Reduction operators and completion of linear rewriting systems - - PowerPoint PPT Presentation

reduction operators and completion of linear rewriting
SMART_READER_LITE
LIVE PREVIEW

Reduction operators and completion of linear rewriting systems - - PowerPoint PPT Presentation

Motivations Reduction operators Confluence and completion Conclusion Reduction operators and completion of linear rewriting systems Cyrille Chenavier INRIA Lille - Nord Europe Valse team February 8, 2019 1/21 INRIA Lille - Nord Europe


slide-1
SLIDE 1

1/21 Motivations Reduction operators Confluence and completion Conclusion

Reduction operators and completion of linear rewriting systems

Cyrille Chenavier

INRIA Lille - Nord Europe Valse team

February 8, 2019

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-2
SLIDE 2

2/21 Motivations Reduction operators Confluence and completion Conclusion

Plan

  • I. Motivations

⊲ Computational problems in algebra and rewriting theory ⊲ Termination, confluence and Gröbner bases

  • II. Reduction operators

⊲ Reduction operators and linear rewriting systems ⊲ Lattice structure of reduction operators

  • III. Confluence and completion

⊲ Lattice formulation of confluence ⊲ Lattice formulation of completion

  • IV. Conclusion and perspectives

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-3
SLIDE 3

3/21 Motivations Reduction operators Confluence and completion Conclusion

Plan

  • I. Motivations

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-4
SLIDE 4

4/21 Motivations Reduction operators Confluence and completion Conclusion

Computational problems in algebra

◮ Computational problems:

⊲ Our running example: how to compute a linear basis of a K-algebra A?

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-5
SLIDE 5

4/21 Motivations Reduction operators Confluence and completion Conclusion

Computational problems in algebra

◮ Computational problems:

⊲ Our running example: how to compute a linear basis of a K-algebra A? ⊲ Development of effective methods: in algebraic geometry, homological algebra, algebraic combinatorics, for polynomial/functional equations, cryptography, · · ·

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-6
SLIDE 6

4/21 Motivations Reduction operators Confluence and completion Conclusion

Computational problems in algebra

◮ Computational problems:

⊲ Our running example: how to compute a linear basis of a K-algebra A? ⊲ Development of effective methods: in algebraic geometry, homological algebra, algebraic combinatorics, for polynomial/functional equations, cryptography, · · ·

◮ These problems concern various algebraic structures:

⊲ (associative, commutative, Lie) algebras, rings of functional operators, ⊲ operads, PROS, monoidal categories, ⊲ · · ·

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-7
SLIDE 7

4/21 Motivations Reduction operators Confluence and completion Conclusion

Computational problems in algebra

◮ Computational problems:

⊲ Our running example: how to compute a linear basis of a K-algebra A? ⊲ Development of effective methods: in algebraic geometry, homological algebra, algebraic combinatorics, for polynomial/functional equations, cryptography, · · ·

◮ These problems concern various algebraic structures:

⊲ (associative, commutative, Lie) algebras, rings of functional operators, ⊲ operads, PROS, monoidal categories, ⊲ · · ·

◮ Rewriting method: present algebraic structures by generators and oriented relations.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-8
SLIDE 8

4/21 Motivations Reduction operators Confluence and completion Conclusion

Computational problems in algebra

◮ Computational problems:

⊲ Our running example: how to compute a linear basis of a K-algebra A? ⊲ Development of effective methods: in algebraic geometry, homological algebra, algebraic combinatorics, for polynomial/functional equations, cryptography, · · ·

◮ These problems concern various algebraic structures:

⊲ (associative, commutative, Lie) algebras, rings of functional operators, ⊲ operads, PROS, monoidal categories, ⊲ · · ·

◮ Rewriting method: present algebraic structures by generators and oriented relations.

⊲ Notion of normal forms. ⊲ Procedures for computing normal forms.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-9
SLIDE 9

5/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ A = K[x, y] the polynomial algebra over two indeterminates.

⊲ As an associative algebra: 2 generators (x and y) and 1 relation (yx − → xy).

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-10
SLIDE 10

5/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ A = K[x, y] the polynomial algebra over two indeterminates.

