On the Essence and Initiality of Conflicts Guilherme Grochau Azzi 1 , - - PowerPoint PPT Presentation

on the essence and initiality of conflicts
SMART_READER_LITE
LIVE PREVIEW

On the Essence and Initiality of Conflicts Guilherme Grochau Azzi 1 , - - PowerPoint PPT Presentation

Introduction Characterization Properties Previous Work Initial Conflicts Conclusions On the Essence and Initiality of Conflicts Guilherme Grochau Azzi 1 , Andrea Corradini 2 and Leila Ribeiro 1 1 Instituto de Informtica Universidade Federal


slide-1
SLIDE 1

1/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

On the Essence and Initiality of Conflicts

Guilherme Grochau Azzi1, Andrea Corradini2 and Leila Ribeiro1

1Instituto de Informática

Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil

2Dipartimento di Informatica

Università di Pisa, Pisa, Italy

11th International Conference on Graph Transformation, June 2018

slide-2
SLIDE 2

2/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Parallel Independence

  • f Transformations

G H1 H1 H

ρ1 ρ2 ρ2 ρ1

slide-3
SLIDE 3

2/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Parallel Independence

  • f Transformations

G H1 H1 H

ρ1 ρ2 ρ2 ρ1

slide-4
SLIDE 4

2/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Parallel Independence

  • f Transformations

G H1 H1 H

ρ1 ρ2 ρ2 ρ1

Conflict

slide-5
SLIDE 5

3/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Motivation

  • Conflicts capture important information about behaviour
  • Enumerating potential conflicts has many applications
  • Critical pairs or initial conflicts
  • Understanding root causes is often important
slide-6
SLIDE 6

4/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: The dpo Approach

Rule: ρ = L

l

֋ K

r

֌ R Match: m : L ֌ G Transformation: G

ρ,m

=⇒ H L K R G D H

m l r k n

PO

g h

PO

slide-7
SLIDE 7

5/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

New Perspective

  • Previous work based on the standard condition for

parallel independence R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

n1 l1 r1 k1 m1 q12 m2 q21 k2 l2 r2 n2 g1 h1 g2 h2

slide-8
SLIDE 8

5/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

New Perspective

  • Previous work based on the standard condition for

parallel independence R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

n1 l1 r1 k1 m1 q12 m2 q21 k2 l2 r2 n2 g1 h1 g2 h2

  • Recently: essential condition for parallel independence

(Corradini et al. 2018)

  • Equivalent to standard condition
slide-9
SLIDE 9

5/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

New Perspective

  • Previous work based on the standard condition for

parallel independence R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

n1 l1 r1 k1 m1 q12 m2 q21 k2 l2 r2 n2 g1 h1 g2 h2

  • Recently: essential condition for parallel independence

(Corradini et al. 2018)

  • Equivalent to standard condition
  • Goal: review characterization of conflicts under new light
slide-10
SLIDE 10

6/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: Adhesive Categories

Subobjects behave like subsets I A B U X

a∩b a b a∪b

slide-11
SLIDE 11

6/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: Adhesive Categories

Subobjects behave like subsets Lemma (Lack and Sobocinski 2005) In adhesive categories, Sub(X) is distributive lattice I A B U X

a∩b a b a∪b

slide-12
SLIDE 12

6/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: Adhesive Categories

Subobjects behave like subsets Lemma (Lack and Sobocinski 2005) In adhesive categories, Sub(X) is distributive lattice Containment existence of mono I A B U X

a∩b a b a∪b

slide-13
SLIDE 13

6/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: Adhesive Categories

Subobjects behave like subsets Lemma (Lack and Sobocinski 2005) In adhesive categories, Sub(X) is distributive lattice Containment existence of mono Intersection pullback I A B U X

a∩b a b a∪b

slide-14
SLIDE 14

6/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: Adhesive Categories

Subobjects behave like subsets Lemma (Lack and Sobocinski 2005) In adhesive categories, Sub(X) is distributive lattice Containment existence of mono Intersection pullback I A B U X

a∩b a b a∪b

slide-15
SLIDE 15

6/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: Adhesive Categories

Subobjects behave like subsets Lemma (Lack and Sobocinski 2005) In adhesive categories, Sub(X) is distributive lattice Containment existence of mono Intersection pullback I A B U X

