combinatorial operads rewrite systems and formal grammars
play

Combinatorial operads, rewrite systems, and formal grammars Samuele - PowerPoint PPT Presentation

Combinatorial operads, rewrite systems, and formal grammars Samuele Giraudo LIGM, Universit Paris-Est Marne-la-Valle Computational Logic and Applications July 12, 2019 1 / 40 Outline Operads Enumeration Generation 2 / 40 Outline


  1. Some suboperads of Path For any m � 0 , an m -Dyck path is a path starting and ending with 0 and m made of steps and . 0 — Example — is a 2 -Dyck path of size 10 . — Proposition — For any m � 0 , the set Dyck ( m ) of all m -Dyck paths is a suboperad of Path . A Motzkin path is a path starting and ending with 0 and made of steps , , and . — Example — is a Motzkin path of size 16 . 10 / 40

  2. Some suboperads of Path For any m � 0 , an m -Dyck path is a path starting and ending with 0 and m made of steps and . 0 — Example — is a 2 -Dyck path of size 10 . — Proposition — For any m � 0 , the set Dyck ( m ) of all m -Dyck paths is a suboperad of Path . A Motzkin path is a path starting and ending with 0 and made of steps , , and . — Example — is a Motzkin path of size 16 . — Proposition — The set Motz of all Motzkin paths is a suboperad of Path . 10 / 40

  3. Algebras over operads Let O be an operad. An algebra over O is a space V equipped, for all x ∈ O ( n ) , with linear maps x : V ⊗ · · · ⊗ V → V � �� � n 11 / 40

  4. Algebras over operads Let O be an operad. An algebra over O is a space V equipped, for all x ∈ O ( n ) , with linear maps x : V ⊗ · · · ⊗ V → V � �� � n such that 1 is the identity map on V 11 / 40

  5. Algebras over operads Let O be an operad. An algebra over O is a space V equipped, for all x ∈ O ( n ) , with linear maps x : V ⊗ · · · ⊗ V → V � �� � n such that 1 is the identity map on V and the compatibility relation x x ◦ i y = . . . . . . v | x | + | y |− 1 v 1 y . . . v | x | + | y |− 1 v 1 . . . v i + | y |− 1 v i holds for any x, y ∈ O , i ∈ [ | x | ] , and v 1 , . . . , v | x | + | y |− 1 ∈ V . 11 / 40

  6. Algebras over operads — Example — Let As be the associative operad defined by As ( n ) := { ⋆ n } for all n � 1 and ⋆ n ◦ i ⋆ m := ⋆ n + m − 1 . 12 / 40

  7. Algebras over operads — Example — Let As be the associative operad defined by As ( n ) := { ⋆ n } for all n � 1 and ⋆ n ◦ i ⋆ m := ⋆ n + m − 1 . This operad is minimally generated by ⋆ 2 . 12 / 40

  8. Algebras over operads — Example — Let As be the associative operad defined by As ( n ) := { ⋆ n } for all n � 1 and ⋆ n ◦ i ⋆ m := ⋆ n + m − 1 . This operad is minimally generated by ⋆ 2 . Any algebra over As is a space V endowed with linear operations ⋆ n of arity n � 1 12 / 40

  9. Algebras over operads — Example — Let As be the associative operad defined by As ( n ) := { ⋆ n } for all n � 1 and ⋆ n ◦ i ⋆ m := ⋆ n + m − 1 . This operad is minimally generated by ⋆ 2 . Any algebra over As is a space V endowed with linear operations ⋆ n of arity n � 1 where ⋆ 2 satisfies, for all v 1 , v 2 , v 3 ∈ V , ( ⋆ 2 ◦ 1 ⋆ 2 ) ( v 1 , v 2 , v 3 ) � ( ⋆ 2 ◦ 2 ⋆ 2 ) ( v 1 , v 2 , v 3 ) 12 / 40

  10. Algebras over operads — Example — Let As be the associative operad defined by As ( n ) := { ⋆ n } for all n � 1 and ⋆ n ◦ i ⋆ m := ⋆ n + m − 1 . This operad is minimally generated by ⋆ 2 . Any algebra over As is a space V endowed with linear operations ⋆ n of arity n � 1 where ⋆ 2 satisfies, for all v 1 , v 2 , v 3 ∈ V , ( ⋆ 2 ◦ 1 ⋆ 2 ) ( v 1 , v 2 , v 3 ) = ⋆ 2 ( ⋆ 2 ( v 1 , v 2 ) , v 3 ) � ( ⋆ 2 ◦ 2 ⋆ 2 ) ( v 1 , v 2 , v 3 ) 12 / 40

