Unification in modal logic Philippe Balbiani CNRS Toulouse - - PowerPoint PPT Presentation

unification in modal logic
SMART_READER_LITE
LIVE PREVIEW

Unification in modal logic Philippe Balbiani CNRS Toulouse - - PowerPoint PPT Presentation

Unification in modal logic Philippe Balbiani CNRS Toulouse University Institut de recherche en informatique de Toulouse ICLA 2019 Indian Institute of Technology Delhi Introduction Unification problem in a logical system L Given a


slide-1
SLIDE 1

Unification in modal logic

Philippe Balbiani

CNRS — Toulouse University Institut de recherche en informatique de Toulouse

ICLA 2019 Indian Institute of Technology — Delhi

slide-2
SLIDE 2

Introduction

Unification problem in a logical system L

◮ Given a formula ψ(x1, . . . , xn) ◮ Determine whether there exists formulas ϕ1, . . ., ϕn such

that ψ(ϕ1, . . . , ϕn) is in L Admissibility problem in a logical system L

◮ Given a rule of inference ϕ1(x1,...,xn), ..., ϕm(x1,...,xn) ψ(x1,...,xn) ◮ Determine whether for all formulas χ1, . . ., χn, if

ϕ1(χ1, . . . , χn), . . ., ϕm(χ1, . . . , χn) are in L then ψ(χ1, . . . , χn) is in L

slide-3
SLIDE 3

Introduction

Why solving unification problem?

Description logic: Given concept definitions C(x1, . . . , xn) and D(x1, . . . , xn)

◮ Determine whether there are some redundancies between

C(x1, . . . , xn) and D(x1, . . . , xn)

◮ Solve C(x1, . . . , xn) ≡ D(x1, . . . , xn)

Example of a unification problem

◮ C = ∀R.∀R.A ⊓ ∀R.X ◮ D = Y ⊓ ∀R.Y ⊓ ∀R.∀S.A

Solution

◮ Replace X by A⊓∀S.A ◮ Replace Y by ∀R.A

slide-4
SLIDE 4

Introduction

Why solving unification problem?

Epistemic planning: Given variable-free epistemic formulas ϕ(p1, . . . , pm) and ψ(p1, . . . , pm)

◮ Determine whether there exists a public announcement χ

such that | = ϕ → χ!ψ

◮ Solve |

= ϕ → x!ψ

Example of a unification problem

◮ K1p ∧ K2(p → q) → K1x!K2q

Solution

◮ Replace x by p

slide-5
SLIDE 5

Introduction

If L is consistent then the following are equivalent:

◮ Formula ϕ(x1, . . . , xn) is unifiable ◮ Rule ϕ(x1,...,xn) ⊥

is non-admissible If L is finitary then the following are equivalent:

◮ Rule ϕ1(x1,...,xn),...,ϕm(x1,...,xn) ψ(x1,...,xn)

is admissible

◮ Formulas ψ(χ1, . . . , χn) is in L for each maximal unifiers

(χ1, . . . , χn) of formulas ϕ1(x1, . . . , xn), . . . , ϕm(x1, . . . , xn)

slide-6
SLIDE 6

Introduction

Unification: two examples

◮ The formula ¬x ∨ x is unifiable in modal logic K ◮ The formula x → x is unifiable in modal logic K

In Classical Logic

◮ Unification is equivalent to satisfiability ◮ Why ? Use the inference rule of Uniform Substitution

In Modal Logic

◮ Unification in S4, S5, etc is not equivalent to satisfiability ◮ Why ? Consider the formula ♦x ∧ ♦¬x and use the

inference rule of Uniform Substitution

slide-7
SLIDE 7

Introduction

In Intuitionistic Propositional Logic

The following rules are admissible but not derivable

◮ ¬x→y∨z (¬x→y)∨(¬x→z) — Harrop rule (1960) ◮ (¬¬x→x)→(x∨¬x) ¬¬x∨¬x

— Lemmon-Scott rule

◮ (x→y)→(x∨¬y) ¬¬x∨¬y

— generalized Lemmon-Scott rule

◮ (x→y)→x∨z ((x→y)→x)∨((x→y)→z) — Mints rule (1972)

In S4

The following rule is is admissible but not derivable

◮ ((♦x→x)→(x∨¬x)) ♦x∨¬x

slide-8
SLIDE 8

Introduction

About Classical Propositional Logic

Classical Propositional Logic is structurally complete

◮ Thus, admissibility in Classical Propositional Logic is

decidable

About intermediate logics

Rybakov (1981): If L is an intermediate logic then the following are equivalent

◮ Rule R is admissible in L ◮ The modal translation of rule R is admissible in the

greatest modal companion of L

slide-9
SLIDE 9

Introduction

Rybakov (1982)

◮ The admissibility problem in extensions of S4.3 is

decidable Rybakov (1984)

◮ The admissibility problem in S4 is decidable

Chagrov (1992)

◮ There exists a decidable normal modal logic with an

undecidable admissibility problem Wolter and Zakharyaschev (2008)

◮ The unification problem for any normal modal logic

between KU and K4U is undecidable

slide-10
SLIDE 10

Introduction

Contents

◮ Definitions ◮ Boolean unification ◮ Modal unification ◮ Unification types in modal logics ◮ Unification in description logics ◮ Recent advances

slide-11
SLIDE 11

Definitions

Substitutions

◮ σ: variable x → formula σ(x)

Applying substitutions to formulas

◮ σ(ϕ(x1, . . . , xn)) = ϕ(σ(x1), . . . , σ(xn))

Composition of substitutions

◮ σ ◦ τ: variable x → formula τ(σ(x))

slide-12
SLIDE 12

Definitions

Let L be a propositional logic

Equivalence relation between substitutions

◮ σ ≃L τ iff for all variables x, σ(x) ↔ τ(x) ∈ L ◮ “σ and τ are L-equivalent” ◮ Example in Classical Propositional Logic :

◮ σ(x) = x ↔ y ◮ τ(x) = (x ∧ y) ∨ (¬x ∧ ¬y)

Partial order between substitutions

◮ σ L τ iff there exists a substitution µ such that σ ◦ µ ≃L τ ◮ “σ is less specific, more general than τ in L” ◮ Example in Classical Propositional Logic :

◮ σ(x) = x ∨ y ◮ τ(x) = (x ∧ y) ∨ (¬x ∧ ¬y)

slide-13
SLIDE 13

Definitions

Unifiers

◮ A substitution σ is a unifier of a formula ϕ iff σ(ϕ) ∈ L

Complete sets of unifiers

◮ A set Σ of unifiers of a formula ϕ is complete iff for all

unifiers τ of ϕ, there exists a unifier σ of ϕ in Σ such that σ L τ Important questions