a∩b a b a∪b

slide-16
SLIDE 16

6/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: Adhesive Categories

Subobjects behave like subsets Lemma (Lack and Sobocinski 2005) In adhesive categories, Sub(X) is distributive lattice Containment existence of mono Intersection pullback Union pushout over intersection I A B U X

a∩b a b a∪b

slide-17
SLIDE 17

6/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: Adhesive Categories

Subobjects behave like subsets Lemma (Lack and Sobocinski 2005) In adhesive categories, Sub(X) is distributive lattice Containment existence of mono Intersection pullback Union pushout over intersection I A B U X

a∩b a b a∪b

slide-18
SLIDE 18

6/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: Adhesive Categories

Subobjects behave like subsets Lemma (Lack and Sobocinski 2005) In adhesive categories, Sub(X) is distributive lattice Containment existence of mono Intersection pullback Union pushout over intersection I A B U X

a∩b a b a∪b

slide-19
SLIDE 19

6/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: Adhesive Categories

Subobjects behave like subsets Lemma (Lack and Sobocinski 2005) In adhesive categories, Sub(X) is distributive lattice Containment existence of mono Intersection pullback Union pushout over intersection Top is X

slide-20
SLIDE 20

6/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: Adhesive Categories

Subobjects behave like subsets Lemma (Lack and Sobocinski 2005) In adhesive categories, Sub(X) is distributive lattice Containment existence of mono Intersection pullback Union pushout over intersection Top is X Bottom usually “empty”, if exists

slide-21
SLIDE 21

7/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: Set-Valued Functor Categories

  • Some results not proven for all adhesive categories
slide-22
SLIDE 22

7/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: Set-Valued Functor Categories

  • Some results not proven for all adhesive categories
  • We use categories et of functors → et with

natural transformations as arrows (essentially presheaves)

  • Generalizes graphs and graph structures

raph = et = V E

s t

slide-23
SLIDE 23

7/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Background: Set-Valued Functor Categories

  • Some results not proven for all adhesive categories
  • We use categories et of functors → et with

natural transformations as arrows (essentially presheaves)

  • Generalizes graphs and graph structures

raph = et = V E

s t

  • Limits, colimits, monos and epis are pointwise
  • Always adhesive
slide-24
SLIDE 24

8/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Outline

  • 1. Characterize conflict between transformations
  • 2. Useful properties of the characterization
  • 3. Compare with conflict reasons of Lambers, Ehrig, and

Orejas (2008)

  • 4. Relate to initial conflicts
slide-25
SLIDE 25

9/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Essential Condition of Parallel Independence

Corradini et al. (2018) H1

t1

⇐= G

t2

=⇒ H2 K1L2 L1L2 L1K2 R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

q12 p2 p1 q21 n1 l1 r1 k1 m1 m2 k2 l2 r2 n2 g1 h1 g2 h2

slide-26
SLIDE 26

9/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Essential Condition of Parallel Independence

Corradini et al. (2018) H1

t1

⇐= G

t2

=⇒ H2 K1L2 L1L2 L1K2 R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

q12 p2 p1 q21 n1 l1 r1 k1 m1 m2 k2 l2 r2 n2 g1 h1 g2 h2

slide-27
SLIDE 27

9/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Essential Condition of Parallel Independence

Corradini et al. (2018) H1

t1

⇐= G

t2

=⇒ H2 K1L2 L1L2 L1K2 R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

q12 p2 p1 q21 n1 l1 r1 k1 m1 m2 k2 l2 r2 n2 g1 h1 g2 h2

slide-28
SLIDE 28

9/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Essential Condition of Parallel Independence

Corradini et al. (2018) H1

t1

⇐= G

t2

=⇒ H2 K1L2 L1L2 L1K2 R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

q12 p2 p1 q21 n1 l1 r1 k1 m1 m2 k2 l2 r2 n2 g1 h1 g2 h2

  • Both morphisms iso ⇒ parallel independence
slide-29
SLIDE 29

9/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Essential Condition of Parallel Independence

Corradini et al. (2018) H1

t1

⇐= G

t2

=⇒ H2 K1L2 L1L2 L1K2 R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

q12 p2 p1 q21 n1 l1 r1 k1 m1 m2 k2 l2 r2 n2 g1 h1 g2 h2

  • Both morphisms iso ⇒ parallel independence
  • Either morphism not iso ⇒ conflict
slide-30
SLIDE 30

9/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Essential Condition of Parallel Independence