⊲ As an associative algebra: 2 generators (x and y) and 1 relation (yx − → xy). ⊲ Monomials over which we cannot apply yx − → xy are called normal forms: xnym.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-11
SLIDE 11

5/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ A = K[x, y] the polynomial algebra over two indeterminates.

⊲ As an associative algebra: 2 generators (x and y) and 1 relation (yx − → xy). ⊲ Monomials over which we cannot apply yx − → xy are called normal forms: xnym. ⊲ In this case: A = K monomials in normal forms .

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-12
SLIDE 12

5/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ A = K[x, y] the polynomial algebra over two indeterminates.

⊲ As an associative algebra: 2 generators (x and y) and 1 relation (yx − → xy). ⊲ Monomials over which we cannot apply yx − → xy are called normal forms: xnym. ⊲ In this case: A = K monomials in normal forms . ⊲ Linear decompositions: obtained by applying yx − → xy as long as it is possible.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-13
SLIDE 13

5/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ A = K[x, y] the polynomial algebra over two indeterminates.

⊲ As an associative algebra: 2 generators (x and y) and 1 relation (yx − → xy). ⊲ Monomials over which we cannot apply yx − → xy are called normal forms: xnym. ⊲ In this case: A = K monomials in normal forms . ⊲ Linear decompositions: obtained by applying yx − → xy as long as it is possible.

◮ A an algebra presented by generators and oriented relations.

⊲ Let NF = monomials in normal forms form . ⊲ Is NF a basis of A?

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-14
SLIDE 14

5/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ A = K[x, y] the polynomial algebra over two indeterminates.

⊲ As an associative algebra: 2 generators (x and y) and 1 relation (yx − → xy). ⊲ Monomials over which we cannot apply yx − → xy are called normal forms: xnym. ⊲ In this case: A = K monomials in normal forms . ⊲ Linear decompositions: obtained by applying yx − → xy as long as it is possible.

◮ A an algebra presented by generators and oriented relations.

⊲ Let NF = monomials in normal forms form . ⊲ Is NF a basis of A? ⊲ That is: is NF a generating family? is NF a free family?

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-15
SLIDE 15

6/21 Motivations Reduction operators Confluence and completion Conclusion

Termination

◮ A = Kx | x − → xx.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-16
SLIDE 16

6/21 Motivations Reduction operators Confluence and completion Conclusion

Termination

◮ A = Kx | x − → xx.

⊲ A = K1 ⊕ Kx and NF = {1}.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-17
SLIDE 17

6/21 Motivations Reduction operators Confluence and completion Conclusion

Termination

◮ A = Kx | x − → xx.

⊲ A = K1 ⊕ Kx and NF = {1}. ⊲ In general, NF is not a generating family of A!

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-18
SLIDE 18

6/21 Motivations Reduction operators Confluence and completion Conclusion

Termination

◮ A = Kx | x − → xx.

⊲ A = K1 ⊕ Kx and NF = {1}. ⊲ In general, NF is not a generating family of A!

  • Definition. Let A be an algebra. A presentation of A is said to be terminating if there

is no infinite sequence of reductions f1 − → f2 − → · · · − → fn − → fn+1 − → · · ·

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-19
SLIDE 19

6/21 Motivations Reduction operators Confluence and completion Conclusion

Termination

◮ A = Kx | x − → xx.

⊲ A = K1 ⊕ Kx and NF = {1}. ⊲ In general, NF is not a generating family of A!

  • Definition. Let A be an algebra. A presentation of A is said to be terminating if there

is no infinite sequence of reductions f1 − → f2 − → · · · − → fn − → fn+1 − → · · · ◮ A an algebra admitting a terminating presentation.

⊲ Every a ∈ A is equal to a normal form.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-20
SLIDE 20

6/21 Motivations Reduction operators Confluence and completion Conclusion

Termination

◮ A = Kx | x − → xx.

⊲ A = K1 ⊕ Kx and NF = {1}. ⊲ In general, NF is not a generating family of A!

  • Definition. Let A be an algebra. A presentation of A is said to be terminating if there

is no infinite sequence of reductions f1 − → f2 − → · · · − → fn − → fn+1 − → · · · ◮ A an algebra admitting a terminating presentation.

