Generalized P lonka Sums Marek Zawadowski University of Warsaw - - PowerPoint PPT Presentation

generalized p lonka sums
SMART_READER_LITE
LIVE PREVIEW

Generalized P lonka Sums Marek Zawadowski University of Warsaw - - PowerPoint PPT Presentation

Generalized P lonka Sums Marek Zawadowski University of Warsaw 88th Workshop on General Algebra June 20, 2014 Marek Zawadowski Generalized P lonka Sums 1 / 22 Operations on algebras Operations on algebras induced by operations on


slide-1
SLIDE 1

Generalized P lonka Sums

Marek Zawadowski

University of Warsaw

88th Workshop on General Algebra June 20, 2014

Marek Zawadowski Generalized P lonka Sums 1 / 22

slide-2
SLIDE 2

Operations on algebras

Operations on algebras induced by operations on universes T - equational theory Alg(T) - category of algebras of the theory T

SetC Set

O Alg(T)C Alg(T)

¯ O

UC

U

1 Examples: products |A × B|

∼ =

− → |A| × |B|, limits, filtered colimits, reduced products, ultraproduct

Marek Zawadowski Generalized P lonka Sums 2 / 22

slide-3
SLIDE 3

Operations on algebras

Operations on algebras induced by operations on universes T - equational theory Alg(T) - category of algebras of the theory T

Set × Set Set

× Alg(T) × Alg(T) Alg(T)

×

U × U

U

1 Examples: products |A × B|

∼ =

− → |A| × |B|, limits, filtered colimits, reduced products, ultraproduct

Marek Zawadowski Generalized P lonka Sums 2 / 22

slide-4
SLIDE 4

Operations on algebras

Operations on algebras induced by operations on universes T - equational theory Alg(T) - category of algebras of the theory T

SetC Set

O Alg(T)C Alg(T)

¯ O

UC

U

1 Examples: products |A × B|

∼ =

− → |A| × |B|, limits, filtered colimits, reduced products, ultraproduct

2 No-Examples: coproducts...but Marek Zawadowski Generalized P lonka Sums 2 / 22

slide-5
SLIDE 5

Operations on algebras

Operations on algebras induced by operations on universes T - equational theory Alg(T) - category of algebras of the theory T

SetC Set

O Alg(T)C Alg(T)

¯ O

UC

U

1 Examples: products |A × B|

∼ =

− → |A| × |B|, limits, filtered colimits, reduced products, ultraproduct

2 No-Examples: coproducts...but 3 P

lonka sums...

Marek Zawadowski Generalized P lonka Sums 2 / 22

slide-6
SLIDE 6

P lonka sums

L - sup-semilattice (∨,⊥) D : L − → Alg(T) - functor - L-diagram of T-algebras P D - P lonka sum of D universe | P D| =

l∈L |D(l)|

Marek Zawadowski Generalized P lonka Sums 3 / 22

slide-7
SLIDE 7

P lonka sums

L - sup-semilattice (∨,⊥) D : L − → Alg(T) - functor - L-diagram of T-algebras P D - P lonka sum of D universe | P D| =

l∈L |D(l)|

Dl1 Dln ... ... l1 ln

Marek Zawadowski Generalized P lonka Sums 3 / 22

slide-8
SLIDE 8

P lonka sums

L - sup-semilattice (∨,⊥) D : L − → Alg(T) - functor - L-diagram of T-algebras P D - P lonka sum of D universe | P D| =

l∈L |D(l)|

Dl1 Dln ... ... l1 ln a1 ∈ an ∈

Marek Zawadowski Generalized P lonka Sums 3 / 22

slide-9
SLIDE 9

P lonka sums

L - sup-semilattice (∨,⊥) D : L − → Alg(T) - functor - L-diagram of T-algebras P D - P lonka sum of D universe | P D| =

l∈L |D(l)|

Dl1 Dln ... ... l1 ln a1 ∈ an ∈ l = l1 ∨ . . . ∨ ln Dl

Marek Zawadowski Generalized P lonka Sums 3 / 22

slide-10
SLIDE 10

P lonka sums

L - sup-semilattice (∨,⊥) D : L − → Alg(T) - functor - L-diagram of T-algebras P D - P lonka sum of D universe | P D| =

l∈L |D(l)|

Dl1 Dln ... ... l1 ln a1 ∈ an ∈ l = l1 ∨ . . . ∨ ln Dl

D(li ≤ l) : Dli → Dl

Marek Zawadowski Generalized P lonka Sums 3 / 22

slide-11
SLIDE 11

P lonka sums

L - sup-semilattice (∨,⊥) D : L − → Alg(T) - functor - L-diagram of T-algebras P D - P lonka sum of D universe | P D| =

l∈L |D(l)|

Dl1 Dln ... ... l1 ln a1 ∈ an ∈ l = l1 ∨ . . . ∨ ln Dl b1, . . . , bn ∈ D(l1 ≤ l)