  11. Algebras over operads — Example — Let As be the associative operad defined by As ( n ) := { ⋆ n } for all n � 1 and ⋆ n ◦ i ⋆ m := ⋆ n + m − 1 . This operad is minimally generated by ⋆ 2 . Any algebra over As is a space V endowed with linear operations ⋆ n of arity n � 1 where ⋆ 2 satisfies, for all v 1 , v 2 , v 3 ∈ V , ( ⋆ 2 ◦ 1 ⋆ 2 ) ( v 1 , v 2 , v 3 ) = ⋆ 2 ( ⋆ 2 ( v 1 , v 2 ) , v 3 ) � ( ⋆ 2 ◦ 2 ⋆ 2 ) ( v 1 , v 2 , v 3 ) = ⋆ 2 ( v 1 , ⋆ 2 ( v 2 , v 3 )) . 12 / 40

  12. Algebras over operads — Example — Let As be the associative operad defined by As ( n ) := { ⋆ n } for all n � 1 and ⋆ n ◦ i ⋆ m := ⋆ n + m − 1 . This operad is minimally generated by ⋆ 2 . Any algebra over As is a space V endowed with linear operations ⋆ n of arity n � 1 where ⋆ 2 satisfies, for all v 1 , v 2 , v 3 ∈ V , ( ⋆ 2 ◦ 1 ⋆ 2 ) ( v 1 , v 2 , v 3 ) = ⋆ 2 ( ⋆ 2 ( v 1 , v 2 ) , v 3 ) � � ( ⋆ 2 ◦ 2 ⋆ 2 ) ( v 1 , v 2 , v 3 ) = ⋆ 2 ( v 1 , ⋆ 2 ( v 2 , v 3 )) . 12 / 40

  13. Algebras over operads — Example — Let As be the associative operad defined by As ( n ) := { ⋆ n } for all n � 1 and ⋆ n ◦ i ⋆ m := ⋆ n + m − 1 . This operad is minimally generated by ⋆ 2 . Any algebra over As is a space V endowed with linear operations ⋆ n of arity n � 1 where ⋆ 2 satisfies, for all v 1 , v 2 , v 3 ∈ V , ( ⋆ 2 ◦ 1 ⋆ 2 ) ( v 1 , v 2 , v 3 ) = ⋆ 2 ( ⋆ 2 ( v 1 , v 2 ) , v 3 ) � � ( ⋆ 2 ◦ 2 ⋆ 2 ) ( v 1 , v 2 , v 3 ) = ⋆ 2 ( v 1 , ⋆ 2 ( v 2 , v 3 )) . Using infix notation for the binary operation ⋆ 2 , we obtain the relation ( v 1 ⋆ 2 v 2 ) ⋆ 2 v 3 = v 1 ⋆ 2 ( v 2 ⋆ 2 v 3 ) , so that algebras over As are associative algebras. 12 / 40

  14. Algebras over operads — Example — Let As be the associative operad defined by As ( n ) := { ⋆ n } for all n � 1 and ⋆ n ◦ i ⋆ m := ⋆ n + m − 1 . This operad is minimally generated by ⋆ 2 . Any algebra over As is a space V endowed with linear operations ⋆ n of arity n � 1 where ⋆ 2 satisfies, for all v 1 , v 2 , v 3 ∈ V , ( ⋆ 2 ◦ 1 ⋆ 2 ) ( v 1 , v 2 , v 3 ) = ⋆ 2 ( ⋆ 2 ( v 1 , v 2 ) , v 3 ) � � ( ⋆ 2 ◦ 2 ⋆ 2 ) ( v 1 , v 2 , v 3 ) = ⋆ 2 ( v 1 , ⋆ 2 ( v 2 , v 3 )) . Using infix notation for the binary operation ⋆ 2 , we obtain the relation ( v 1 ⋆ 2 v 2 ) ⋆ 2 v 3 = v 1 ⋆ 2 ( v 2 ⋆ 2 v 3 ) , so that algebras over As are associative algebras. In the same way, there are operads for ◮ Lie alg.; ◮ duplicial alg. [ Loday , 2008] ; ◮ pre-Lie alg. [ Chapoton, Livernet , 2001] ; ◮ diassociative alg. [ Loday , 2001] ; ◮ dendriform alg. [ Loday , 2001] ; ◮ brace alg. 12 / 40