Corradini et al. (2018) H1

t1

⇐= G

t2

=⇒ H2 K1L2 L1L2 L1K2 R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

q12 p2 p1 q21 n1 l1 r1 k1 m1 m2 k2 l2 r2 n2 g1 h1 g2 h2

  • Both morphisms iso ⇒ parallel independence
  • Either morphism not iso ⇒ conflict
  • K1L2 → L1L2 not iso ⇒ t1 disables t2
slide-31
SLIDE 31

10/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Example: Conflict

slide-32
SLIDE 32

10/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Example: Conflict

slide-33
SLIDE 33

10/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Example: Conflict

slide-34
SLIDE 34

10/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Example: Conflict

slide-35
SLIDE 35

11/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Determining the Root Cause

  • Useful concept: initial pushout over f : X → Y

B X C Y

b f f c

  • “Categorical diff” for a morphism
slide-36
SLIDE 36

11/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Determining the Root Cause

  • Useful concept: initial pushout over f : X → Y

B X C Y

b f f c

  • “Categorical diff” for a morphism
  • Context c : C ֌ Y contains “modified stuff”
  • Boundary b : B ֌ C contains “points of contact”
slide-37
SLIDE 37

12/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Example: Initial Pushout

slide-38
SLIDE 38

12/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Example: Initial Pushout

slide-39
SLIDE 39

12/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Example: Initial Pushout

slide-40
SLIDE 40

13/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Conflict and Disabling Essences

Definition Given transformations (t1, t2) : H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2: B1 C1 K1L2 L1L2 R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

b1 c1 q12 p2 p1 n1 l1 r1 k1 m1 m2 k2 l2 r2 n2 g1 h1 g2 h2

slide-41
SLIDE 41

13/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Conflict and Disabling Essences

Definition Given transformations (t1, t2) : H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2: B1 C1 K1L2 L1L2 R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

b1 c1 q12 p2 p1 n1 l1 r1 k1 m1 m2 k2 l2 r2 n2 g1 h1 g2 h2

slide-42
SLIDE 42

13/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Conflict and Disabling Essences

Definition Given transformations (t1, t2) : H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2: B1 C1 K1L2 L1L2 R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

b1 c1 q12 p2 p1 n1 l1 r1 k1 m1 m2 k2 l2 r2 n2 g1 h1 g2 h2

slide-43
SLIDE 43

13/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Conflict and Disabling Essences

Definition Given transformations (t1, t2) : H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2:

  • Disabling essence for (t1, t2) is c1 ∈ Sub(L1L2)

B1 C1 K1L2 L1L2 R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

b1 c1 q12 p2 p1 n1 l1 r1 k1 m1 m2 k2 l2 r2 n2 g1 h1 g2 h2

slide-44
SLIDE 44

13/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Conflict and Disabling Essences

Definition Given transformations (t1, t2) : H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2:

  • Disabling essence for (t1, t2) is c1 ∈ Sub(L1L2)
  • Disabling essence for (t2, t1) is c2 ∈ Sub(L1L2)

B1 C1 K1L2 L1L2 R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

b1 c1 q12 p2 p1 n1 l1 r1 k1 m1 m2 k2 l2 r2 n2 g1 h1 g2 h2

slide-45
SLIDE 45

13/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Conflict and Disabling Essences

Definition Given transformations (t1, t2) : H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2:

  • Disabling essence for (t1, t2) is c1 ∈ Sub(L1L2)
  • Disabling essence for (t2, t1) is c2 ∈ Sub(L1L2)
  • Conflict essence for (t1, t2) is c = c1 ∪ c2

B1 C1 K1L2 L1L2 R1 K1 L1 L2 K2 R2 H1 D1 G D2 H2

b1 c1 q12 p2 p1 n1 l1 r1 k1 m1 m2 k2 l2 r2 n2 g1 h1 g2 h2

slide-46
SLIDE 46

14/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Example: Parallel Independence

slide-47
SLIDE 47

14/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Example: Parallel Independence

slide-48
SLIDE 48

14/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Example: Parallel Independence

slide-49
SLIDE 49

14/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Example: Parallel Independence

slide-50
SLIDE 50

14/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Example: Parallel Independence

slide-51
SLIDE 51

14/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Example: Parallel Independence

No conflict =⇒ no element caused a conflict

slide-52
SLIDE 52

15/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Empty Essences

Recall: bottom subobject generalizes “emptiness”

slide-53
SLIDE 53

15/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Empty Essences

Recall: bottom subobject generalizes “emptiness” Consider (t1, t2) : H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2