⊲ Every a ∈ A is equal to a normal form.

◮ When the presentation of A is terminating, NF is a generating family of A!

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-21
SLIDE 21

7/21 Motivations Reduction operators Confluence and completion Conclusion

Confluence

◮ A = Kx, y | yy − → yx.

⊲ yxy, yxx ∈ NF and yxy = yxx in A

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-22
SLIDE 22

7/21 Motivations Reduction operators Confluence and completion Conclusion

Confluence

◮ A = Kx, y | yy − → yx.

⊲ yxy, yxx ∈ NF and yxy = yxx in A: yyy

  • yxy

yyx

  • yxx

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-23
SLIDE 23

7/21 Motivations Reduction operators Confluence and completion Conclusion

Confluence

◮ A = Kx, y | yy − → yx.

⊲ yxy, yxx ∈ NF and yxy = yxx in A: yyy

  • yxy

yyx

  • yxx
  • Definition. Let A be an algebra. A presentation of A is said to be confluent if

f1

  • f

  • g

f2

  • INRIA Lille - Nord Europe

Reduction operators and completion of linear rewriting systems

slide-24
SLIDE 24

7/21 Motivations Reduction operators Confluence and completion Conclusion

Confluence

◮ A = Kx, y | yy − → yx.

⊲ yxy, yxx ∈ NF and yxy = yxx in A: yyy

  • yxy

yyx

  • yxx
  • Definition. Let A be an algebra. A presentation of A is said to be confluent if

f1

  • f

  • g

f2

  • ◮ When the presentation of A is confluent and terminating, NF is a linear basis of A!

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-25
SLIDE 25

8/21 Motivations Reduction operators Confluence and completion Conclusion

Gröbner Bases

◮ Gröbner bases appear in

⊲ Lie algebras [Shirshov 1962], ⊲ Commutative algebras [Buchberger 1965], ⊲ Associative algebras [Bokut 1976, Bergman 1978, Mora 1992], ⊲ Operads [Dotsenko-Khoroshkin 2010], ⊲ · · ·

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-26
SLIDE 26

8/21 Motivations Reduction operators Confluence and completion Conclusion

Gröbner Bases

◮ Gröbner bases appear in

⊲ Lie algebras [Shirshov 1962], ⊲ Commutative algebras [Buchberger 1965], ⊲ Associative algebras [Bokut 1976, Bergman 1978, Mora 1992], ⊲ Operads [Dotsenko-Khoroshkin 2010], ⊲ · · ·

◮ The relations are oriented w.r.t a monomial order <:

⊲ If f = lm (f ) − r(f ), then lm (f ) − → r(f ). ⊲ A monomial order ensures termination.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-27
SLIDE 27

8/21 Motivations Reduction operators Confluence and completion Conclusion

Gröbner Bases

◮ Gröbner bases appear in

⊲ Lie algebras [Shirshov 1962], ⊲ Commutative algebras [Buchberger 1965], ⊲ Associative algebras [Bokut 1976, Bergman 1978, Mora 1992], ⊲ Operads [Dotsenko-Khoroshkin 2010], ⊲ · · ·

◮ The relations are oriented w.r.t a monomial order <:

⊲ If f = lm (f ) − r(f ), then lm (f ) − → r(f ). ⊲ A monomial order ensures termination.

◮ A = KX | R, where the elements of R are oriented w.r.t a monomial order.

⊲ R is called a Gröbner basis if it induces a confluent presentation.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-28
SLIDE 28

9/21 Motivations Reduction operators Confluence and completion Conclusion

Objectives

◮ Representations of rewriting systems by reduction operators:

⊲ Formalisation of noncommutative Gröbner bases [Bergman, 1978], ⊲ Applications to Koszul duality [Berger 1998].

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-29
SLIDE 29

9/21 Motivations Reduction operators Confluence and completion Conclusion

Objectives

◮ Representations of rewriting systems by reduction operators:

⊲ Formalisation of noncommutative Gröbner bases [Bergman, 1978], ⊲ Applications to Koszul duality [Berger 1998].