D(li ≤ l) : Dli → Dl bi := D(li ≤ l)(ai) ∈ Dl

Marek Zawadowski Generalized P lonka Sums 3 / 22

slide-12
SLIDE 12

P lonka sums

L - sup-semilattice (∨,⊥) D : L − → Alg(T) - functor - L-diagram of T-algebras P D - P lonka sum of D universe | P D| =

l∈L |D(l)|

Dl1 Dln ... ... l1 ln a1 ∈ an ∈ l = l1 ∨ . . . ∨ ln Dl b1, . . . , bn ∈ D(l1 ≤ l)

b := f Dl(b1, . . . , bn)

D(li ≤ l) : Dli → Dl bi := D(li ≤ l)(ai) ∈ Dl

Marek Zawadowski Generalized P lonka Sums 3 / 22

slide-13
SLIDE 13

P lonka sums

L - sup-semilattice (∨,⊥) D : L − → Alg(T) - functor - L-diagram of T-algebras P D - P lonka sum of D universe | P D| =

l∈L |D(l)|

Dl1 Dln ... ... l1 ln a1 ∈ an ∈ l = l1 ∨ . . . ∨ ln Dl b1, . . . , bn ∈ D(l1 ≤ l)

b := f Dl(b1, . . . , bn)

D(li ≤ l) : Dli → Dl bi := D(li ≤ l)(ai) ∈ Dl f

P D(a1, . . . , an) := b

Marek Zawadowski Generalized P lonka Sums 3 / 22

slide-14
SLIDE 14

P lonka sums

Theorem [J. P lonka 1967] If T is regular, then P lonka sum of T algebras is a T-algebra.

Marek Zawadowski Generalized P lonka Sums 4 / 22

slide-15
SLIDE 15

A categorist’s look at P lonka sums

Marek Zawadowski Generalized P lonka Sums 5 / 22

slide-16
SLIDE 16

A categorist’s look at P lonka sums

Why regular theories?

Marek Zawadowski Generalized P lonka Sums 5 / 22

slide-17
SLIDE 17

A categorist’s look at P lonka sums

Why regular theories? Why sup-semilattices?

Marek Zawadowski Generalized P lonka Sums 5 / 22

slide-18
SLIDE 18

A categorist’s look at P lonka sums

Why regular theories? Why sup-semilattices?

1 The theory of sup-semilattices is the terminal object in the

category of regular theories, i.e. there is a unique regular interpretation from any regular theory to the theory of sup-semilattices. We can take any regular interpretation I : R − → T between regular theories instead!

Marek Zawadowski Generalized P lonka Sums 5 / 22

slide-19
SLIDE 19

A categorist’s look at P lonka sums

Why regular theories? Why sup-semilattices?

1 The theory of sup-semilattices is the terminal object in the

category of regular theories, i.e. there is a unique regular interpretation from any regular theory to the theory of sup-semilattices. We can take any regular interpretation I : R − → T between regular theories instead!

2 Any T-algebra A gives rise to a P

lonka sum on the category

  • f algebras Alg(R) with the arity being the category of regular

polynomial over A. Any sup-semilattice is a posetal reflection

  • f its category of regular polynomials.

Marek Zawadowski Generalized P lonka Sums 5 / 22

slide-20
SLIDE 20

A categorist’s look at P lonka sums

Why regular theories? Why sup-semilattices?

1 The theory of sup-semilattices is the terminal object in the

category of regular theories, i.e. there is a unique regular interpretation from any regular theory to the theory of sup-semilattices. We can take any regular interpretation I : R − → T between regular theories instead!

2 Any T-algebra A gives rise to a P

lonka sum on the category

  • f algebras Alg(R) with the arity being the category of regular

polynomial over A. Any sup-semilattice is a posetal reflection

  • f its category of regular polynomials.