Theorem The conflict essence for (t1, t2) is ⊥ ∈ Sub(L1L2) if and only if t1 and t2 are parallel independent.

slide-54
SLIDE 54

16/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Extension

  • Same transformation in “larger context”

G H G H

f t f′ t

≡ L K R G D H G D H

m l k r n f d g h f′ g h

slide-55
SLIDE 55

16/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Extension

  • Same transformation in “larger context”

G H G H

f t f′ t

≡ L K R G D H G D H

m l k r n f d g h f′ g h

  • Lower pushouts ensure t behaves like t
slide-56
SLIDE 56

17/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Essence Inheritance

Theorem If extension diagrams below exist, (t1, t2) and (t1, t2) have the same disabling and conflict essences. H1 G H2 H1 G H2

f t1 t2 t1 t2

slide-57
SLIDE 57

17/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Essence Inheritance

Theorem If extension diagrams below exist, (t1, t2) and (t1, t2) have the same disabling and conflict essences. H1 G H2 H1 G H2

f t1 t2 t1 t2

C L1 C L2

p1◦c p2◦c

  • p1◦c

p2◦c

slide-58
SLIDE 58

17/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Essence Inheritance

Theorem If extension diagrams below exist, (t1, t2) and (t1, t2) have the same disabling and conflict essences. H1 G H2 H1 G H2

f t1 t2 t1 t2

C L1 C L2

p1◦c p2◦c

  • p1◦c

p2◦c

slide-59
SLIDE 59

17/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Essence Inheritance

In categories of set-valued functors (also graphs, typed graphs...) Theorem If extension diagrams below exist, (t1, t2) and (t1, t2) have the same disabling and conflict essences. H1 G H2 H1 G H2

f t1 t2 t1 t2

C L1 C L2

p1◦c p2◦c

  • p1◦c

p2◦c

slide-60
SLIDE 60

17/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Essence Inheritance

In categories of set-valued functors (also graphs, typed graphs...) Theorem If extension diagrams below exist, (t1, t2) and (t1, t2) have the same disabling and conflict essences. H1 G H2 H1 G H2

f t1 t2 t1 t2

C L1 C L2

p1◦c p2◦c

  • p1◦c

p2◦c

Conflicts are preserved and reflected by extension.

slide-61
SLIDE 61

18/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Disabling Reasons

Essences are not the first proposed characterization Given transformations (t1, t2) : H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2 Definition (Lambers, Ehrig, and Orejas 2008) The disabling reason L1 ← S1 → L2 for (t1, t2) Bl1 Cl1 S1 K1 L1 L2 G

l1 bl1 cl1

  • 1

s12 b∗ l1 m1 m2

slide-62
SLIDE 62

18/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Disabling Reasons

Essences are not the first proposed characterization Given transformations (t1, t2) : H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2 Definition (Lambers, Ehrig, and Orejas 2008) The disabling reason L1 ← S1 → L2 for (t1, t2) is obtained from the initial pushout over l1, Bl1 Cl1 S1 K1 L1 L2 G

l1 bl1 cl1

  • 1

s12 b∗ l1 m1 m2

slide-63
SLIDE 63

18/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Disabling Reasons

Essences are not the first proposed characterization Given transformations (t1, t2) : H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2 Definition (Lambers, Ehrig, and Orejas 2008) The disabling reason L1 ← S1 → L2 for (t1, t2) is obtained from the initial pushout over l1, then pullback of (m1 ◦ cl1, m2). Bl1 Cl1 S1 K1 L1 L2 G

l1 bl1 cl1

  • 1

s12 b∗ l1 m1 m2

slide-64
SLIDE 64

18/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Disabling Reasons

Essences are not the first proposed characterization Given transformations (t1, t2) : H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2 Definition (Lambers, Ehrig, and Orejas 2008) The disabling reason L1 ← S1 → L2 for (t1, t2) is obtained from the initial pushout over l1, then pullback of (m1 ◦ cl1, m2). Conflict condition: There is no b∗ making diagram commute. Bl1 Cl1 S1 K1 L1 L2 G

l1 bl1 cl1

  • 1

s12 b∗ l1 m1 m2

slide-65
SLIDE 65

18/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Disabling Reasons

Essences are not the first proposed characterization Given transformations (t1, t2) : H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2 Definition (Lambers, Ehrig, and Orejas 2008) The disabling reason L1 ← S1 → L2 for (t1, t2) is obtained from the initial pushout over l1, then pullback of (m1 ◦ cl1, m2). Conflict condition: There is no b∗ making diagram commute. Conflict reason is union of relevant disabling reasons. Bl1 Cl1 S1 K1 L1 L2 G

l1 bl1 cl1

  • 1

s12 b∗ l1 m1 m2

slide-66
SLIDE 66

19/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Comparing Reasons and Essences