◮ Given a formula, has it a unifier? ◮ If so, has it a minimal complete set of unifiers? ◮ If so, how large is this set? Is this set effectively calculable?

slide-14
SLIDE 14

◮ Definitions ◮ Boolean unification ◮ Modal unification ◮ Unification types in modal logics ◮ Unification in description logics ◮ Recent advances

slide-15
SLIDE 15

Boolean unification

Syntax

◮ ϕ ::= x | p | ⊥ | ¬ϕ | (ϕ ∨ ψ)

Abbreviations for ⊤, ∧, etc

◮ As usual

Example of a Boolean unification problem

◮ (x ↔ y) ↔ (x ∨ y)

Solution

◮ σ(x) = ⊤ and σ(y) = ⊤

slide-16
SLIDE 16

Boolean unification

Syntax

◮ ϕ ::= x | p | ⊥ | ¬ϕ | (ϕ ∨ ψ)

Abbreviations for ⊤, ∧, etc

◮ As usual

Example of a Boolean unification problem

◮ (x → y) ∧ (¬x → z)

Solutions

◮ σ(x) = ⊥, σ(y) = y and σ(z) = ⊤ ◮ σ(x) = ⊤, σ(y) = ⊤ and σ(z) = z ◮ σ(x) = x ∧ y, σ(y) = (x ∧ y) ∨ (y ∧ z) and σ(z) = x ∧ y → z

slide-17
SLIDE 17

Boolean unification

Syntax

◮ ϕ ::= x | p | ⊥ | ¬ϕ | (ϕ ∨ ψ)

Abbreviations for ⊤, ∧, etc

◮ As usual

Example of a Boolean unification problem

◮ (x → p) ∧ (q → y)

Solutions

◮ σ(x) = ⊥ and σ(y) = ⊤ ◮ σ(x) = p and σ(y) = q ◮ σ(x) = p ∧ x and σ(y) = q ∨ y

slide-18
SLIDE 18

Boolean unification

Proposition

Without parameters, Boolean unification is NP-complete

◮ ϕ(¯

x) is CPL-unifiable ⇐ ⇒ ∃¯ xϕ(¯ x) is QBF-valid With parameters, Boolean unification is ΠP

2 -complete ◮ ϕ(¯

p, ¯ x) is CPL-unifiable ⇐ ⇒ ∀¯ p∃¯ xϕ(¯ p, ¯ x) is QBF-valid Baader (1998)

slide-19
SLIDE 19

Boolean unification

Projective formulas

◮ A formula ϕ is said to be projective iff it has a unifier σ

such that ϕ → (σ(x) ↔ x) is in CPL Any unifier σ of ϕ satisfying the above condition is called a projective unifier of ϕ Lemma Projective unifiers are closed under compositions Lemma Projective unifiers are most general unifiers

slide-20
SLIDE 20

Boolean unification

Lemma Unifiable formulas are projective Proof: Consider a unifier σ of ϕ

◮ Let ǫ be the substitution such that

ǫ(x) = (ϕ ∧ x) ∨ (¬ϕ ∧ σ(x))

◮ Fact:

  • 1. ϕ → (ǫ(ψ) ↔ ψ) is in CPL
  • 2. ¬ϕ → (ǫ(ψ) ↔ σ(ψ)) is in CPL

◮ Thus, ǫ is a projective unifier of ϕ

slide-21
SLIDE 21

Boolean unification

Lemma Projective unifiers are closed under compositions Lemma Projective unifiers are most general unifiers Lemma Unifiable formulas are projective Proposition Boolean unification is unitary, i.e. every unifiable formula has a most general unifier

slide-22
SLIDE 22

Boolean unification

◮ Baader, F.: On the complexity of Boolean unification.

Information Processing Letters 67 (1998) 215–220.

◮ Baader, F., Ghilardi, S.: Unification in modal and

description logics. Logic Journal of the IGPL 19 (2011) 705–730.

◮ Martin, U., Nipkow, T.: Boolean unification — the story so

  • far. Journal of Symbolic Computation 7 (1989) 275–293.
slide-23
SLIDE 23

◮ Definitions ◮ Boolean unification ◮ Modal unification ◮ Unification types in modal logics ◮ Unification in description logics ◮ Recent advances

slide-24
SLIDE 24

Modal unification

Syntax

◮ ϕ ::= x | p | ⊥ | ¬ϕ | (ϕ ∨ ψ) | ϕ

Abbreviation

◮ ♦ϕ ::= ¬¬ϕ

Examples of modal unification problems

◮ ¬x ∨ x ◮ x → x ◮ (x → p) ∧ (x → (p → x))

slide-25
SLIDE 25

Modal unification

Semantics

◮ Frame: directed graph F = (W, R) ◮ Models: M = (W, R, V) where V: x → V(x) ⊆ W

Truth conditions in a model

◮ M, s |

= x iff s ∈ V(x)

◮ M, s|

=⊥

◮ M, s |

= ¬ϕ iff M, s| =ϕ

◮ M, s |

= ϕ ∨ ψ iff M, s | = ϕ or M, s | = ψ

◮ M, s |

= ϕ iff ∀t ∈ W, if sRt then M, t | = ϕ

slide-26
SLIDE 26

Modal unification

Semantics

◮ Frame: directed graph F = (W, R) ◮ Models: M = (W, R, V) where V: x → V(x) ⊆ W

Validity in a frame

◮ ϕ is valid in frame F iff ϕ is true at every node of every

model based on F

slide-27
SLIDE 27

Modal unification

Normal modal logics

◮ A set L of formula is a normal modal logic iff

  • 1. L contains all tautologies
  • 2. L contains (ϕ → ψ) → (ϕ → ψ)
  • 3. L is closed under modus ponens: ϕ, ϕ→ψ

ψ

  • 4. L is closed under uniform substitution:

ϕ σ(ϕ)

  • 5. L is closed under generalization:

ϕ ϕ

slide-28
SLIDE 28

Modal unification

Examples of normal modal logics

◮ Least normal modal logic: K ◮ Additional axiom D: ♦⊤ ◮ Additional axiom T: ϕ → ϕ ◮ Additional axiom 4: ϕ → ϕ ◮ Additional axiom 5: ♦ϕ → ♦ϕ

slide-29
SLIDE 29

Modal unification

Remark that the following statements are equivalent

◮ Formula ϕ(x1, . . . , xn) is unifiable ◮ Rule ϕ(x1,...,xn) ⊥

is non-admissible Thus, unification can be reduced to non-admissibility Ghilardi (1999) observed that in many normal modal logics L, Admissibility can be reduced to unification