◮ Objective: extend the functional approach.

⊲ We introduce a lattice interpretation of the confluence property. ⊲ We deduce a lattice interpretation of the completion procedure.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-30
SLIDE 30

9/21 Motivations Reduction operators Confluence and completion Conclusion

Objectives

◮ Representations of rewriting systems by reduction operators:

⊲ Formalisation of noncommutative Gröbner bases [Bergman, 1978], ⊲ Applications to Koszul duality [Berger 1998].

◮ Objective: extend the functional approach.

⊲ We introduce a lattice interpretation of the confluence property. ⊲ We deduce a lattice interpretation of the completion procedure. ⊲ The functional approach concerns general linear rewriting systems!

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-31
SLIDE 31

10/21 Motivations Reduction operators Confluence and completion Conclusion

Plan

  • II. Reduction operators

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-32
SLIDE 32

11/21 Motivations Reduction operators Confluence and completion Conclusion

Definition

◮ (G, <) a fixed well-ordered set.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-33
SLIDE 33

11/21 Motivations Reduction operators Confluence and completion Conclusion

Definition

◮ (G, <) a fixed well-ordered set.

⊲ For algebras: G is a set of monomials and < is a monomial order.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-34
SLIDE 34

11/21 Motivations Reduction operators Confluence and completion Conclusion

Definition

◮ (G, <) a fixed well-ordered set.

⊲ For algebras: G is a set of monomials and < is a monomial order. ⊲ In our examples: (G, <) is a totally ordered finite set.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-35
SLIDE 35

11/21 Motivations Reduction operators Confluence and completion Conclusion

Definition

◮ (G, <) a fixed well-ordered set.

⊲ For algebras: G is a set of monomials and < is a monomial order. ⊲ In our examples: (G, <) is a totally ordered finite set.

  • Notations. ∀v, v ′ ∈ KG

⊲ lm (v): the greatest basis element occurring in the decomposition of v. ⊲ We let v < v ′ if lm (v) < lm (v ′).

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-36
SLIDE 36

11/21 Motivations Reduction operators Confluence and completion Conclusion

Definition

◮ (G, <) a fixed well-ordered set.

⊲ For algebras: G is a set of monomials and < is a monomial order. ⊲ In our examples: (G, <) is a totally ordered finite set.

  • Notations. ∀v, v ′ ∈ KG

⊲ lm (v): the greatest basis element occurring in the decomposition of v. ⊲ We let v < v ′ if lm (v) < lm (v ′).

  • Definition. An endomorphism T of KG is a reduction operator relative to (G, <) if

⊲ T is a projector, ⊲ ∀g ∈ G, we have T(g) ≤ g.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-37
SLIDE 37

12/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ (G, <) = g1 < g2 < g3 < g4

  • ,

T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-38
SLIDE 38

12/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ (G, <) = g1 < g2 < g3 < g4

  • ,

T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

◮ v = (λ1, λ2, λ3, λ4) ∈ KG.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-39
SLIDE 39

12/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ (G, <) = g1 < g2 < g3 < g4

  • ,

T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

◮ v = (λ1, λ2, λ3, λ4) ∈ KG.

⊲ T1 reduces v as follows v − →

T1

T1(v) = (λ1 + λ2, 0, λ3 + λ4, 0) .

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-40
SLIDE 40

12/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ (G, <) = g1 < g2 < g3 < g4

  • ,

T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

◮ v = (λ1, λ2, λ3, λ4) ∈ KG.

⊲ T1 reduces v as follows v − →

T1

T1(v) = (λ1 + λ2, 0, λ3 + λ4, 0) . ⊲ T2 reduces v as follows v − →

T2

T2(v) = (λ1, λ2 + λ4, λ3, 0) .

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-41
SLIDE 41

13/21 Motivations Reduction operators Confluence and completion Conclusion

Lattice Structure

  • Proposition. The map

ker: RO (G, <) − →

  • subspaces of KG
  • ,

T − → ker(T) is a bijection.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-42
SLIDE 42

13/21 Motivations Reduction operators Confluence and completion Conclusion

Lattice Structure

  • Proposition. The map

ker: RO (G, <) − →

  • subspaces of KG
  • ,

T − → ker(T) is a bijection.