3 As P

lonka sum is induced by an operation on universes of algebras, it is given by a morphism of monads. This allows us for some simplifications: to consider free algebras only and move between algebras over different categories (the rest will be taken care off by ‘abstract nonsense’).

Marek Zawadowski Generalized P lonka Sums 5 / 22

slide-21
SLIDE 21

Plan of the talk

Plan

1 The category of regular equational theories 2 Monads and their algebras 3 The category of semi-analytic monads 4 More on morphisms of monads 5 Category of regular polynomials over an algebra 6 Morphism of monads that induce (generalized) P

lonka sums

7 Examples Marek Zawadowski Generalized P lonka Sums 6 / 22

slide-22
SLIDE 22

Regular equational theories

L - signature

Marek Zawadowski Generalized P lonka Sums 7 / 22

slide-23
SLIDE 23

Regular equational theories

L - signature

  • xn = x1, . . . , xn - context is

Marek Zawadowski Generalized P lonka Sums 7 / 22

slide-24
SLIDE 24

Regular equational theories

L - signature

  • xn = x1, . . . , xn - context is

A regular term in context t : xn is a term such that variables that occurs in t are exactly xn;

Marek Zawadowski Generalized P lonka Sums 7 / 22

slide-25
SLIDE 25

Regular equational theories

L - signature

  • xn = x1, . . . , xn - context is

A regular term in context t : xn is a term such that variables that occurs in t are exactly xn; A regular equation in context s = t : xn if both s : xn and t : xn are regular terms in context

Marek Zawadowski Generalized P lonka Sums 7 / 22

slide-26
SLIDE 26

Regular equational theories

L - signature

  • xn = x1, . . . , xn - context is

A regular term in context t : xn is a term such that variables that occurs in t are exactly xn; A regular equation in context s = t : xn if both s : xn and t : xn are regular terms in context T = L, A is a regular equational theory, if A is a set of regular equations in contexts over signature L.

Marek Zawadowski Generalized P lonka Sums 7 / 22

slide-27
SLIDE 27

Regular equational theories

L - signature

  • xn = x1, . . . , xn - context is

A regular term in context t : xn is a term such that variables that occurs in t are exactly xn; A regular equation in context s = t : xn if both s : xn and t : xn are regular terms in context T = L, A is a regular equational theory, if A is a set of regular equations in contexts over signature L. A regular interpretation of regular equational theories I : T → T′ sends n-ary symbols f in L to regular terms in contexts I(f ) : xn in T′ so that for any equation s = t : xn in T we have T ⊢ I(s) = I(t) : xn

Marek Zawadowski Generalized P lonka Sums 7 / 22

slide-28
SLIDE 28

Monads (on Set)

For any equational theory T, the forgetful functor U has a left adjoint F, the free T-algebra functor: Alg(T) Set

U

F Thus F(X) is the free T-algebra on the set X. It can be constructed as the set of terms with additional constants from the set X divided by the provable equality in theory T. The unit of the adjunction ηX : X → UF(X) is an embedding of generators, and the counit εA : FU(A) → A is an evaluation of the terms over the universe of A in algebra A.

Marek Zawadowski Generalized P lonka Sums 8 / 22

slide-29
SLIDE 29

Monads (on Set)

definition

The composed endofunctor T = UF : Set → Set together with natural transformations η : 1Set → T and µ = UεF : T 2 → T make the diagrams T 2 T

µ T 3 T 2

µT

T (µ)

µ T T 2

ηT 1

❅ ❅ ❅ ❘

T

µ T

T (η) 1

  • commute. The left square expresses the fact that ’evaluation

commutes with substitution’. A monad on the category Set is an endofunctor T on Set together with two natural transformations η and µ as above making the above diagrams commute.

Marek Zawadowski Generalized P lonka Sums 9 / 22

slide-30
SLIDE 30

Algebras for monads

An algebra for a monad T or T -algebra is a set A (the universe of the algebra) together with a function (structure map) α : T (A) → A such that T (A)

✛ α

T 2(A)

µ

T (α) A T (A)

ηA 1A

❅ ❅ ❅ ❘

A

α commutes.