  15. Scope of operads As main benefits, operads ◮ offer a formalism to compute over operations; ◮ allow us to work virtually with all the structures of a type; ◮ lead to discover the underlying combinatorics of types of algebras. 13 / 40

  16. Scope of operads As main benefits, operads ◮ offer a formalism to compute over operations; ◮ allow us to work virtually with all the structures of a type; ◮ lead to discover the underlying combinatorics of types of algebras. Endowing a set of combinatorial objects with an operad structure helps to ◮ highlight elementary building block for the objects; ◮ build combinatorial structures (graded graphs, posets, latices, etc. ); ◮ enumerative prospects and discovery of statistics. 13 / 40

  17. Outline Enumeration 14 / 40

  18. Syntax trees An alphabet is a graded set G := � n � 1 G ( n ) . 15 / 40

  19. Syntax trees An alphabet is a graded set G := � n � 1 G ( n ) . Let S ( G ) be the set of G -syntax trees, defined recursively by ◮ ∈ S ( G ) ; ◮ if a ∈ G and t 1 , . . . , t | a | ∈ S ( G ) , then a ( t 1 , . . . , t | a | ) ∈ S ( G ) . — Example — Let G := G (2) ⊔ G (3) such that G (2) = { a , b } and G (3) = { c } . denotes the G -tree c c b c ( , c ( a ( , ) , , b ( a ( , ) , c ( , , ) ) ) , b ( , b ( , ) ) ) a b b having degree 8 and arity 12 . a c 15 / 40

  20. Syntax trees An alphabet is a graded set G := � n � 1 G ( n ) . Let S ( G ) be the set of G -syntax trees, defined recursively by ◮ ∈ S ( G ) ; ◮ if a ∈ G and t 1 , . . . , t | a | ∈ S ( G ) , then a ( t 1 , . . . , t | a | ) ∈ S ( G ) . Let t ∈ S ( G ) . Some definitions: ◮ is the leaf; ◮ the degree deg( t ) of t is its number of internal nodes; ◮ the arity | t | of t is its number of leaves. — Example — Let G := G (2) ⊔ G (3) such that G (2) = { a , b } and G (3) = { c } . denotes the G -tree c c b c ( , c ( a ( , ) , , b ( a ( , ) , c ( , , ) ) ) , b ( , b ( , ) ) ) a b b having degree 8 and arity 12 . a c 15 / 40

  21. Compositions of syntax trees Let t , s ∈ S ( G ) . For each i ∈ [ | t | ] , the partial composition t ◦ i s is the tree obtained by grafing the root of s onto the i th leaf of t . — Example — c c c b b ◦ 5 = c b a a c b b a b a c 16 / 40

  22. Compositions of syntax trees Let t , s ∈ S ( G ) . For each i ∈ [ | t | ] , the partial composition t ◦ i s is the tree obtained by grafing the root of s onto the i th leaf of t . — Example — c c c b b ◦ 5 = c b a a c b b a b a c � � Let t , s 1 , ..., s | t | be G -trees. The full composition t ◦ s 1 , . . . , s | t | is obtained by grafing simultaneously the roots of each s i onto the i th leaf of t . — Example —   b a b ◦  =  b , ,  a a c  a a a c b 16 / 40

  23. Free operads Let G be an alphabet. 17 / 40

  24. Free operads Let G be an alphabet. The free operad on G is the operad on the set S ( G ) wherein ◮ elements of arity n are the G -trees of arity n ; 17 / 40

  25. Free operads Let G be an alphabet. The free operad on G is the operad on the set S ( G ) wherein ◮ elements of arity n are the G -trees of arity n ; ◮ the partial composition map ◦ i is the one of the G -trees; 17 / 40

  26. Free operads Let G be an alphabet. The free operad on G is the operad on the set S ( G ) wherein ◮ elements of arity n are the G -trees of arity n ; ◮ the partial composition map ◦ i is the one of the G -trees; ◮ the unit is . 17 / 40

  27. Free operads Let G be an alphabet. The free operad on G is the operad on the set S ( G ) wherein ◮ elements of arity n are the G -trees of arity n ; ◮ the partial composition map ◦ i is the one of the G -trees; ◮ the unit is . Let c : G → S ( G ) be the natural injection (made implicit in the sequel). 17 / 40

