completeness for cartesian bicategories
play

Completeness for Cartesian bicategories Relational algebra with - PowerPoint PPT Presentation

Completeness for Cartesian bicategories Relational algebra with string diagrams Filippo Bonchi, Jens Seeber , Pawe l Soboci nski IMT School for Advanced Studies Lucca Glasgow - 18 th December, 2018 Contents 1 Cartesian bicategories 2


  1. The syntactic Frobenius theory Signature Σ, each σ ∈ Σ equipped with arity and coarity, σ : n → m . Freely generated (syntactic) Cartesian bicategory CB Σ has objects N and morphisms . . . . S 1 . . � � � � � . . . Mor( CB Σ ) ::= ǫ . . . � � � � S 1 S 2 � . . . . . � � � � � . . S 2 . . � � � � . . . . � � � � σ . . � � � � modulo the laws of Cartesian bicategories. A model M : CB Σ → Rel consists of • a set V = M (1)

  2. The syntactic Frobenius theory Signature Σ, each σ ∈ Σ equipped with arity and coarity, σ : n → m . Freely generated (syntactic) Cartesian bicategory CB Σ has objects N and morphisms . . . . S 1 . . � � � � � . . . Mor( CB Σ ) ::= ǫ . . . � � � � S 1 S 2 � . . . . . � � � � � . . S 2 . . � � � � . . . . � � � � σ . . � � � � modulo the laws of Cartesian bicategories. A model M : CB Σ → Rel consists of • a set V = M (1) • relations M ( σ ) ⊆ V n × V m for σ ∈ Σ, σ : n → m

  3. Example Frobenius theories Example • ≤ R

  4. Example Frobenius theories Example • ensures that R is reflexive ≤ R

  5. Example Frobenius theories Example • ensures that R is reflexive ≤ R • ≤ R R R

  6. Example Frobenius theories Example • ensures that R is reflexive ≤ R • ensures that R is transitive ≤ R R R

  7. Example Frobenius theories Example • ensures that R is reflexive ≤ R • ensures that R is transitive ≤ R R R • R and ≤ ≤ R R R

  8. Example Frobenius theories Example • ensures that R is reflexive ≤ R • ensures that R is transitive ≤ R R R • R and ≤ ≤ R R R ensure that R is a function.

  9. Example Frobenius theories Example • ensures that R is reflexive ≤ R • ensures that R is transitive ≤ R R R • R and ≤ ≤ R R R ensure that R is a function. • ≤

  10. Example Frobenius theories Example • ensures that R is reflexive ≤ R • ensures that R is transitive ≤ R R R • R and ≤ ≤ R R R ensure that R is a function. • ensures that the underlying set is nonempty. ≤

  11. Presentations Definition Fix a signature Σ and let E be a set of (well-typed) inequalities of morphisms in CB Σ .

  12. Presentations Definition Fix a signature Σ and let E be a set of (well-typed) inequalities of morphisms in CB Σ . The Frobenius theory CB Σ /E has the morphisms of CB Σ taken modulo E .

  13. Presentations Definition Fix a signature Σ and let E be a set of (well-typed) inequalities of morphisms in CB Σ . The Frobenius theory CB Σ /E has the morphisms of CB Σ taken modulo E . Lemma A model of CB Σ /E is the same thing as a model of CB Σ satisfying E .

  14. Presentations Definition Fix a signature Σ and let E be a set of (well-typed) inequalities of morphisms in CB Σ . The Frobenius theory CB Σ /E has the morphisms of CB Σ taken modulo E . Lemma A model of CB Σ /E is the same thing as a model of CB Σ satisfying E . Lemma Every Frobenius theory is of the shape CB Σ /E for some Σ , E .

  15. Contents 1 Cartesian bicategories 2 Frobenius theories 3 Completeness

  16. Σ-structures • A model M : CB Σ → Rel consists of • a set V = M (1) • relations M ( σ ) ⊆ V n × V m for σ ∈ Σ, σ : n → m

  17. Σ-structures • A model M : CB Σ → Rel consists of • a set V = M (1) • relations M ( σ ) ⊆ V n × V m for σ ∈ Σ, σ : n → m • In model theory, these are called Σ-structures.