◮ Assume that for a unifiable formula ϕ, one can compute a

finite complete set Σ of unifiers

◮ Thus, to decide whether the rule ϕ ψ is admissible in L, it is

enough to enumerate Σ and to check whether σ(ψ) is in L for all σ in Σ

slide-30
SLIDE 30

Modal unification

Lemma The unification problem is trivially decidable (NP-complete) for any normal modal logic containing ♦⊤

◮ KD, KT, S4, S4.3, S5

From the results of Rybakov 1984, 1997

◮ The unification and admissibility problems are decidable

for intuitionistic logic, GL and S4 From the results of Je˘ r´ abek 2005, 2007

◮ The admissibility problem is coNEXPTIME-complete for

intuitionistic logic, GL and S4

slide-31
SLIDE 31

Modal unification

From the results of Chagrov 1992

◮ Only one — rather artificial — example of a decidable

unimodal logic for which the admissibility problem is undecidable

slide-32
SLIDE 32

Modal unification

Admissibility in Alt1 × Alt1 is undecidable

Syntax

◮ ϕ ::= x | ⊥¬ϕ | (ϕ ∨ ψ) | [h]ϕ | [v]ϕ

Abbreviations

◮ hϕ ::= ¬[h]¬ϕ ◮ vϕ ::= ¬[v]¬ϕ

slide-33
SLIDE 33

Modal unification

Admissibility in Alt1 × Alt1 is undecidable

Semantics

◮ Frame: grid F = (I, J) where I, J ≥ 1 ◮ Models: M = (I, J, V) where V:

x → V(x) ⊆ {1, . . . , I} × {1, . . . , J} Truth conditions in a model

◮ M, (i, j) |

= x iff (i, j) ∈ V(x)

◮ M, (i, j) |

= [h]ϕ iff if i < I then M, (i + 1, j) | = ϕ

◮ M, (i, j) |

= [v]ϕ iff if j < J then M, (i, j + 1) | = ϕ Satisfiability is

◮ NP-complete

slide-34
SLIDE 34

Modal unification

Admissibility in Alt1 × Alt1 is undecidable

Semantics

◮ Frame: grid F = (I, J) where I, J ≥ 1 ◮ Models: M = (I, J, V) where V:

x → V(x) ⊆ {1, . . . , I} × {1, . . . , J} Truth conditions in a model

◮ M, (i, j) |

= x iff (i, j) ∈ V(x)

◮ M, (i, j) |

= [h]ϕ iff if i < I then M, (i + 1, j) | = ϕ

◮ M, (i, j) |

= [v]ϕ iff if j < J then M, (i, j + 1) | = ϕ Admissibility is

◮ undecidable

slide-35
SLIDE 35

Modal unification

Admissibility in Alt1 × Alt1 is undecidable

Computability of admissibility

◮ undecidable

Proof

◮ Reduction of the domino-tiling problem

(∆, H, V, ∆u, ∆d, ∆l, ∆r) where

◮ ∆ is a finite set of domino-types ◮ H and V are binary relations on ∆ ◮ ∆u, ∆d, ∆l, ∆r are subsets of ∆

slide-36
SLIDE 36

Modal unification

Ku: least normal modal logic with the universal modality K4u: least normal modal logic with the universal modality that contains the extra formula

◮ x → x

From the results of Wolter and Zakharyaschev 2008

◮ The unification problem for modal logics between Ku and

K4u is undecidable

slide-37
SLIDE 37

Modal unification

The unification and admissibility problems for K itself . . .

◮ . . . still remain open

Unfortunately, nothing is known about

◮ The decidability status of the unification and admissibility

problems for

◮ Basic modal logic K ◮ Various multimodal logics ◮ Various hybrid logics ◮ Various description logics

slide-38
SLIDE 38

◮ Definitions ◮ Boolean unification ◮ Modal unification ◮ Unification types in modal logics ◮ Unification in description logics ◮ Recent advances

slide-39
SLIDE 39

Unification types in modal logics

Unification types in propositional logic

Let L be a propositional logic and ϕ be a formula An L-unifier of ϕ is a substitution σ such that

◮ σ(ϕ) ∈ L

We shall say that ϕ is of type unitary (1) for L iff

◮ There exists a complete minimal set Σ of L-unifiers of ϕ ◮ Card(Σ) = 1

slide-40
SLIDE 40

Unification types in modal logics

Unification types in propositional logic

Let L be a propositional logic and ϕ be a formula An L-unifier of ϕ is a substitution σ such that

◮ σ(ϕ) ∈ L

We shall say that ϕ is of type finitary (ω) for L iff

◮ There exists a complete minimal set Σ of L-unifiers of ϕ ◮ Card(Σ) = 1 but Σ is finite

slide-41
SLIDE 41

Unification types in modal logics

Unification types in propositional logic

Let L be a propositional logic and ϕ be a formula An L-unifier of ϕ is a substitution σ such that

◮ σ(ϕ) ∈ L

We shall say that ϕ is of type infinitary (∞) for L iff

◮ There exists a complete minimal set Σ of L-unifiers of ϕ ◮ Σ is infinite

slide-42
SLIDE 42

Unification types in modal logics

Unification types in propositional logic

Let L be a propositional logic and ϕ be a formula An L-unifier of ϕ is a substitution σ such that

◮ σ(ϕ) ∈ L

We shall say that ϕ is of type nullary (0) for L iff

◮ There exists no complete minimal set of L-unifiers of ϕ

slide-43
SLIDE 43

Unification types in modal logics

Unification types in propositional logic

Let L be a propositional logic We shall say that L is of type unitary/finitary iff

◮ For all formulas ϕ, ϕ is of type unitary/finitary for L

Examples

◮ Unification in classical propositional logic is unitary ◮ Unification in intuitionistic propositional logic is finitary

slide-44
SLIDE 44

Unification types in modal logics

Unification types in propositional logic

Let L be a propositional logic We shall say that L is of type infinitary/nullary iff

◮ There exists a formula ϕ such that ϕ is of type

infinitary/nullary for L Example

◮ Unification in modal logic K is nullary

slide-45
SLIDE 45

Unification types in modal logics

Unification in intuitionistic propositional logic

We have seen: CPL-unification is unitary, i.e.

◮ For all formulas ϕ(x1, . . . , xn), the cardinality of a minimal

complete set of CPL-unifiers is at most 1 Ghilardi (1999) has demonstrated that IPL-unification is finitary, i.e.