  • Notation. ker−1 :

subspaces of KG − → RO (G, <) the inverse of ker. Lattice structure.

  • RO (G, <) , , ∧, ∨
  • is a lattice where

⊲ T1 T2 if ker (T2) ⊆ ker (T1). ⊲ T1 ∧ T2 = ker−1 (ker(T1) + ker(T2)). ⊲ T1 ∨ T2 = ker−1 (ker(T1) ∩ ker(T2)).

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-43
SLIDE 43

14/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ (G, <) = g1 < g2 < g3 < g4

  • ,

T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-44
SLIDE 44

14/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ (G, <) = g1 < g2 < g3 < g4

  • ,

T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

◮ By definition, ker (T1 ∧ T2) = ker (T1) + ker (T2)

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-45
SLIDE 45

14/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ (G, <) = g1 < g2 < g3 < g4

  • ,

T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

◮ By definition, ker (T1 ∧ T2) = ker (T1) + ker (T2), so that

⊲ ker (T1 ∧ T2) = K{g2 − g1} + K{g4 − g3}

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-46
SLIDE 46

14/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ (G, <) = g1 < g2 < g3 < g4

  • ,

T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

◮ By definition, ker (T1 ∧ T2) = ker (T1) + ker (T2), so that

⊲ ker (T1 ∧ T2) = K{g2 − g1} + K{g4 − g3} + K{g4 − g2}.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-47
SLIDE 47

14/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ (G, <) = g1 < g2 < g3 < g4

  • ,

T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

◮ By definition, ker (T1 ∧ T2) = ker (T1) + ker (T2), so that

⊲ ker (T1 ∧ T2) = K{g2 − g1} + K{g4 − g3} + K{g4 − g2}. ⊲ Hence, ker (T1 ∧ T2) is spanned by the rows of the matrix

  • −1

1 −1 1 −1 1

  • .

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-48
SLIDE 48

14/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ (G, <) = g1 < g2 < g3 < g4

  • ,

T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

◮ By definition, ker (T1 ∧ T2) = ker (T1) + ker (T2), so that

⊲ ker (T1 ∧ T2) = K{g2 − g1} + K{g4 − g3} + K{g4 − g2}. ⊲ Hence, by Gaussian elimination, ker (T1 ∧ T2) is spanned by the rows of the matrix

  • −1

1 −1 1 −1 1

  • .

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-49
SLIDE 49

14/21 Motivations Reduction operators Confluence and completion Conclusion

Example

◮ (G, <) = g1 < g2 < g3 < g4

  • ,

T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

◮ By definition, ker (T1 ∧ T2) = ker (T1) + ker (T2), so that

⊲ ker (T1 ∧ T2) = K{g2 − g1} + K{g4 − g3} + K{g4 − g2}. ⊲ Hence, by Gaussian elimination, ker (T1 ∧ T2) is spanned by the rows of the matrix

  • −1

1 −1 1 −1 1

  • .

◮ We deduce T1 ∧ T2 =

  

1 1 1 1

   .

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-50
SLIDE 50

15/21 Motivations Reduction operators Confluence and completion Conclusion

Plan

  • III. Confluence and completion

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-51
SLIDE 51

16/21 Motivations Reduction operators Confluence and completion Conclusion

Obstructions to confluence

◮ ∀T ∈ RO (G, <), we let NF (T) = {g | T(g) = g}.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-52
SLIDE 52

16/21 Motivations Reduction operators Confluence and completion Conclusion

Obstructions to confluence

◮ ∀T ∈ RO (G, <), we let NF (T) = {g | T(g) = g}. ◮ Given P = (T1, T2) ⊂ RO (G, <), we have NF (T1 ∧ T2) ⊆ NF (T1) ∩ NF (T2) .