  18. Σ-structures • A model M : CB Σ → Rel consists of • a set V = M (1) • relations M ( σ ) ⊆ V n × V m for σ ∈ Σ, σ : n → m • In model theory, these are called Σ-structures. • A morphism between Σ-structures is a function between the underlying sets respecting the relations.

  19. Σ-structures • A model M : CB Σ → Rel consists of • a set V = M (1) • relations M ( σ ) ⊆ V n × V m for σ ∈ Σ, σ : n → m • In model theory, these are called Σ-structures. • A morphism between Σ-structures is a function between the underlying sets respecting the relations. • One can view a set as a Σ-structure – with empty relations

  20. Σ-structures • A model M : CB Σ → Rel consists of • a set V = M (1) • relations M ( σ ) ⊆ V n × V m for σ ∈ Σ, σ : n → m • In model theory, these are called Σ-structures. • A morphism between Σ-structures is a function between the underlying sets respecting the relations. • One can view a set as a Σ-structure – with empty relations • For S a Σ-structure, a morphism n → S is an n -tuple in S

  21. Universal models Example We can translate a morphism R : n → m in CB Σ to a finite ι R ω R model U R with n − → U R ← − − m (called universal model).

  22. Universal models Example We can translate a morphism R : n → m in CB Σ to a finite ι R ω R model U R with n − → U R ← − − m (called universal model). σ σ τ

  23. Universal models Example We can translate a morphism R : n → m in CB Σ to a finite ι R ω R model U R with n − → U R ← − − m (called universal model). z σ y σ x τ

  24. Universal models Example We can translate a morphism R : n → m in CB Σ to a finite ι R ω R model U R with n − → U R ← − − m (called universal model). z σ y σ x τ U R (1) = { x, y, z }

  25. Universal models Example We can translate a morphism R : n → m in CB Σ to a finite ι R ω R model U R with n − → U R ← − − m (called universal model). z σ y σ x τ U R (1) = { x, y, z } U R ( σ ) = { ( x, y ) , ( y, z ) } , U R ( τ ) = { x }

  26. Universal models Example We can translate a morphism R : n → m in CB Σ to a finite ι R ω R model U R with n − → U R ← − − m (called universal model). z σ y σ x τ U R (1) = { x, y, z } U R ( σ ) = { ( x, y ) , ( y, z ) } , U R ( τ ) = { x } ι R = x, ω R = y

  27. Connection with Completeness Theorem (SYCO 1) ι R ω R The assignment of R : n → m to n − → U R ← − − m is a bijection

  28. Connection with Completeness Theorem (SYCO 1) ι R ω R The assignment of R : n → m to n − → U R ← − − m is a bijection between morphisms in CB Σ and discrete cospans of finite Σ -structures.

  29. Connection with Completeness Theorem (SYCO 1) ι R ω R The assignment of R : n → m to n − → U R ← − − m is a bijection between morphisms in CB Σ and discrete cospans of finite Σ -structures. S ≤ R if and only if there is α : U R → U S such that U R ι R ω R n m α ι S ω S U S

  30. Connection with Completeness Theorem (SYCO 1) ι R ω R The assignment of R : n → m to n − → U R ← − − m is a bijection between morphisms in CB Σ and discrete cospans of finite Σ -structures. S ≤ R if and only if there is α : U R → U S such that U R ι R ω R n m α ι S ω S U S Connects semantics to syntax.

  31. The ( · ) E construction Idea: Saturate a Σ-structure with respect to the axioms E .

  32. The ( · ) E construction Idea: Saturate a Σ-structure with respect to the axioms E . Theorem There is a functor ( · ) E : Mod CB Σ → Mod CB Σ with a natural transformation ζ S : S → S E with the following property:

  33. The ( · ) E construction Idea: Saturate a Σ-structure with respect to the axioms E . Theorem There is a functor ( · ) E : Mod CB Σ → Mod CB Σ with a natural transformation ζ S : S → S E with the following property: If A ≤ B is an axiom in E , and ( x, y ) ∈ S ( A ) then ( ζ ( x ) , ζ ( y )) ∈ S E ( B ) .