◮ For all formulas ϕ(x1, . . . , xn), the cardinality of a minimal

complete set of IPL-unifiers is finite Example The formulas x ∨ ¬x is IPL-unifiable with the 2 following most general IPL-unifiers

◮ σ(x) = ⊥ ◮ τ(x) = ⊤

slide-46
SLIDE 46

Unification types in modal logics

Unification in intuitionistic propositional logic

We have seen:

◮ The complexity of Boolean unification is NP-complete

It can be easily proved that:

◮ The complexity of IPL-unification is NP-complete too

Lemma For all formulas ϕ, the following statements are equivalent:

◮ ϕ is IPL-unifiable ◮ ϕ is CPL-unifiable

slide-47
SLIDE 47

Unification types in modal logics

Unification in intuitionistic propositional logic

Proposition IPL-unification is NP-complete Proof: By the above Lemma Remark For IPL-unification with constants, see

◮ Rybakov, V.: Rules of inference with parameters for

intuitionistic logic. The Journal of Symbolic Logic 57 (1992) 912–923. Proposition (Ghilardi 1999) IPL-unification is finitary, i.e.

◮ For all formulas ϕ(x1, . . . , xn), the cardinality of a minimal

complete set of IPL-unifiers is finite Proof: We will demonstrate a similar result for K4

slide-48
SLIDE 48

Unification types in modal logics

Unification in K4

Modal logic K4

◮ Syntax

◮ ϕ ::= x | p | ⊥ | ¬ϕ | (ϕ ∨ ψ) | ϕ

◮ Abbreviations

◮ ♦ϕ ::= ¬¬ϕ ◮ +ϕ ::= ϕ ∧ ϕ

◮ Semantics

◮ Frame: directed graph F = (W, R) where R is transitive ◮ Model: M = (W, R, V) where V: p → V(p) ⊆ W

slide-49
SLIDE 49

Unification types in modal logics

Unification in K4

Modal logic K4

◮ Syntax

◮ ϕ ::= x | p | ⊥ | ¬ϕ | (ϕ ∨ ψ) | ϕ

◮ Abbreviations

◮ ♦ϕ ::= ¬¬ϕ ◮ +ϕ ::= ϕ ∧ ϕ

◮ Truth conditions in a model M = (W, R, V)

◮ M, x |

= p iff x ∈ V(p)

◮ M, x |

= ϕ iff ∀y ∈ W, if xRy then M, y | = ϕ

slide-50
SLIDE 50

Unification types in modal logics

Unification in K4

Proposition (Rybakov 1984, 1997) K4-unification is decidable Proof: Later Proposition (Ghilardi 2000) K4-unification is finitary, i.e.

◮ For all formulas ϕ(x1, . . . , xn), the cardinality of a minimal

complete set of K4-unifiers is finite Proof: Later

slide-51
SLIDE 51

Unification types in modal logics

Unification in K4

A formula of the kind +ϕ(x1, . . . , xn) is said to be projective iff there exists a substitution σ such that

  • 1. σ is a K4-unifier of +ϕ
  • 2. +ϕ → (xi ↔ σ(xi)) ∈ K4 for each i such that 1 ≤ i ≤ n

Remark This definition resembles the definition of a unifier being transparent as used in

◮ Dzik, W.: Transparent unifiers in modal logics with

self-conjugate operators. Bulletin of the Section of Logic 35 (2006) 73–83.

slide-52
SLIDE 52

Unification types in modal logics

Unification in K4

A formula of the kind +ϕ(x1, . . . , xn) is said to be projective iff there exists a substitution σ such that

  • 1. σ is a K4-unifier of +ϕ
  • 2. +ϕ → (xi ↔ σ(xi)) ∈ K4 for each i such that 1 ≤ i ≤ n

Example To see that the formula +ϕ = +x is projective, it suffices to consider the substitution σ(x) = ⊤

◮ σ(+ϕ) = +⊤ ◮ +ϕ → (x ↔ σ(x)) = +x → (x ↔ ⊤)

slide-53
SLIDE 53

Unification types in modal logics

Unification in K4

A formula of the kind +ϕ(x1, . . . , xn) is said to be projective iff there exists a substitution σ such that

  • 1. σ is a K4-unifier of +ϕ
  • 2. +ϕ → (xi ↔ σ(xi)) ∈ K4 for each i such that 1 ≤ i ≤ n

Remark The following statements hold:

◮ Such σ is a most general K4-unifier for +ϕ ◮ +ϕ → (ψ ↔ σ(ψ)) ∈ K4 for each formula ψ(x1, . . . , xn) ◮ The set of all substitutions satisfying condition 2 is closed

under compositions

slide-54
SLIDE 54

Unification types in modal logics

Unification in K4

A formula of the kind +ϕ(x1, . . . , xn) is said to be projective iff there exists a substitution σ such that

  • 1. σ is a K4-unifier of +ϕ
  • 2. +ϕ → (xi ↔ σ(xi)) ∈ K4 for each i such that 1 ≤ i ≤ n

For all A ⊆ {1, . . . , n}, let θA

ϕ be the substitution defined by ◮ θA ϕ(xi) = +ϕ → xi if i ∈ A ◮ θA ϕ(xi) = +ϕ ∧ xi if i ∈ A

Remark The substitution θA

ϕ satisfies condition 2

slide-55
SLIDE 55

Unification types in modal logics

Unification in K4

A formula of the kind +ϕ(x1, . . . , xn) is said to be projective iff there exists a substitution σ such that

  • 1. σ is a K4-unifier of +ϕ
  • 2. +ϕ → (xi ↔ σ(xi)) ∈ K4 for each i such that 1 ≤ i ≤ n

For all A ⊆ {1, . . . , n}, let θA

ϕ be the substitution defined by ◮ θA ϕ(xi) = +ϕ → xi if i ∈ A ◮ θA ϕ(xi) = +ϕ ∧ xi if i ∈ A

Given an arbitrary enumeration A1, . . . , A2n of the subsets of {1, . . . , n}, let θϕ = θA1

ϕ ◦ . . . ◦ θA2n ϕ

slide-56
SLIDE 56

Unification types in modal logics

Unification in K4

Proposition For all formulas of the kind +ϕ(x1, . . . , xn), if d = depth(ϕ) and N is the number of non-∼d-equivalent models over x1, . . . , xn, the following statements are equivalent:

◮ θϕ2N is a K4-unifier of +ϕ ◮ +ϕ is projective ◮ Ghilardi, S.: Best solving modal equations. Annals of Pure

and Applied Logic 102 (2000) 183–198. Corollary It is decidable to determine whether a given formula

  • f the kind +ϕ is projective
slide-57
SLIDE 57

Unification types in modal logics

Unification in K4

Lemma For all formulas ϕ and for all substitutions σ, if σ is a K4-unifier of ϕ

◮ There exists a formula of the kind +ψ,

depth(ψ) ≤ depth(ϕ), such that

◮ +ψ is projective ◮ σ is a K4-unifier of +ψ ◮ +ψ → ϕ ∈ K4

◮ Ghilardi, S.: Best solving modal equations. Annals of Pure

and Applied Logic 102 (2000) 183–198.

slide-58
SLIDE 58

Unification types in modal logics

Unification in K4

Proposition (Ghilardi 2000) K4-unification is finitary, i.e.