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-53
SLIDE 53

16/21 Motivations Reduction operators Confluence and completion Conclusion

Obstructions to confluence

◮ ∀T ∈ RO (G, <), we let NF (T) = {g | T(g) = g}. ◮ Given P = (T1, T2) ⊂ RO (G, <), we have NF (T1 ∧ T2) ⊆ NF (T1) ∩ NF (T2) . ◮ In general, the inclusion is strict

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-54
SLIDE 54

16/21 Motivations Reduction operators Confluence and completion Conclusion

Obstructions to confluence

◮ ∀T ∈ RO (G, <), we let NF (T) = {g | T(g) = g}. ◮ Given P = (T1, T2) ⊂ RO (G, <), we have NF (T1 ∧ T2) ⊆ NF (T1) ∩ NF (T2) . ◮ In general, the inclusion is strict:

⊲ T1 = ker−1 (K{g2 − g1, g4 − g3}) and T2 = ker−1 (K{g4 − g2}).

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-55
SLIDE 55

16/21 Motivations Reduction operators Confluence and completion Conclusion

Obstructions to confluence

◮ ∀T ∈ RO (G, <), we let NF (T) = {g | T(g) = g}. ◮ Given P = (T1, T2) ⊂ RO (G, <), we have NF (T1 ∧ T2) ⊆ NF (T1) ∩ NF (T2) . ◮ In general, the inclusion is strict:

⊲ T1 = ker−1 (K{g2 − g1, g4 − g3}) and T2 = ker−1 (K{g4 − g2}). ⊲ T1 ∧ T2 = ker−1 (K{g4 − g1, g3 − g1, g2 − g1}).

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-56
SLIDE 56

16/21 Motivations Reduction operators Confluence and completion Conclusion

Obstructions to confluence

◮ ∀T ∈ RO (G, <), we let NF (T) = {g | T(g) = g}. ◮ Given P = (T1, T2) ⊂ RO (G, <), we have NF (T1 ∧ T2) ⊆ NF (T1) ∩ NF (T2) . ◮ In general, the inclusion is strict:

⊲ T1 = ker−1 (K{g2 − g1, g4 − g3}) and T2 = ker−1 (K{g4 − g2}). ⊲ T1 ∧ T2 = ker−1 (K{g4 − g1, g3 − g1, g2 − g1}). ⊲ g3 ∈ NF (T1) ∩ NF (T2) and g3 / ∈ NF (T1 ∧ T2).

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-57
SLIDE 57

16/21 Motivations Reduction operators Confluence and completion Conclusion

Obstructions to confluence

◮ ∀T ∈ RO (G, <), we let NF (T) = {g | T(g) = g}. ◮ Given P = (T1, T2) ⊂ RO (G, <), we have NF (T1 ∧ T2) ⊆ NF (T1) ∩ NF (T2) . ◮ In general, the inclusion is strict:

⊲ T1 = ker−1 (K{g2 − g1, g4 − g3}) and T2 = ker−1 (K{g4 − g2}). ⊲ T1 ∧ T2 = ker−1 (K{g4 − g1, g3 − g1, g2 − g1}). ⊲ g3 ∈ NF (T1) ∩ NF (T2) and g3 / ∈ NF (T1 ∧ T2).

  • Remark. The obstruction to confluence is ∄ g3 −

→ g1: g4

T2

  • T1
  • g3

g2

T1

  • g1

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-58
SLIDE 58

17/21 Motivations Reduction operators Confluence and completion Conclusion

Confluence

◮ Let F ⊂ RO (G, <).

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-59
SLIDE 59

17/21 Motivations Reduction operators Confluence and completion Conclusion

Confluence

◮ Let F ⊂ RO (G, <). ∧F = ker−1

  • T∈F

ker (T)

  • and NF (F) =
  • T∈F

NF (T) .

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-60
SLIDE 60

17/21 Motivations Reduction operators Confluence and completion Conclusion

Confluence

◮ Let F ⊂ RO (G, <). ∧F = ker−1

  • T∈F

ker (T)

  • and NF (F) =
  • T∈F

NF (T) .

  • Lemma. NF (∧F)

⊆ NF (F).

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-61
SLIDE 61

17/21 Motivations Reduction operators Confluence and completion Conclusion

Confluence

◮ Let F ⊂ RO (G, <). ∧F = ker−1

  • T∈F

ker (T)

  • and NF (F) =
  • T∈F

NF (T) .