  34. The ( · ) E construction Idea: Saturate a Σ-structure with respect to the axioms E . Theorem There is a functor ( · ) E : Mod CB Σ → Mod CB Σ with a natural transformation ζ S : S → S E with the following property: If A ≤ B is an axiom in E , and ( x, y ) ∈ S ( A ) then ( ζ ( x ) , ζ ( y )) ∈ S E ( B ) . Definition An algebra for the pointed endofunctor ( · ) E is a Σ-structure S with a morphism α : S E → S such that α ◦ ζ S = id S

  35. The ( · ) E construction Idea: Saturate a Σ-structure with respect to the axioms E . Theorem There is a functor ( · ) E : Mod CB Σ → Mod CB Σ with a natural transformation ζ S : S → S E with the following property: If A ≤ B is an axiom in E , and ( x, y ) ∈ S ( A ) then ( ζ ( x ) , ζ ( y )) ∈ S E ( B ) . Definition An algebra for the pointed endofunctor ( · ) E is a Σ-structure S with a morphism α : S E → S such that α ◦ ζ S = id S Lemma ( · ) E -algebras are models for CB Σ /E .

  36. Example � � Take Σ = ∅ , E = ≤

  37. Example � � Take Σ = ∅ , E = , Mod CB Σ /E is the category of ≤ non-empty sets.

  38. Example � � Take Σ = ∅ , E = , Mod CB Σ /E is the category of ≤ non-empty sets. • S E = S + 1

  39. Example � � Take Σ = ∅ , E = , Mod CB Σ /E is the category of ≤ non-empty sets. • S E = S + 1 • ( · ) E –algebras are pointed sets

  40. Example � � Take Σ = ∅ , E = , Mod CB Σ /E is the category of ≤ non-empty sets. • S E = S + 1 • ( · ) E –algebras are pointed sets The category of ( · ) E –algebras is better behaved than Mod CB Σ /E .

  41. The free algebra ( · ) E -Alg Mod CB Σ U Mod CB Σ /E

  42. The free algebra ( · ) E -Alg Mod CB Σ ⊥ U Mod CB Σ /E

  43. The free algebra ( · ) Eω ( · ) E -Alg Mod CB Σ ⊥ U Mod CB Σ /E

  44. The free algebra ( · ) Eω ( · ) E -Alg Mod CB Σ ⊥ U Mod CB Σ /E S S E S E 2 · · · S E ω

  45. The free algebra ( · ) Eω ( · ) E -Alg Mod CB Σ ⊥ U Mod CB Σ /E S S E S E 2 · · · S E ω

  46. From semantics to syntax Theorem S ≤ R in CB Σ if and only if there is α : U R → U S such that U R ι R ω R n m α ι S ω S U S

  47. From semantics to syntax Theorem S ≤ R in CB Σ /E if and only if there is α : ( U R ) E ω → ( U S ) E ω such that ( U R ) E ω n m α ( U S ) E ω

  48. From semantics to syntax Proof sketch. ( U R ) E ω n m α ( U S ) E ω

  49. From semantics to syntax Proof sketch. U R n m α ( U S ) E ω

  50. From semantics to syntax Proof sketch. U R n m α ( U S ) E ω • U R is compact

  51. From semantics to syntax Proof sketch. U R n m α ( U S ) E k • U R is compact

  52. From semantics to syntax Proof sketch. U R n m α ( U S ) E k • U R is compact • n → ( U S ) E i ← m correspond to string diagrams S i

  53. From semantics to syntax Proof sketch. U R n m α ( U S ) E k • U R is compact • n → ( U S ) E i ← m correspond to string diagrams S i • S k ≤ R in CB Σ , hence in CB Σ /E

  54. From semantics to syntax Proof sketch. U R n m α ( U S ) E k • U R is compact • n → ( U S ) E i ← m correspond to string diagrams S i • S k ≤ R in CB Σ , hence in CB Σ /E • S i +1 is obtained by blindly applying all axioms to S i

  55. From semantics to syntax Proof sketch. U R n m α ( U S ) E k • U R is compact • n → ( U S ) E i ← m correspond to string diagrams S i • S k ≤ R in CB Σ , hence in CB Σ /E • S i +1 is obtained by blindly applying all axioms to S i • S = S 0 ≤ S 1 ≤ S 2 ≤ · · · ≤ S k ≤ R in CB Σ /E

Recommend


More recommend