◮ For all formulas ϕ(x1, . . . , xn), the cardinality of a minimal

complete set of K4-unifiers is finite Corollary K4-unification is decidable Proof: Given a formula ϕ

◮ Determine whether there exists a formula of the kind +ψ,

depth(ψ) ≤ depth(ϕ), such that

◮ +ψ is projective ◮ +ψ → ϕ ∈ K4

slide-59
SLIDE 59

Unification types in modal logics

Unification in S5

Modal logic S5

◮ Syntax

◮ ϕ ::= x | p | ⊥ | ¬ϕ | (ϕ ∨ ψ) | ϕ

◮ Abbreviations

◮ ♦ϕ ::= ¬¬ϕ

◮ Semantics

◮ Frame: partition F = (W, R), i.e. R is an equivalence

relation

◮ Model: M = (W, R, V) where V: x → V(x) ⊆ W

slide-60
SLIDE 60

Unification types in modal logics

Unification in S5

Modal logic S5

◮ Syntax

◮ ϕ ::= x | p | ⊥ | ¬ϕ | (ϕ ∨ ψ) | ϕ

◮ Abbreviations

◮ ♦ϕ ::= ¬¬ϕ

◮ Truth conditions in a model M = (W, R, V)

◮ M, x |

= p iff x ∈ V(p)

◮ M, x |

= ϕ iff ∀y ∈ W, if xRy then M, y | = ϕ

slide-61
SLIDE 61

Unification types in modal logics

Unification in S5

Modal logic S5

◮ Syntax

◮ ϕ ::= x | p | ⊥ | ¬ϕ | (ϕ ∨ ψ) | ϕ

◮ Abbreviations

◮ ♦ϕ ::= ¬¬ϕ

◮ Important properties of modal logic S5

◮ For all formulas ϕ, ψ, (ϕ ∧ ψ) ↔ ϕ ∧ ψ ∈ S5 ◮ For all formulas ϕ, ψ, (♦ϕ ∨ ψ) ↔ ♦ϕ ∨ ψ ∈ S5 ◮ For all variable-free formulas ϕ, either ¬ϕ ∈ S5, or ϕ ∈ S5

slide-62
SLIDE 62

Unification types in modal logics

Unification in S5

Remark The unification problem is NP-complete for S5 Remark In case we allow extra parameters in the formulas, S5-unification becomes a more serious problem The formula ϕ(p1, . . . , pm, x1, . . . , xn) with parameters p1, . . . , pm and variables x1, . . . , xn is S5-unifiable iff there exists formulas χ1, . . . , χn such that ϕ(p1, . . . , pm, χ1, . . . , χn) ∈ S5 Remark If ϕ(p1, . . . , pm, x1, . . . , xn) is S5-unifiable then there exists an S5-unifier based only on parameters p1, . . . , pm

slide-63
SLIDE 63

Unification types in modal logics

Unification in S5

Remark In case we allow extra parameters in the formulas, S5-unification becomes a more serious problem Proposition S5-unification with parameters is in ΠEXP

2

Claim S5-unification with parameters is coNEXPTIME-hard

slide-64
SLIDE 64

Unification types in modal logics

Unification in S5

Proposition (Dzik 2003) S5-unification is unitary, i.e.

◮ For all formulas ϕ(x1, . . . , xn), the cardinality of a minimal

complete set of S5-unifiers is at most 1 Proof: Assume ϕ(x1, . . . , xn) is S5-unifiable

◮ Thus, there exists a ground substitution σ such that

σ(ϕ) ∈ S5

◮ Let τ be the substitution defined by

◮ τ(xi) = ϕ → xi if σ(xi) ∈ S5 ◮ τ(xi) = ϕ ∧ xi if ¬σ(xi) ∈ S5

◮ It can be proved that τ is a most general S5-unifier of ϕ

slide-65
SLIDE 65

Unification types in modal logics

Unification in S5

Remark The proofs that the unification problems in classical propositional logic and in S5 are unitary are based on the

◮ Fact Given a unifiable formula ϕ(x1, . . . , xn),

◮ There exists a unifier σ of ϕ such that for all i, if 1 ≤ i ≤ n,

ϕ → (xi ↔ σ(xi)) ∈ L

Remark This fact is used, for example, by Dzik (transparent unifiers) and Ghilardi (projective formulas) in

◮ Dzik, W.: Transparent unifiers in modal logics with

self-conjugate operators. Bulletin of the Section of Logic 35 (2006) 73–83.

◮ Ghilardi, S.: Best solving modal equations. Annals of Pure

and Applied Logic 102 (2000) 183–198.

slide-66
SLIDE 66

Unification types in modal logics

Unification in S5

Remark The proofs that the unification problems in classical propositional logic and in S5 are unitary are based on the

◮ Fact Given a unifiable formula ϕ(x1, . . . , xn),

◮ There exists a unifier σ of ϕ such that for all i, if 1 ≤ i ≤ n,

ϕ → (xi ↔ σ(xi)) ∈ L

Remark It is true that if L satisfies the above fact, L-unification is unitary but the converse is not always true

◮ S4.2Grz-unification is unitary (Ghilardi 2000) ◮ S4.2Grz does not satisfy the above fact (Dzik 2006)

slide-67
SLIDE 67

Unification types in modal logics

Unification in K

Modal logic K

◮ Syntax

◮ ϕ ::= x | ⊥ | ¬ϕ | (ϕ ∨ ψ) | ϕ

◮ Abbreviations

◮ ♦ϕ ::= ¬¬ϕ ◮ <nϕ ::= 0ϕ ∧ . . . ∧ n−1ϕ for each n ∈ N

◮ Semantics

◮ Frame: directed graph F = (W, R) ◮ Model: M = (W, R, V) where V: p → V(p) ⊆ W

slide-68
SLIDE 68

Unification types in modal logics

Unification in K

Modal logic K

◮ Syntax

◮ ϕ ::= x | ⊥ | ¬ϕ | (ϕ ∨ ψ) | ϕ

◮ Abbreviations

◮ ♦ϕ ::= ¬¬ϕ ◮ <nϕ ::= 0ϕ ∧ . . . ∧ n−1ϕ for each n ∈ N

◮ Truth conditions in a model M = (W, R, V)

◮ M, x |

= p iff x ∈ V(p)

◮ M, x |

= ϕ iff ∀y ∈ W, if xRy then M, y | = ϕ

slide-69
SLIDE 69

Unification types in modal logics

Unification in K

Open question Is K-unification decidable? K-unification is not unitary since

◮ σ⊤(x) = ⊤ and σ⊥(x) = ⊥ constitute a minimal complete

set of unifiers in K of the formula ♦x → x Our purpose: demonstrate that K-unification is nullary, i.e.