  • Lemma. NF (∧F)

⊆ NF (F).

  • Notation. ObsF = NF (F) \ NF (∧F).

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-62
SLIDE 62

17/21 Motivations Reduction operators Confluence and completion Conclusion

Confluence

◮ Let F ⊂ RO (G, <). ∧F = ker−1

  • T∈F

ker (T)

  • and NF (F) =
  • T∈F

NF (T) .

  • Lemma. NF (∧F)

⊆ NF (F).

  • Notation. ObsF = NF (F) \ NF (∧F).
  • Definition. F is said to be confluent if ObsF = ∅.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-63
SLIDE 63

17/21 Motivations Reduction operators Confluence and completion Conclusion

Confluence

◮ Let F ⊂ RO (G, <). ∧F = ker−1

  • T∈F

ker (T)

  • and NF (F) =
  • T∈F

NF (T) .

  • Lemma. NF (∧F)

⊆ NF (F).

  • Notation. ObsF = NF (F) \ NF (∧F).
  • Definition. F is said to be confluent if ObsF = ∅.
  • Proposition. F is confluent if and only if it is so for

− →

F

=

  • T∈F

− →

T

.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-64
SLIDE 64

18/21 Motivations Reduction operators Confluence and completion Conclusion

Completion

◮ P = (T1, T2), where T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-65
SLIDE 65

18/21 Motivations Reduction operators Confluence and completion Conclusion

Completion

◮ P = (T1, T2), where T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

◮ We complete P with? g4

T2

  • T1
  • g3

g2

T1

  • g1

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-66
SLIDE 66

18/21 Motivations Reduction operators Confluence and completion Conclusion

Completion

◮ P = (T1, T2), where T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

◮ We complete P with g4

T2

  • T1
  • g3

CP

  • g2

T1

  • g1

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-67
SLIDE 67

18/21 Motivations Reduction operators Confluence and completion Conclusion

Completion

◮ P = (T1, T2), where T1 =

  

1 1 1 1

  

and T2 =

  

1 1 1 1

   .

◮ We complete P with g4

T2

  • T1
  • g3

CP

  • g2

T1

  • g1

◮ Formally C P =

  

1 1 1 1

   .

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-68
SLIDE 68

19/21 Motivations Reduction operators Confluence and completion Conclusion

Lattice description of the completion procedure

◮ Let F ⊂ RO (G, <). ◮ F is completed by C F ∈ RO (G, <), defined as follows C F(g) =

  • ∧ F(g),

if g ∈ ObsF g,

  • therwise.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-69
SLIDE 69

19/21 Motivations Reduction operators Confluence and completion Conclusion

Lattice description of the completion procedure

◮ Let F ⊂ RO (G, <). ◮ F is completed by C F ∈ RO (G, <), defined as follows C F(g) =

  • ∧ F(g),

if g ∈ ObsF g,

  • therwise.

◮ What is the procedure for computing C F?

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-70
SLIDE 70

19/21 Motivations Reduction operators Confluence and completion Conclusion

Lattice description of the completion procedure

◮ Let F ⊂ RO (G, <). ◮ F is completed by C F ∈ RO (G, <), defined as follows C F(g) =

  • ∧ F(g),

if g ∈ ObsF g,

  • therwise.

◮ What is the procedure for computing C F?

⊲ Compute ∧F by “Gaussian elimination”.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-71
SLIDE 71

19/21 Motivations Reduction operators Confluence and completion Conclusion

Lattice description of the completion procedure

◮ Let F ⊂ RO (G, <). ◮ F is completed by C F ∈ RO (G, <), defined as follows C F(g) =

  • ∧ F(g),

if g ∈ ObsF g,

  • therwise.

◮ What is the procedure for computing C F?

⊲ Compute ∧F by “Gaussian elimination”. ⊲ For every g such that g / ∈ NF (∧F) and g ∈ NF (F): CF (g) = ∧F(g).

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-72
SLIDE 72

19/21 Motivations Reduction operators Confluence and completion Conclusion

Lattice description of the completion procedure

◮ Let F ⊂ RO (G, <). ◮ F is completed by C F ∈ RO (G, <), defined as follows C F(g) =

  • ∧ F(g),

if g ∈ ObsF g,

  • therwise.

