbols φ | ∼S

L ψ or simply φ |

∼S ψ, if for all σ ∈ S, ⊢L σ(φ) = ⇒ ⊢L σ(ψ). L is said to be S-structurally complete if every S-admissible rule of L is derivable, i.e., φ | ∼S

L ψ =

⇒ φ ⊢L ψ.

19/19

slide-119
SLIDE 119

Definition

Let S be a set of well-behaved substitutions of a logic L. A rule φ/ψ of L is said to be a S-admissible rule, in symbols φ | ∼S

L ψ or simply φ |

∼S ψ, if for all σ ∈ S, ⊢L σ(φ) = ⇒ ⊢L σ(ψ). L is said to be S-structurally complete if every S-admissible rule of L is derivable, i.e., φ | ∼S

L ψ =

⇒ φ ⊢L ψ. Clearly, if L is S-structurally complete, then φ | ∼S

L ψ ⇐

⇒ φ ⊢L ψ.

19/19

slide-120
SLIDE 120

Definition

Let S be a set of well-behaved substitutions of a logic L. A rule φ/ψ of L is said to be a S-admissible rule, in symbols φ | ∼S

L ψ or simply φ |

∼S ψ, if for all σ ∈ S, ⊢L σ(φ) = ⇒ ⊢L σ(ψ). L is said to be S-structurally complete if every S-admissible rule of L is derivable, i.e., φ | ∼S

L ψ =

⇒ φ ⊢L ψ. Clearly, if L is S-structurally complete, then φ | ∼S

L ψ ⇐

⇒ φ ⊢L ψ.

Theorem

PID and PD are F-structurally complete, where F is the set of all flat substitutions.

  • Proof. Since φ ≡

i∈I ΘXi, where each ΘXi is S-projective, for every

formula φ of the logics.

19/19