◮ There exists a formula ϕ such that there exists no

complete minimal set of K-unifiers of ϕ Method (Je˘ r´ abek, 2014) Study the K-unifiers of

◮ x → x

slide-70
SLIDE 70

Unification types in modal logics

Unification in K

Method (Je˘ r´ abek, 2014) Study the K-unifiers of

◮ x → x

Consider the following substitutions

◮ σn(x) = <nx ∧ n⊥ for each n ∈ N ◮ σ⊤(x) = ⊤

Lemma

◮ σn is a K-unifier of x → x for each n ∈ N ◮ σ⊤ is a K-unifier of x → x

slide-71
SLIDE 71

Unification types in modal logics

Unification in K

Method (Je˘ r´ abek, 2014) Study the K-unifiers of

◮ x → x

Consider the following substitutions

◮ σn(x) = <nx ∧ n⊥ for each n ∈ N ◮ σ⊤(x) = ⊤

Lemma For all K-unifiers σ of x → x and for all n ∈ N, the following statements are equivalent:

◮ σ ≤K σn ◮ σ(x) → n⊥ ∈ K

slide-72
SLIDE 72

Unification types in modal logics

Unification in K

Method (Je˘ r´ abek, 2014) Study the K-unifiers of

◮ x → x

Consider the following substitutions

◮ σn(x) = <nx ∧ n⊥ for each n ∈ N ◮ σ⊤(x) = ⊤

Lemma For all substitutions σ, the following statements are equivalent:

◮ σ ≤K σ⊤ ◮ σ(x) ∈ K

slide-73
SLIDE 73

Unification types in modal logics

Unification in K

Proposition (Je˘ r´ abek, 2014) For all formulas ϕ, depth(ϕ) = n,

◮ If ϕ → ϕ ∈ K then either ϕ → n⊥ ∈ K, or ϕ ∈ K

Corollary The following substitutions form a complete set of K-unifiers for the formula x → x

◮ σn(x) = <nx ∧ n⊥ for each n ∈ N ◮ σ⊤(x) = ⊤

Corollary K-unification is nullary, i.e.

◮ There exists a formula ϕ such that there exists no

complete minimal set of K-unifiers of ϕ Proof: Take ϕ = x → x

slide-74
SLIDE 74

◮ Definitions ◮ Boolean unification ◮ Modal unification ◮ Unification types in modal logics ◮ Unification in description logics ◮ Recent advances

slide-75
SLIDE 75

Unification in description logics

Syntax of the basic Boolean description language FL0

◮ C ::= A | ⊤ | (C ⊓ D) | ∀R.C — concept descriptions ◮ A denotes an arbitrary atomic concept ◮ R denotes an arbitrary atomic role

Example of FL0-concept description

◮ Woman ⊓ ∀child.Woman

See

◮ Baader, F.: Terminological cycles in KL-ONE-based

knowledge representation languages. In: AAAI’90 Proceedings of the eighth National conference on Artificial

  • intelligence. AAAI Press (1990) 621–626.
slide-76
SLIDE 76

Unification in description logics

Syntax of the basic Boolean description language FL0

◮ C ::= A | ⊤ | (C ⊓ D) | ∀R.C — concept descriptions

An interpretation is a pair I = (∆I, ·I) where

◮ ∆I is a non-empty set — the domain of I ◮ ·I is the interpretation function

◮ A → AI ⊆ ∆I ◮ R → RI ⊆ ∆I × ∆I

slide-77
SLIDE 77

Unification in description logics

An interpretation is a pair I = (∆I, ·I) where

◮ ∆I is a non-empty set — the domain of I ◮ ·I is the interpretation function

◮ A → AI ⊆ ∆I ◮ R → RI ⊆ ∆I × ∆I

The interpretation function ·I is inductively extended to concept descriptions

◮ (A)I = AI ◮ (⊤)I = ∆I ◮ (C ⊓ D)I = (C)I ∩ (D)I ◮ (∀R.C)I = {d ∈ ∆I: ∀e ∈ ∆I, if (d, e) ∈ RI then

e ∈ (C)I}

slide-78
SLIDE 78

Unification in description logics

Two concept descriptions C, D are equivalent (C ≡ D) iff

◮ (C)I = (D)I holds for all interpretations I

The concept description D subsumes the concept description C (C ⊑ D) iff

◮ (C)I ⊆ (D)I holds for all interpretations I

Proposition Equivalence and subsumption of FL0-concept descriptions can be decided in polynomial time Proof:

◮ Levesque, H., Brachman, R.: Expressiveness and

tractability in knowledge representation and reasoning. Computational Intelligence 3 (1987) 78–93.

slide-79
SLIDE 79

Unification in description logics

We partition the set of all atomic concepts into

◮ A set of concept variables — denoted X, Y, . . . ◮ A set of concept constants — denoted A, B, . . .

Syntax of the basic Boolean description language FL0 with variables and constants

◮ C ::= X | A | ⊤ | (C ⊓ D) | ∀R.C — concept descriptions

slide-80
SLIDE 80

Unification in description logics

We partition the set of all atomic concepts into

◮ A set of concept variables — denoted X, Y, . . . ◮ A set of concept constants — denoted A, B, . . .