  28. Free operads Let G be an alphabet. The free operad on G is the operad on the set S ( G ) wherein ◮ elements of arity n are the G -trees of arity n ; ◮ the partial composition map ◦ i is the one of the G -trees; ◮ the unit is . Let c : G → S ( G ) be the natural injection (made implicit in the sequel). Free operads satisfy the following universality property. f For any alphabet G , any operad O , O G and any map f : G → O preserving the arities, there exists a unique op- c φ erad morphism φ : S ( G ) → O such that f = φ ◦ c . S ( G ) 17 / 40

  29. Factors and prefixes Let t , s ∈ S ( G ) . 18 / 40

  30. Factors and prefixes Let t , s ∈ S ( G ) . If t decomposes as � � �� t = r ◦ i s ◦ r 1 , . . . , r [ s | for some trees r , r 1 , ..., r | s | , and i ∈ [ | r | ] , then s is a factor of t . This property is denoted by s � f t . — Example — b c c b � f b a b a c b 18 / 40

  31. Factors and prefixes Let t , s ∈ S ( G ) . If t decomposes as � � �� t = r ◦ i s ◦ r 1 , . . . , r [ s | for some trees r , r 1 , ..., r | s | , and i ∈ [ | r | ] , then s is a factor of t . This property is denoted by s � f t . If in the previous decomposition r = , then s is a prefix of t . This property is denoted by s � p t . — Example — b b c c b b c b � f � p c b a b a c a a c b b b b 18 / 40

  32. Patern avoidance and enumeration ✚ A G -tree t avoids a G -tree s if s ✚ � f t . 19 / 40

  33. Patern avoidance and enumeration ✚ A G -tree t avoids a G -tree s if s ✚ � f t . For any P ⊆ S ( G ) , let A( P ) = { t ∈ S ( G ) : for all s ∈ P , s ✚ � f t } . ✚ 19 / 40

  34. Patern avoidance and enumeration ✚ A G -tree t avoids a G -tree s if s ✚ � f t . For any P ⊆ S ( G ) , let A( P ) = { t ∈ S ( G ) : for all s ∈ P , s ✚ � f t } . ✚ — Qestion — Enumerate A( P ) w.r.t. the arities of the trees. 19 / 40

  35. Patern avoidance and enumeration ✚ A G -tree t avoids a G -tree s if s ✚ � f t . For any P ⊆ S ( G ) , let A( P ) = { t ∈ S ( G ) : for all s ∈ P , s ✚ � f t } . ✚ — Example — � � ◮ A a a b b is enumerated by 1 , 2 , 4 , 8 , 16 , 32 , 64 , 128 , . . . . a b a b — Qestion — Enumerate A( P ) w.r.t. the arities of the trees. 19 / 40

  36. Patern avoidance and enumeration ✚ A G -tree t avoids a G -tree s if s ✚ � f t . For any P ⊆ S ( G ) , let A( P ) = { t ∈ S ( G ) : for all s ∈ P , s ✚ � f t } . ✚ — Example — � � ◮ A a a b b is enumerated by 1 , 2 , 4 , 8 , 16 , 32 , 64 , 128 , . . . . a b a b � � ◮ A a c c a is enumerated by 1 , 1 , 2 , 4 , 9 , 21 , 51 , 127 , . . . ( A001006 ) . c a c a — Qestion — Enumerate A( P ) w.r.t. the arities of the trees. 19 / 40

  37. Patern avoidance and enumeration ✚ A G -tree t avoids a G -tree s if s ✚ � f t . For any P ⊆ S ( G ) , let A( P ) = { t ∈ S ( G ) : for all s ∈ P , s ✚ � f t } . ✚ — Example — � � ◮ A a a b b is enumerated by 1 , 2 , 4 , 8 , 16 , 32 , 64 , 128 , . . . . a b a b � � ◮ A a c c a is enumerated by 1 , 1 , 2 , 4 , 9 , 21 , 51 , 127 , . . . ( A001006 ) . c a c a   b ◮ A  is enumerated by 1 , 2 , 5 , 13 , 35 , 96 , 267 , 750 ,. . . ( A005773 ) . a b b  b a a a b — Qestion — Enumerate A( P ) w.r.t. the arities of the trees. 19 / 40

  38. Formal power series of trees For any P , Q ⊆ S ( G ) , let � F ( P , Q ) := t . t ∈ S ( G ) t ∈ A( P ) � p t ✚ ∀ s ∈Q , s ✚ This is the formal sum of all the G -trees avoiding as factors all paterns of P and avoiding as prefixes all paterns of Q . 20 / 40