  • Non-empty reasons exist even with parallel

independence

slide-67
SLIDE 67

19/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Comparing Reasons and Essences

  • Non-empty reasons exist even with parallel

independence

slide-68
SLIDE 68

19/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Comparing Reasons and Essences

  • Non-empty reasons exist even with parallel

independence

slide-69
SLIDE 69

19/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Comparing Reasons and Essences

  • Non-empty reasons exist even with parallel

independence

slide-70
SLIDE 70

19/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Comparing Reasons and Essences

  • Non-empty reasons exist even with parallel

independence

  • Isolated boundary nodes (Lambers, Born, et al. 2018)
slide-71
SLIDE 71

19/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Comparing Reasons and Essences

  • Non-empty reasons exist even with parallel

independence

  • Isolated boundary nodes (Lambers, Born, et al. 2018)
  • Inheritance also doesn’t hold
slide-72
SLIDE 72

20/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Essence ⊆ Reason

slide-73
SLIDE 73

20/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Essence ⊆ Reason

Remark Conflict reason determines s ∈ Sub(L1L2). L1 S L1L2 G L2

m1 s2 s1 s p1 p2 m2

slide-74
SLIDE 74

20/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Essence ⊆ Reason

Remark Conflict reason determines s ∈ Sub(L1L2). L1 S L1L2 G L2

m1 s2 s1 s p1 p2 m2

Theorem If c ∈ Sub(L1L2) is disabling essence and s ∈ Sub(L1L2) disabling reason, then c ⊆ s. The same holds if c is conflict essence and s conflict reason.

slide-75
SLIDE 75

21/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Initial Conflicts

  • We now understand individual conflicting

transformations

  • We want overview of potential conflicts for rules
slide-76
SLIDE 76

21/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Initial Conflicts

  • We now understand individual conflicting

transformations

  • We want overview of potential conflicts for rules
  • Lambers, Born, et al. (2018) proposed initial conflicts

(w.r.t extension) J1 I J2 H1 G H2 H1 G H2

slide-77
SLIDE 77

22/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Initial Conflicts

  • Initial conflicts are subset of critical pairs, often much

smaller!

slide-78
SLIDE 78

22/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Initial Conflicts

  • Initial conflicts are subset of critical pairs, often much

smaller!

  • Initial conflicts capture all conflicts ⇐⇒ every

transformation pair is extension of some initial transformation pair

slide-79
SLIDE 79

22/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Initial Conflicts

  • Initial conflicts are subset of critical pairs, often much

smaller!

  • Initial conflicts capture all conflicts ⇐⇒ every

transformation pair is extension of some initial transformation pair

  • But: no categorical construction yet
slide-80
SLIDE 80

23/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Constructing Initial Transformation Pairs

Conflict essences and initial transformation pairs are closely related (in categories of set-valued functors)

slide-81
SLIDE 81

23/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Constructing Initial Transformation Pairs

Conflict essences and initial transformation pairs are closely related (in categories of set-valued functors) Theorem Given H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2, the pushout of its conflict essence determines its initial transformation pair. C R1 K1 L1 L2 K2 R2 J1 E1 I E2 J2 H1 D1 G D2 H2

p1◦c p2◦c r1 l1 m1 m1 m2 m2 l2 r2

slide-82
SLIDE 82

23/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Constructing Initial Transformation Pairs

Conflict essences and initial transformation pairs are closely related (in categories of set-valued functors) Theorem Given H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2, the pushout of its conflict essence determines its initial transformation pair. C R1 K1 L1 L2 K2 R2 J1 E1 I E2 J2 H1 D1 G D2 H2

p1◦c p2◦c r1 l1 m1 m1 m2 m2 l2 r2

slide-83
SLIDE 83

23/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Constructing Initial Transformation Pairs