Marek Zawadowski Generalized P lonka Sums 10 / 22

slide-31
SLIDE 31

Algebras for monads

An algebra for a monad T or T -algebra is a set A (the universe of the algebra) together with a function (structure map) α : T (A) → A such that T (A)

✛ α

T 2(A)

µ

T (α) A T (A)

ηA 1A

❅ ❅ ❅ ❘

A

α commutes. A morphism of T -algebras h : (A, α) → (A′, α′) is a function h : A → A′ compatible with the structure maps.

Marek Zawadowski Generalized P lonka Sums 10 / 22

slide-32
SLIDE 32

Algebras for monads

An algebra for a monad T or T -algebra is a set A (the universe of the algebra) together with a function (structure map) α : T (A) → A such that T (A)

✛ α

T 2(A)

µ

T (α) A T (A)

ηA 1A

❅ ❅ ❅ ❘

A

α commutes. A morphism of T -algebras h : (A, α) → (A′, α′) is a function h : A → A′ compatible with the structure maps. So we have the categories of algebras for monad Alg(T ) as well.

Marek Zawadowski Generalized P lonka Sums 10 / 22

slide-33
SLIDE 33

Semi-analytic monads

The monads (R, η, µ) arising from regular equational theories T are more special then arbitrary monads on Set. They are characterized by some additional abstract conditions (finitary, preserves pullbacks along monomorphisms). They have much more specific presentations. There is a functor R : S → Set where S of finite sets n = {1, . . . , n} and surjections to Set (R(n) is the set of regular terms in context xn divided by provable equations), for set X, R(X) =

  • n∈ω

X n

  • ⊗n Rn

where X n

  • is the set of monomorphisms from n to X

⊗n is the tensor over symmetric group Sn.

Marek Zawadowski Generalized P lonka Sums 11 / 22

slide-34
SLIDE 34

The category of semi-analytic monads

A morphism of monads τ : (T , η, µ) → (T ′, η′, µ′) is a natural transformation τ : T → T ′ compatible with η’s and µ’s, i.e. the diagram T ′2

µ′ T 2

✛ µ ❄

τ ∗ τ 1Set T

η η′

❅ ❅ ❅ ❘

T ′

τ commutes. It induces a functor between categories of algebras: Alg(τ) : Alg(T ′) − → Alg(T ) (A, α : T ′(A) → A) → (A, α ◦ τA : T (A) → A) A morphism of semi-analytic monads τ is a morphism of monads such that naturality squares for monomorphisms are pullbacks.

Marek Zawadowski Generalized P lonka Sums 12 / 22

slide-35
SLIDE 35

Regular theories vs semi-analytic monads

Theorem (S. Szawiel, MZ) The category SanMnd of semi-analytic monads on Set is equivalent to the category of regular theories RegET. This correspondence respects categories of algebras.

Marek Zawadowski Generalized P lonka Sums 13 / 22

slide-36
SLIDE 36

Moving between algebras over different categories

more on morphisms of monads

We can consider morphism between monads defined on different

  • categories. They induce functors between categories of algebras.

C C′

K

(K, τ)

✝ ✆ ✻

(T ′, η′, µ′) (T , η, µ)

✝ ✆ ✻

1 K : C′ → C a functor 2 τ : T K → KT ′ a natural transformation 3 compatible with η’s and µ’s, i.e. the diagram

KT ′2

K(µ′) T 2K

µK

τ ∗ τ K T K

ηK K(η′)

❅ ❅ ❅ ❘

KT ′

τ

commutes.

Marek Zawadowski Generalized P lonka Sums 14 / 22

slide-37
SLIDE 37

Moving between algebras over different categories

more on morphisms of monads

They induce functors between categories of algebras: Alg(τ) : Alg(T ′) − → Alg(T ) is given by K(A, α : T ′(A) → A) = (K(A), T K(A)

τA

− → KT ′(A)

K(α)

− → K(A))

Marek Zawadowski Generalized P lonka Sums 15 / 22

slide-38
SLIDE 38

Lift of a monad to the category of diagrams

If (T , η, µ) is a monad on Set and C is a small category, then we have a monad ( ˆ T , ˆ η, ˆ µ) on SetC, the lift of the monad T to SetC. It is defined by composition, for a functor F : C → Set: ˆ T (F) = T ◦F, ˆ ηF = ηF : F → T ◦F, ˆ µF = µF : T 2◦F → T ◦F where ˆ ηF is the component of the natural transformation ˆ η at a functor F and ηF is the wiskering of the natural transformation η : 1Set → T along the functor F; the same applies to the definition of ˆ µ. The category of algebras for the lifted monads Alg( ˆ T ) is equivalent to the category of C-diagrams of algebras for T , i.e. Alg(T )C.