  39. Formal power series of trees For any P , Q ⊆ S ( G ) , let � F ( P , Q ) := t . t ∈ S ( G ) t ∈ A( P ) � p t ✚ ∀ s ∈Q , s ✚ This is the formal sum of all the G -trees avoiding as factors all paterns of P and avoiding as prefixes all paterns of Q . Since ◮ F ( P , ∅ ) is the formal sum of all the trees of A( P ) ; 20 / 40

  40. Formal power series of trees For any P , Q ⊆ S ( G ) , let � F ( P , Q ) := t . t ∈ S ( G ) t ∈ A( P ) � p t ✚ ∀ s ∈Q , s ✚ This is the formal sum of all the G -trees avoiding as factors all paterns of P and avoiding as prefixes all paterns of Q . Since ◮ F ( P , ∅ ) is the formal sum of all the trees of A( P ) ; ◮ the linear map t �→ z | t | sends F ( P , ∅ ) to the generating series of A( P ) ; 20 / 40

  41. Formal power series of trees For any P , Q ⊆ S ( G ) , let � F ( P , Q ) := t . t ∈ S ( G ) t ∈ A( P ) � p t ✚ ∀ s ∈Q , s ✚ This is the formal sum of all the G -trees avoiding as factors all paterns of P and avoiding as prefixes all paterns of Q . Since ◮ F ( P , ∅ ) is the formal sum of all the trees of A( P ) ; ◮ the linear map t �→ z | t | sends F ( P , ∅ ) to the generating series of A( P ) ; the series F ( P , Q ) contains all the enumerative data about the trees avoiding P . 20 / 40

  42. System of equations When G , P , and Q satisfy some conditions, F ( P , Q ) expresses as an inclusion-exclusion formula involving simpler terms F ( P , S i ) . — Theorem — The series F ( P , Q ) satisfies � � ( − 1) 1+ ℓ a ¯ F ( P , Q ) = + ◦ [ F ( P , S 1 ) , . . . , F ( P , S k )] . k � 1 ℓ � 1 a ∈ G ( k ) � R (1) ,..., R ( ℓ ) � ⊆ M (( P∪Q ) a ) ( S 1 ,..., S k )= R (1) ∔ ··· ∔ R ( ℓ ) 21 / 40

  43. System of equations When G , P , and Q satisfy some conditions, F ( P , Q ) expresses as an inclusion-exclusion formula involving simpler terms F ( P , S i ) . — Theorem — The series F ( P , Q ) satisfies � � ( − 1) 1+ ℓ a ¯ F ( P , Q ) = + ◦ [ F ( P , S 1 ) , . . . , F ( P , S k )] . k � 1 ℓ � 1 a ∈ G ( k ) � R (1) ,..., R ( ℓ ) � ⊆ M (( P∪Q ) a ) ( S 1 ,..., S k )= R (1) ∔ ··· ∔ R ( ℓ ) This leads to a system of equations for the generating series of A( P ) . Indeed, the generating series of A( P ) is the series F ( P , ∅ ) where � � ( − 1) 1+ ℓ � F ( P , Q ) = z + F ( P , S i ) . k � 1 ℓ � 1 i ∈ [ k ] a ∈ G ( k ) � R (1) ,..., R ( ℓ ) � ⊆ M (( P∪Q ) a ) ( S 1 ,..., S k )= R (1) ∔ ··· ∔ R ( ℓ ) 21 / 40

  44. System of equations — Example — � � For P := a , we obtain the system of formal power series of trees a b F ( P , ∅ ) = + a ¯ ◦ [ F ( P , { a } ) , F ( P , ∅ )] + a ¯ ◦ [ F ( P , ∅ ) , F ( P , { b } )] − a ¯ ◦ [ F ( P , { a } ) , F ( P , { b } )] + b ¯ ◦ [ F ( P , ∅ ) , F ( P , ∅ )] , F ( P , { a } ) = + b ¯ ◦ [ F ( P , ∅ ) , F ( P , ∅ )] , F ( P , { b } ) = + a ¯ ◦ [ F ( P , { a } ) , F ( P , ∅ )] + a ¯ ◦ [ F ( P , ∅ ) , F ( P , { b } )] − a ¯ ◦ [ F ( P , { a } ) , F ( P , { b } )] . 22 / 40