Conflict essences and initial transformation pairs are closely related (in categories of set-valued functors) Theorem Given H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2, the pushout of its conflict essence determines its initial transformation pair. C R1 K1 L1 L2 K2 R2 J1 E1 I E2 J2 H1 D1 G D2 H2

p1◦c p2◦c r1 l1 m1 m1 m2 m2 l2 r2

slide-84
SLIDE 84

23/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Constructing Initial Transformation Pairs

Conflict essences and initial transformation pairs are closely related (in categories of set-valued functors) Theorem Given H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2, the pushout of its conflict essence determines its initial transformation pair. C R1 K1 L1 L2 K2 R2 J1 E1 I E2 J2 H1 D1 G D2 H2

p1◦c p2◦c r1 l1 m1 m1 m2 m2 l2 r2

slide-85
SLIDE 85

23/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Constructing Initial Transformation Pairs

Conflict essences and initial transformation pairs are closely related (in categories of set-valued functors) Theorem Given H1

ρ1,m1

⇐= G

ρ2,m2

=⇒ H2, the pushout of its conflict essence determines its initial transformation pair. C R1 K1 L1 L2 K2 R2 J1 E1 I E2 J2 H1 D1 G D2 H2

p1◦c p2◦c r1 l1 m1 m1 m2 m2 l2 r2

slide-86
SLIDE 86

24/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Overview

Conflict Conflict Essence Conflict Reason Critical Pair Essential Critical Pair Initial Conflict

represented by unique is a has unique uniquely determines represented by unique is a is a represented by unique has unique contained in uniquely determines

Available for:

Adhesive Categories et raphT

slide-87
SLIDE 87

24/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Overview

Conflict Conflict Essence Conflict Reason Critical Pair Essential Critical Pair Initial Conflict

represented by unique is a has unique uniquely determines represented by unique is a is a represented by unique has unique contained in uniquely determines

Available for:

Adhesive Categories et raphT

slide-88
SLIDE 88

24/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Overview

Conflict Conflict Essence Conflict Reason Critical Pair Essential Critical Pair Initial Conflict

represented by unique is a has unique uniquely determines represented by unique is a is a represented by unique has unique contained in uniquely determines

Available for:

Adhesive Categories et raphT

Open Problem: in all adhesive categories?

slide-89
SLIDE 89

25/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Conclusions

  • Essential condition allowed powerful characterization for

root causes of conflicts

  • Lots of future work!
  • Constraints and application conditions
  • Compare with notions of granularity (Born et al. 2017)
  • Attributed graphs and other adhesive categories
  • Sesqui-Pushout and AGREE
slide-90
SLIDE 90

26/29 Introduction Characterization Properties Previous Work Initial Conflicts Conclusions

Thank you!

Questions?

slide-91
SLIDE 91

27/29 References Subobjects

References I

Born, Kristopher et al. (2017). “Granularity of Conflicts and Dependencies in Graph Transformation Systems”. In: ICGT.

  • Vol. 10373. LNCS. Springer, pp. 125–141. doi:

10.1007/978-3-319-61470-0_8. url: https://doi.org/10.1007/978-3-319-61470-0_8. Corradini, Andrea et al. (2018). “On the Essence of Parallel Independence for the Double-Pushout and Sesqui-Pushout Approaches”. In: Graph Transformation, Specifications, and

  • Nets. Vol. 10800. LNCS. Springer, pp. 1–18. doi:

10.1007/978-3-319-75396-6_1. url: https://doi.org/10.1007/978-3-319-75396-6_1. Lack, Stephen and Pawel Sobocinski (2005). “Adhesive and quasiadhesive categories”. In: ITA 39.3, pp. 511–545. doi: 10.1051/ita:2005028. url: https://doi.org/10.1051/ita:2005028.

slide-92
SLIDE 92

28/29 References Subobjects

References II

Lambers, Leen, Kristopher Born, et al. (2018). “Initial Conflicts and Dependencies: Critical Pairs Revisited”. In: Graph Transformation, Specifications, and Nets. Vol. 10800. LNCS. Springer, pp. 105–123. doi: 10.1007/978-3-319-75396-6_6. url: https://doi.org/10.1007/978-3-319-75396-6_6. Lambers, Leen, Hartmut Ehrig, and Fernando Orejas (2008). “Efficient Conflict Detection in Graph Transformation Systems by Essential Critical Pairs”. In: ENTCS 211, pp. 17–26. doi: 10.1016/j.entcs.2008.04.026. url: https://doi.org/10.1016/j.entcs.2008.04.026.

slide-93
SLIDE 93

29/29 References Subobjects

Notes