Now, an interpretation is a pair I = (∆I, ·I) where

◮ ∆I is a non-empty set — the domain of I ◮ ·I is the interpretation function

◮ X → X I ⊆ ∆I ◮ A → AI ⊆ ∆I ◮ R → RI ⊆ ∆I × ∆I

slide-81
SLIDE 81

Unification in description logics

A substitution σ is a mapping from the set of all concept variables into the set of all FL0-concept descriptions This mapping is inductively extended to concept descriptions

◮ σ(A) = A ◮ σ(⊤) = ⊤ ◮ σ(C ⊓ D) = σ(C) ⊓ σ(D) ◮ σ(∀R.C) = ∀R.σ(C)

slide-82
SLIDE 82

Unification in description logics

The substitution σ is a unifier of FL0-concept descriptions C and D iff

◮ σ(C) ≡ σ(D)

The FL0-concept descriptions C and D are unifiable iff they have a unifier Example The substitution σ defined by

◮ σ(X) = A ⊓ ∀S.A and σ(Y) = ∀R.A

is a unifier of the FL0-concept descriptions

◮ C = ∀R.∀R.A ⊓ ∀R.X ◮ D = Y ⊓ ∀R.Y ⊓ ∀R.∀S.A

slide-83
SLIDE 83

Unification in description logics

A substitution is ground iff

◮ The FL0-concept descriptions it substitutes for the

variables do not contain variables Remark For all FL0-concept descriptions C, D, the following statements are equivalent:

◮ There exists a unifier of C and D ◮ There exists a ground unifier of C and D

slide-84
SLIDE 84

Unification in description logics

Lemma For all FL0-concept descriptions C1, . . . , Cn, D1, . . . , Dn and for all pairwise distinct roles R1, . . . , Rn, the following statements are equivalent:

◮ C1 ≡ D1, . . ., Cn ≡ Dn ◮ ∀R1.C1 ⊓ . . . ⊓ ∀Rn.Cn ≡ ∀R1.D1 ⊓ . . . ⊓ ∀Rn.Dn

slide-85
SLIDE 85

Unification in description logics

Given finite sets S0, . . . , Sn, T0, . . . , Tn of words over the alphabet of role names, we consider the equation

◮ S0 ∪ S1 · X1 ∪ . . . ∪ Sn · Xn = T0 ∪ T1 · X1 ∪ . . . ∪ Tn · Xn

where

◮ ∪ stands for set union ◮ · stands for element-wise concatenation of sets of words

Examples

◮ {R} ∪ {RS} · X = {RSS} ∪ {R} · X ◮ {RR} ∪ {RS} · Y = {RSR, RR} ∪ {R} · Y

slide-86
SLIDE 86

Unification in description logics

Given two FL0-concept descriptions C, D, let

◮ X1, . . . , Xn be the concept variables that occur in C, D ◮ A1, . . . , Ak be the concept constants that occur in C, D

Abbreviating ∀R1. . . . ∀Rm. by ∀R1 . . . Rm., the FL0-concept descriptions C, D can be rewritten

◮ C ≡ ∀S0,1.A1 ⊓ . . . ⊓ ∀S0,k.Ak ⊓ ∀S1.X1 ⊓ . . . ⊓ ∀Sn.Xn ◮ D ≡ ∀T0,1.A1 ⊓ . . . ⊓ ∀T0,k.Ak ⊓ ∀T1.X1 ⊓ . . . ⊓ ∀Tn.Xn

for finite sets of words S0,i, Sj, T0,i, Tj

slide-87
SLIDE 87

Unification in description logics

Theorem (Baader and Narendran 2001) Let C, D be FL0-concept descriptions such that

◮ C ≡ ∀S0,1.A1 ⊓ . . . ⊓ ∀S0,k.Ak ⊓ ∀S1.X1 ⊓ . . . ⊓ ∀Sn.Xn ◮ D ≡ ∀T0,1.A1 ⊓ . . . ⊓ ∀T0,k.Ak ⊓ ∀T1.X1 ⊓ . . . ⊓ ∀Tn.Xn

The following statements are equivalent:

◮ The FL0-concept descriptions C and D are unifiable ◮ For all i, if 1 ≤ i ≤ k, the linear equation EC,D(Ai)

◮ S0,i ∪ S1 · X1,i ∪ . . . ∪ Sn · Xn,i = T0,i ∪ T1 · X1,i ∪ . . . ∪ Tn · Xn,i

has a solution

slide-88
SLIDE 88

Unification in description logics

Example Let C, D be the following FL0-concept descriptions

◮ C = ∀R.(A1 ⊓ ∀R.A2) ⊓ ∀R.∀S.X1 ◮ D = ∀R.∀S.(∀S.A1 ⊓ ∀R.A2) ⊓ ∀R.X1 ⊓ ∀R.∀R.A2

Then

◮ C ≡ C′ = ∀{R}.A1 ⊓ ∀{RR}.A2 ⊓ ∀{RS}.X1 ◮ D ≡ D′ = ∀{RSS}.A1 ⊓ ∀{RSR, RR}.A2 ⊓ ∀{R}.X1

The unification of C′, D′ leads to the two linear equations

◮ {R} ∪ {RS} · X1,1 = {RSS} ∪ {R} · X1,1 ◮ {RR} ∪ {RS} · X1,2 = {RSR, RR} ∪ {R} · X1,2

slide-89
SLIDE 89

Unification in description logics

Theorem (Baader and Narendran 2001) Solvability of linear equations can be decided in deterministic exponential time Corollary (Baader and Narendran 2001) Solvability of unification problems in FL0 can be decided in deterministic exponential time

slide-90
SLIDE 90

◮ Definitions ◮ Boolean unification ◮ Modal unification ◮ Unification types in modal logics ◮ Unification in description logics ◮ Recent advances

slide-91
SLIDE 91

Recent advances

Description logic EL

◮ Unification in EL is NP-complete ◮ Unification in EL−⊤ is PSPACE-complete

Baader, F ., Binh, N., Borgwardt, S., Morawska, B.: Deciding unifiability and computing local unifiers in the description logic EL without top constructor. Notre Dame Journal of Formal Logic 57 (2016) 443–476.

slide-92
SLIDE 92

Recent advances

KD = K + ♦⊤

KD is nullary

◮ x → p ◮ x → (p → x)

Balbiani, P ., Gencer, C ¸ .: KD is nullary. Journal of Applied Non-Classical Logics 27 (2018) 196–205.

slide-93
SLIDE 93

Recent advances

KT = K + ϕ → ϕ

KT is nullary

◮ x → p ◮ x → (q → y) ◮ y → q ◮ y → (p → x)

Balbiani, P .: Remarks about the unification type of several non-symmetric non-transitive modal logics. Logic Journal of the IGPL (to appear).

slide-94
SLIDE 94

Recent advances

KB = K + ϕ → ♦ϕ

KB is nullary

◮ x → (¬p∧¬q → (p∧¬q → (¬p∧q → (¬p∧¬q → x))))

Balbiani, P ., Gencer, C ¸ .: About the unification type of simple symmetric modal logics. Submitted for publication.

slide-95
SLIDE 95

Recent advances

Alt1 = K + ♦ϕ → ϕ

◮ Alt1 is nullary for unification ◮ The unification problem (without parameters) in Alt1 is

decidable (in PSPACE) Balbiani, P ., Tinchev, T.: Unification in modal logic Alt1. In Beklemishev, L., Demri, S., M´ at´ e, A. (editors): Advances in Modal Logic. Volume 11. College Publications (2016) 117–134.