  45. System of equations — Example — � � For P := a , we obtain the system of formal power series of trees a b F ( P , ∅ ) = + a ¯ ◦ [ F ( P , { a } ) , F ( P , ∅ )] + a ¯ ◦ [ F ( P , ∅ ) , F ( P , { b } )] − a ¯ ◦ [ F ( P , { a } ) , F ( P , { b } )] + b ¯ ◦ [ F ( P , ∅ ) , F ( P , ∅ )] , F ( P , { a } ) = + b ¯ ◦ [ F ( P , ∅ ) , F ( P , ∅ )] , F ( P , { b } ) = + a ¯ ◦ [ F ( P , { a } ) , F ( P , ∅ )] + a ¯ ◦ [ F ( P , ∅ ) , F ( P , { b } )] − a ¯ ◦ [ F ( P , { a } ) , F ( P , { b } )] . This leads to the system of generating series F ( P , ∅ ) = z + F ( P , { a } ) F ( P , ∅ ) + F ( P , ∅ ) F ( P , { b } ) − F ( P , { a } ) F ( P , { b } ) + F ( P , ∅ ) F ( P , ∅ ) , F ( P , { a } ) = z + F ( P , ∅ ) F ( P , ∅ ) , F ( P , { b } ) = z + F ( P , { a } ) F ( P , ∅ ) + F ( P , ∅ ) F ( P , { b } ) − F ( P , { a } ) F ( P , { b } ) . 22 / 40

  46. System of equations — Example — � � For P := a , we obtain the system of formal power series of trees a b F ( P , ∅ ) = + a ¯ ◦ [ F ( P , { a } ) , F ( P , ∅ )] + a ¯ ◦ [ F ( P , ∅ ) , F ( P , { b } )] − a ¯ ◦ [ F ( P , { a } ) , F ( P , { b } )] + b ¯ ◦ [ F ( P , ∅ ) , F ( P , ∅ )] , F ( P , { a } ) = + b ¯ ◦ [ F ( P , ∅ ) , F ( P , ∅ )] , F ( P , { b } ) = + a ¯ ◦ [ F ( P , { a } ) , F ( P , ∅ )] + a ¯ ◦ [ F ( P , ∅ ) , F ( P , { b } )] − a ¯ ◦ [ F ( P , { a } ) , F ( P , { b } )] . This leads to the system of generating series F ( P , ∅ ) = z + F ( P , { a } ) F ( P , ∅ ) + F ( P , ∅ ) F ( P , { b } ) − F ( P , { a } ) F ( P , { b } ) + F ( P , ∅ ) F ( P , ∅ ) , F ( P , { a } ) = z + F ( P , ∅ ) F ( P , ∅ ) , F ( P , { b } ) = z + F ( P , { a } ) F ( P , ∅ ) + F ( P , ∅ ) F ( P , { b } ) − F ( P , { a } ) F ( P , { b } ) . As a consequence, F ( P , ∅ ) satisfies z − F ( P , ∅ ) + (2 + z ) F ( P , ∅ ) 2 − F ( P , ∅ ) 3 + F ( P , ∅ ) 4 = 0 . 22 / 40

  47. Operads and presentations Let O be an operad. A congruence of O is an equivalence relation ≡ on O preserving the arities and such that x ≡ x ′ and y ≡ y ′ imply x ◦ i y ≡ x ′ ◦ i y ′ for all i ∈ [ | x | ] . 23 / 40

  48. Operads and presentations Let O be an operad. A congruence of O is an equivalence relation ≡ on O preserving the arities and such that x ≡ x ′ and y ≡ y ′ imply x ◦ i y ≡ x ′ ◦ i y ′ for all i ∈ [ | x | ] . A presentation of O is a pair ( G , ≡ ) such that G is an alphabet and ≡ is a congruence of O satisfying O ≃ S ( G ) / ≡ . 23 / 40

  49. Operads and presentations Let O be an operad. A congruence of O is an equivalence relation ≡ on O preserving the arities and such that x ≡ x ′ and y ≡ y ′ imply x ◦ i y ≡ x ′ ◦ i y ′ for all i ∈ [ | x | ] . A presentation of O is a pair ( G , ≡ ) such that G is an alphabet and ≡ is a congruence of O satisfying O ≃ S ( G ) / ≡ . — Example — The operad Motz admits the presentation ( G , ≡ ) where � � G := , 23 / 40