Marek Zawadowski Generalized P lonka Sums 16 / 22

slide-39
SLIDE 39

Category of regular polynomials over an algebra

Let T = (T , η, µ) be a semi-analytic monad, T : S → Set the coefficient functor for T so that for any set X T (X) =

  • n∈ω

X n

  • ⊗n Tn

(A, α : T (A) → A) and T -algebra. We define a category A of regular polynomials over T -algebra A. The objects of A are elements of A. A morphism in A is an equivalence class of triples [ a, i, r]∼ : a(i) → α([ a, r]∼) where a : n → A is an injection, i ∈ n, r ∈ Tn, for some n ∈ ω. Note that [ a, r]∼ is an element of T (A). We identify triples

  • a ◦ σ, i, r ∼

a, σ(i), T(σ)(r) where σ ∈ Sn.

Marek Zawadowski Generalized P lonka Sums 17 / 22

slide-40
SLIDE 40

Generalized P lonka sum

Setup

Generalized P lonka sum π : R → T a morphism of semi-analytic monads, defined by a natural transformation π : R → T in SetS (A, α) be a T -algebra A the category of regular polynomials over (A, α) ˆ R is the lift of the monad R to the category SetA We shall define a morphism of monads (

  • A, λ) : R → ˆ

R that gives rise to an operation Alg(R)A ∼ = Alg( ˆ R) − → Alg(R)

Marek Zawadowski Generalized P lonka Sums 18 / 22

slide-41
SLIDE 41

Generalized P lonka sum

Let F : A → Alg(R) be a functor, and F : A → Set the composition of F with the forgetful functor. We shall define the component λF : R(

  • a∈A

F(a)) − →

  • a∈A

R(F(a))

  • f λ

λ : R ◦

  • A −

  • A ◦ ˆ

R [ x, r]∼ ∈ R(

a∈A F(a))

  • x : (n] →

a∈A F(a) is an injection

r ∈ Rn p :

a∈A F(a) → A the projection from the coproduct to the

index set

Marek Zawadowski Generalized P lonka Sums 19 / 22

slide-42
SLIDE 42

Generalized P lonka sum

definition

(m] (n]

✛ ✛

s A

✻ ✻

  • a

(k]

✲ ✲

g F(b)

✻ ✻

  • z
  • a∈A F(a)

✛ p

  • i∈(n] F(

a(s(i)))

a(s(i))]i∈(n]

[F(ψi)]i∈(n]

❅ ❅ ❅ ❅ ■

  • x
  • x′
  • a ◦ s a surjection-injection factorization of p ◦

x κa : F(a) →

a∈A F(a) is the injection into coproduct

  • x′ - the unique making the triangle in the middle commute

We put b = α([ a, πm(R(s)(r))]) and, for i ∈ (n], we have a morphism ψi = [ a, s(i), πm(R(s)(r))] : a(s(i)) − → b in the category A.

Marek Zawadowski Generalized P lonka Sums 20 / 22

slide-43
SLIDE 43

Generalized P lonka sum

Finally, we put λR

F ([

x, r]∼) = [ z, R(g)(r)]∼ Theorem (

A, λR) : R → ˆ

R is a lax morphism of monads.

Marek Zawadowski Generalized P lonka Sums 21 / 22

slide-44
SLIDE 44

Examples

Examples

1 Identity interpretation 1 : R → R of a semi-analytic monad

  • R. P

lonka sum of identity interpretation over an R-algebra A

  • f a constant diagram F equal to an R-algebra B is the

product A × B.

2 The usual P

lonka sum comes from the unique semi-analytic interpretation of a R → S, of any semi-analytic monad in the monda for sup-semilattices.

3 More sophisticated examples. Let 2S be the monad

corresponding to the theory of two theories of sup-semilattices taken together. Let R be the monad arising from a regular theory that is a ’sum’ of two regular equational theories (having nothing to do one with the other). Then we have morphism of semi-analytic monads R → 2S such that the two parts of R are interpreted in different parts of 2S.

Marek Zawadowski Generalized P lonka Sums 22 / 22