slide-96
SLIDE 96

Recent advances

Normal extensions of K5 = K + ♦ϕ → ♦ϕ

◮ These modal logics are unitary for unification

K + k⊥ for k ≥ 2

◮ These modal logics are finitary for unification

Balbiani, P ., Rostamigiv, M., Tinchev, T.: About the unification type of some locally tabular modal logics. Submitted for publication.

slide-97
SLIDE 97

Recent advances

Unification in Dynamic Epistemic Logics

Syntax

◮ ϕ ::= x | p | ⊥ | ¬ϕ | (ϕ ∨ ψ) | Kaϕ | [ϕ]ψ

Abbreviations

◮ ˆ

Kaϕ ::= ¬Ka¬ϕ

◮ ϕψ ::= ¬[ϕ]¬ψ

Readings

◮ Kaϕ: “agent a knows that ϕ holds” ◮ [ϕ]ψ: “if ϕ holds then ψ will hold after ϕ is announced” ◮ ˆ

Kaϕ: “it is compatible with a’s knowledge that ϕ holds”

◮ ϕψ: “ϕ holds and ψ will hold after ϕ is announced”

slide-98
SLIDE 98

Recent advances

Unification in Dynamic Epistemic Logics

Example of unification problems ϕ(¯ p) → xKaψ(¯ p)

◮ ϕ(¯

p) describes an initial situation

◮ x is the announcement ◮ Kaψ(¯

p) — with ψ(¯ p) Boolean formula — is a goal formula

Other examples of unification problems

◮ ϕ → xKaψ ◮ ϕ → x(Ka1ψ1 ∧ . . . ∧ Kanψn) ◮ ϕ → xKa1 . . . Kanψ ◮ ϕ → KbxKaψ ◮ ϕ → Kbx(Ka1Kbψ1∧. . .∧KanKbψn∧Ka1 ˆ

Kbχ1∧. . .∧Kan ˆ Kbχn)

slide-99
SLIDE 99

Conclusion

Applications to description logics

◮ Baader, F., Fern´

andez Gil, O., Morawska, B.: Hybrid unification in the description logic EL. In Fontaine, P ., Ringeissein, C., Schmidt, R. (editors): Frontiers of Combining Systems. Springer (2013) 295–310.

◮ Baader, F., Morawska, B.: Unification in the description

logic EL. In Treinen, R. (editor): Rewriting Techniques and

  • Applications. Springer (2009) 350–364.

◮ Baader, F., Narendran, P.: Unification of concept terms in

description logics. Journal of Symbolic Computation 31 (2001) 277–305.

slide-100
SLIDE 100

Conclusion

Applications to epistemic logics and temporal logics

◮ Babenyshev, S., Rybakov, V.: Unification in linear

temporal logic LTL. Annals of Pure and Applied Logic 162 (2011) 991–1000.

◮ Rybakov, V.: Logical consecutions in discrete linear

temporal logic. The Journal of Symbolic Logic 70 (2005) 1137–1149.

◮ Rybakov, V.: Multi-modal and temporal logics with

universal formula — reduction of admissibility to validity and unification. Journal of Logic and Computation 18 (2008) 509–519.

slide-101
SLIDE 101

Conclusion

Admissibility and unification in other non-classical logics

◮ Cintula, P., Metcalfe, G.: Structural completeness in fuzzy

  • logics. Notre Dame Journal of Formal Logic 50 (2009)

153–182.

◮ Dzik, W.: Unification of some substructural logics of

BL-algebras and hoops. Reports on Mathematical Logic 43 (2008) 73–3.

◮ Jer´

abek, E.: Admissible rules of Łukasiewicz logic. Journal

  • f Logic and Computation 20 (2010) 425–447.

◮ Odintsov, S., Rybakov, V.: Unification and admissible

rules for paraconsistent minimal Johanssons logic J and positive intuitionistic logic IPC+. Annals of Pure and Applied logic 164 (2013) 771–784.

slide-102
SLIDE 102

Conclusion

Proof-theoretic approaches

◮ Iemhoff, R.: On the admissible rules of Intuitionistic

Propositional Logic. The Journal of Symbolic Logic 66 (2001) 281–294.

◮ Iemhoff, R.: A syntactic approach to unification in

transitive reflexive modal logics. Notre Dame Journal of Formal Logic 57 (2016) 233–247.

◮ Iemhoff, R., Metcalfe, G.: Hypersequent systems for the

admissible rules of modal and intermediate logics. In Artemov, S., Nerode, A. (editors): Logical Foundations of Computer Science. Springer (2009) 230–245.

◮ Iemhoff, R., Metcalfe, G.: Proof theory for admissible

  • rules. Annals of Pure and Applied Logic 159 (2009)

171–186.

slide-103
SLIDE 103

Conclusion

Decidability/complexity and proof procedures

◮ Babenyshev, S., Rybakov, V., Schmidt, R., Tishkovsky,

D.: A tableau method for checking rule admissibility in S4. Electronic Notes in Theoretical Computer Science 262 (2010) 17–32.

◮ Cintula, P., Metcalfe, G.: Admissible rules in the

implication-negation fragment of intuitionistic logic. Annals

  • f Pure and Applied Logic 162 (2010) 162–171.

◮ Ghilardi, S.: A resolution/tableaux algorithm for projective

approximations in IPC. Logic Journal of the IGPL 10 (2002) 229–243.

◮ Je˘

r´ abek, E.: Complexity of admissible rules. Archive for Mathematical Logic 46 (2007) 73–92.

slide-104
SLIDE 104

Conclusion

K-unification

◮ Je˘

r´ abek, E.: Blending margins: the modal logic K has nullary unification type. Journal of Logic and Computation 25 (2015) 1231–1240.

◮ Wolter, F., Zakharyaschev, M.: Undecidability of the

unification and admissibility problems for modal and description logics. ACM Transactions on Computational Logic 9 (2008) 25:1–25:20.

slide-105
SLIDE 105

Conclusion

Some open problems

◮ Decidability of

◮ parameter-free unification in modal logic K, KB ? ◮ unification with parameters in modal logics KD, KDB ? ◮ unification with parameters in modal logics KT, KTB ? ◮ unification with parameters in modal logics Alt1, Alt2 ? ◮ unification in implication fragments ?

◮ Type of

◮ KB, KD, KDB, KT, KTB for parameter-free unification ? ◮ S5 ⊗ S5 and other fusions of modal logics ? ◮ S4.2 × S4.2 and other products of modal logics ? ◮ K + k⊥ and other locally tabular modal logics ? ◮ unification in implication fragments ?

slide-106
SLIDE 106

Thank you