  50. Operads and presentations Let O be an operad. A congruence of O is an equivalence relation ≡ on O preserving the arities and such that x ≡ x ′ and y ≡ y ′ imply x ◦ i y ≡ x ′ ◦ i y ′ for all i ∈ [ | x | ] . A presentation of O is a pair ( G , ≡ ) such that G is an alphabet and ≡ is a congruence of O satisfying O ≃ S ( G ) / ≡ . — Example — The operad Motz admits the presentation ( G , ≡ ) where � � G := , and ≡ is the smallest operad congruence satisfying ◦ 1 ≡ ◦ 2 , ◦ 1 ≡ ◦ 2 , ◦ 1 ≡ ◦ 3 , ◦ 1 ≡ ◦ 3 . 23 / 40

  51. Operads and paterns Let O be an operad admiting a presentation ( G , ≡ ) . 24 / 40

  52. Operads and paterns Let O be an operad admiting a presentation ( G , ≡ ) . A basis of O is a subset B of S ( G ) such that for any [ t ] ≡ ∈ S ( G ) / ≡ , there exists a unique s ∈ [ t ] ≡ ∩ B . 24 / 40

  53. Operads and paterns Let O be an operad admiting a presentation ( G , ≡ ) . A basis of O is a subset B of S ( G ) such that for any [ t ] ≡ ∈ S ( G ) / ≡ , there exists a unique s ∈ [ t ] ≡ ∩ B . In most cases, B can be described as set of G -trees avoiding a subset P B of S ( G ) . 24 / 40

  54. Operads and paterns Let O be an operad admiting a presentation ( G , ≡ ) . A basis of O is a subset B of S ( G ) such that for any [ t ] ≡ ∈ S ( G ) / ≡ , there exists a unique s ∈ [ t ] ≡ ∩ B . In most cases, B can be described as set of G -trees avoiding a subset P B of S ( G ) . — Example — The set B , described as the set of G -trees avoiding � � P B := ◦ 1 ◦ 1 ◦ 1 ◦ 1 , , , , is a basis of Motz . 24 / 40

  55. Operads and paterns Let O be an operad admiting a presentation ( G , ≡ ) . A basis of O is a subset B of S ( G ) such that for any [ t ] ≡ ∈ S ( G ) / ≡ , there exists a unique s ∈ [ t ] ≡ ∩ B . In most cases, B can be described as set of G -trees avoiding a subset P B of S ( G ) . — Example — The set B , described as the set of G -trees avoiding � � P B := ◦ 1 ◦ 1 ◦ 1 ◦ 1 , , , , is a basis of Motz . Rewrite systems on G -trees are good tools to compute bases (we find terminating and confluent orientations ⇒ of ≡ ). 24 / 40

  56. Operads and enumeration Let X be a family of combinatorial objects we want enumerate. The approach using operads consists in 1. endowing X with the structure of an operad O X ; 25 / 40

  57. Operads and enumeration Let X be a family of combinatorial objects we want enumerate. The approach using operads consists in 1. endowing X with the structure of an operad O X ; 2. exhibiting a presentation ( G , ≡ ) of O X and a basis B ; 25 / 40

  58. Operads and enumeration Let X be a family of combinatorial objects we want enumerate. The approach using operads consists in 1. endowing X with the structure of an operad O X ; 2. exhibiting a presentation ( G , ≡ ) of O X and a basis B ; 3. computing the series F ( P B , ∅ ) where P B is a set of G -trees satisfying A ( P B ) = B . 25 / 40

  59. Operads and enumeration Let X be a family of combinatorial objects we want enumerate. The approach using operads consists in 1. endowing X with the structure of an operad O X ; 2. exhibiting a presentation ( G , ≡ ) of O X and a basis B ; 3. computing the series F ( P B , ∅ ) where P B is a set of G -trees satisfying A ( P B ) = B . — Example — To enumerate Motzkin paths (w.r.t. their sizes), we consider their operad structure Motz . 25 / 40

  60. Operads and enumeration Let X be a family of combinatorial objects we want enumerate. The approach using operads consists in 1. endowing X with the structure of an operad O X ; 2. exhibiting a presentation ( G , ≡ ) of O X and a basis B ; 3. computing the series F ( P B , ∅ ) where P B is a set of G -trees satisfying A ( P B ) = B . — Example — To enumerate Motzkin paths (w.r.t. their sizes), we consider their operad structure Motz . � � a c c Let a := , c := , and P := a . c a c a 25 / 40