◮ What is the procedure for computing C F?

⊲ Compute ∧F by “Gaussian elimination”. ⊲ For every g such that g / ∈ NF (∧F) and g ∈ NF (F): CF (g) = ∧F(g). ⊲ For every all other g ∈ G: CF (g) = g.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-73
SLIDE 73

19/21 Motivations Reduction operators Confluence and completion Conclusion

Lattice description of the completion procedure

◮ Let F ⊂ RO (G, <). ◮ F is completed by C F ∈ RO (G, <), defined as follows C F(g) =

  • ∧ F(g),

if g ∈ ObsF g,

  • therwise.

◮ What is the procedure for computing C F?

⊲ Compute ∧F by “Gaussian elimination”. ⊲ For every g such that g / ∈ NF (∧F) and g ∈ NF (F): CF (g) = ∧F(g). ⊲ For every all other g ∈ G: CF (g) = g.

  • Theorem. Letting ∨F = ker−1 (NF (F)), we have:

C F = (∧F) ∨ ∨F .

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-74
SLIDE 74

20/21 Motivations Reduction operators Confluence and completion Conclusion

Plan

  • IV. Conclusion

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-75
SLIDE 75

21/21 Motivations Reduction operators Confluence and completion Conclusion

Summary and perspectives

◮ Summary of results (arXiv:1605.00174):

⊲ We equipped RO (G, <) with a lattice structure. ⊲ We introduced lattice formulations of confluence and completion.

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-76
SLIDE 76

21/21 Motivations Reduction operators Confluence and completion Conclusion

Summary and perspectives

◮ Summary of results (arXiv:1605.00174):

⊲ We equipped RO (G, <) with a lattice structure. ⊲ We introduced lattice formulations of confluence and completion.

◮ This lattice approach provides applications to higher-dimensional algebra:

⊲ Construction of a contracting homotopy for the Koszul complex (arXiv:1504.03222).

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-77
SLIDE 77

21/21 Motivations Reduction operators Confluence and completion Conclusion

Summary and perspectives

◮ Summary of results (arXiv:1605.00174):

⊲ We equipped RO (G, <) with a lattice structure. ⊲ We introduced lattice formulations of confluence and completion.

◮ This lattice approach provides applications to higher-dimensional algebra:

⊲ Construction of a contracting homotopy for the Koszul complex (arXiv:1504.03222). ⊲ Procedure for computing syzygies for linear rewriting systems (arXiv:1708.08709).

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-78
SLIDE 78

21/21 Motivations Reduction operators Confluence and completion Conclusion

Summary and perspectives

◮ Summary of results (arXiv:1605.00174):

⊲ We equipped RO (G, <) with a lattice structure. ⊲ We introduced lattice formulations of confluence and completion.

◮ This lattice approach provides applications to higher-dimensional algebra:

⊲ Construction of a contracting homotopy for the Koszul complex (arXiv:1504.03222). ⊲ Procedure for computing syzygies for linear rewriting systems (arXiv:1708.08709).

◮ Further works:

⊲ Reduction operators for left modules. ⊲ Applications to algebraic study of linear functionnal systems (e.g. Ore extensions).

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems

slide-79
SLIDE 79

21/21 Motivations Reduction operators Confluence and completion Conclusion

Summary and perspectives

◮ Summary of results (arXiv:1605.00174):

⊲ We equipped RO (G, <) with a lattice structure. ⊲ We introduced lattice formulations of confluence and completion.

◮ This lattice approach provides applications to higher-dimensional algebra:

⊲ Construction of a contracting homotopy for the Koszul complex (arXiv:1504.03222). ⊲ Procedure for computing syzygies for linear rewriting systems (arXiv:1708.08709).

◮ Further works:

⊲ Reduction operators for left modules. ⊲ Applications to algebraic study of linear functionnal systems (e.g. Ore extensions).

◮ Thank you for listening!

INRIA Lille - Nord Europe Reduction operators and completion of linear rewriting systems