  61. Operads and enumeration Let X be a family of combinatorial objects we want enumerate. The approach using operads consists in 1. endowing X with the structure of an operad O X ; 2. exhibiting a presentation ( G , ≡ ) of O X and a basis B ; 3. computing the series F ( P B , ∅ ) where P B is a set of G -trees satisfying A ( P B ) = B . — Example — To enumerate Motzkin paths (w.r.t. their sizes), we consider their operad structure Motz . � � a c c Let a := , c := , and P := a . c a c a We have F ( P , ∅ ) = + a ¯ ◦ [ F ( P , { a , c } ) , F ( P , ∅ )] + c ¯ ◦ [ F ( P , { a , c } ) , F ( P , ∅ ) , F ( P , ∅ )] , F ( P , { a , c } ) = , so that, the generating series of Motzkin paths satisfies F ( P , ∅ ) = z + z F ( P , ∅ ) + z F ( P , ∅ ) 2 . 25 / 40

  62. Outline Generation 26 / 40

  63. Context-free grammars Let A = V ⊔ T be a set where V is a set of variables and T is a set of terminal symbols. 27 / 40

  64. Context-free grammars Let A = V ⊔ T be a set where V is a set of variables and T is a set of terminal symbols. A rule is a pair ( x, v ) ∈ V × A ∗ . A set R of rules specifies a rewrite rule → on A ∗ by seting u x w → u v w for any u, w ∈ A ∗ provided that ( x, v ) ∈ R . 27 / 40

  65. Context-free grammars Let A = V ⊔ T be a set where V is a set of variables and T is a set of terminal symbols. A rule is a pair ( x, v ) ∈ V × A ∗ . A set R of rules specifies a rewrite rule → on A ∗ by seting u x w → u v w for any u, w ∈ A ∗ provided that ( x, v ) ∈ R . — Example — Let V := { x, y } , T := { a , b , c } , and R := { ( x, b ) , ( x, x a y ) , ( y, ac ) } . We have b xx → b x a yx → bba yx → bbaac x. 27 / 40

  66. Regular tree grammars Let V be a set of variables and T be an alphabet of terminal symbols. 28 / 40

  67. Regular tree grammars Let V be a set of variables and T be an alphabet of terminal symbols. A ( V , T ) -tree is a planar rooted tree where internal nodes are labeled on T and leaves are labeled on V . 28 / 40

  68. Regular tree grammars Let V be a set of variables and T be an alphabet of terminal symbols. A ( V , T ) -tree is a planar rooted tree where internal nodes are labeled on T and leaves are labeled on V . A rule is a pair ( x, t ) where x ∈ V and t is a ( V , T ) -tree. A set R of rules specifies a rewrite rule → on the set of all ( V , T ) -trees by seting → s s x t for any ( V , T ) -tree s having a leaf labeled by x , provided that ( x, t ) ∈ R . 28 / 40

  69. Regular tree grammars Let V be a set of variables and T be an alphabet of terminal symbols. A ( V , T ) -tree is a planar rooted tree where internal nodes are labeled on T and leaves are labeled on V . A rule is a pair ( x, t ) where x ∈ V and t is a ( V , T ) -tree. A set R of rules specifies a rewrite rule → on the set of all ( V , T ) -trees by seting → s s x t for any ( V , T ) -tree s having a leaf labeled by x , provided that ( x, t ) ∈ R . — Example — �� � �� � b Let V := { x, y } , T := { a , b } where | a | := 1 , | b | := 2 , and R := x, , y, . a x b y x y b We have b a a b → → . x x a a a b x x y x b y x 28 / 40

  70. General generation Objectives: ◮ Introduce generating systems for any kind of combinatorial objects; ◮ Retrieve the generation of words and of trees as special cases; ◮ Develop a toolbox for the enumeration of combinatorial objects. 29 / 40

  71. General generation Objectives: ◮ Introduce generating systems for any kind of combinatorial objects; ◮ Retrieve the generation of words and of trees as special cases; ◮ Develop a toolbox for the enumeration of combinatorial objects. — Key idea — Use colored operads, where ◮ colors play the role of variables and terminal symbols; ◮ Formal series on colored operad and their operations support enumeration. 29 / 40

  72. Colored operads Colored operads are algebraic structures formalizing the notion of partial operations and their composition. 30 / 40

  73. Colored operads Colored operads are algebraic structures formalizing the notion of partial operations and their composition. A colored operad is a quadruplet ( C , C , ◦ i , 1 c ) where 1. C is a finite set of colors; 30 / 40

Recommend


More recommend