Log-space computability of the conjugacy problem in wreath products - PowerPoint PPT Presentation

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Log-space computability of the conjugacy problem in wreath products Svetla Vassileva McGill University City University of New York GAGTA May 29, 2013

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Why space? • Handling large data sets. • RAM vs. external storage • DNA sequencing • working with databases • Time complexity can really be due to space issues. • Gröbner bases • Start with basis for ideal and “blow it up” by adding polynomials • The number of polynomials we add is unbounded � �� � space ⇒ the time complexity is large

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Time vs. space Fact: log-space ⊆ P-time. • P-time is not always very practical • if polynomial is more than quadratic, the algorithm is not practical • the degree of the polynomial varies with the model of computation • P-time is too large as a class • P-time has many subclasses • aim for “tighter” bound on the complexity class • log-space is “tighter” than P-time

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Log-space transducers input tape read only work tape read/write output tape write only

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Example: sorting is in log-space 12 10 11 15 latest printed p = 4 q = 1 current candidate 15

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Log-space ⇒ P-time. • Configurations cannot be repeated. • Total number of configurations ≤ k ( n + 2 c log n ) ∼ n c • P-time ? ⇒ log-space: open problem.

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Log-space functions can be composed . . . x 1 x 2 x 3 x n g : f :

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Log-space functions can be composed . . . x 1 x 2 x 3 x n f ◦ g : p = i g ( x )[ i ]

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Some log-space computable problems • WP in linear groups is log-space decidable (Zalcstein, Lipton). • Normal forms in free groups are log-space computable (Elder, Elston, Ostheimer). • Normal forms in abelian groups are log-space computable (EEO). • Normal forms in wreath products are log-space computable (EEO). • WP in Grigorchuk group is log-space decidable (EEO). • Normal forms in RAAG are log-space computable (Diekert, Kausch, Lohrey).

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Wreath products The restricted wreath product is the group: A ≀ B = { bf | b ∈ B , f ∈ A ( B ) } , with multiplication defined by bf · cg = bc f c g , where • f c ( x ) = f ( xc − 1 ) for x ∈ B . • A ( B ) is the set of all functions from B to A of finite support . • Multiplication in A ( B ) is given by f · g ( x ) = f ( x ) g ( x ) . • 1 A ( B ) is the function 1 : B → 1 A . Remark. B acts on A ( B ) , so A ≀ B ≃ B ⋉ A ( B )

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries A presentation for A ≀ B Let A = � X | R A � , B = � Y | R B � . Then � � X ∪ Y | R A , R B , [ a b 1 1 , a b 2 A ≀ B = 2 ] , where a 1 , a 2 ∈ A and b 1 , b 2 ∈ B . � a if x = b a b � f a , b ( x ) = 1 otherwise. • Any function f ∈ A ( B ) can be given as { ( b 1 , a 1 ) , . . . , ( b n , a n ) } • Equivalently, f = f a 1 , b 1 . . . f a n , b n = f b 1 a n , 1 � a b 1 a 1 , 1 . . . f b n 1 . . . a b n n .

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Normal forms in wreath products Given a word w = b 1 a 1 . . . b k a k in generators X and Y , we can rewrite it as w = bf . • w = b 1 . . . b n · a b 2 ... b k a b 3 ... b n · · · a b n n − 1 a n 1 2 • w = b · A B 1 1 . . . A B k k , where • b = b 1 . . . b n ∈ B • A 1 , . . . , A k � = 1 • B i � = B j whenever i � = j • A B 1 1 . . . A B k k can be viewed as a function f : B i �→ A i

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Conjugacy in wreath products • Let x = bf , y = cg ∈ A ≀ B be given. • There exists z = dh ∈ A ≀ B such that z − 1 xz = y iff d − 1 bd = c and g d = h b fh − 1 . • g d = h b fh − 1 ⇔ ∀ x ∈ B , g d ( x ) = h b fh − 1 ( x ) . • Problems: • ∀ x ∈ B is a lot of elements to check for (but finite support). • Get rid of h . • Get rid of d .

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries A conjugacy criterion • T = { t i } – set of � b � - coset representatives for supp ( f ) ∪ supp ( g ) • S = { s i } – set of � c � - coset representatives for supp ( f ) ∪ supp ( g ) • Define � � f ( t i b j ) and γ i ( f ) = f ( s i c j ) . β i ( f ) = j j Theorem (Matthews (modified)) In A ≀ B, bf ∼ cg if and only if • b ∼ c in B and • β i ( f ) ∼ γ i ( g ) in A for all i.

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries CP in wreath products Theorem (V.) Suppose that • the conjugacy problem in A is log-space decidable, • the conjugacy problem in B is log-space decidable and • the power problem in B is computable in log-space. Then the conjugacy problem in A ≀ B is also log-space decidable. Power problem in G : Given two words x and y in generators of G , find the smallest integer n such that x n = y .

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Direct corollaries Corollary The conjugacy problem in a wreath product of two abelian groups is log-space decidable. Example. The conjugacy problem in the lamplighter group Z ≀ Z 2 is decidable in log-space. Corollary The conjugacy problem in the wreath product F ≀ Z 2 of a free group F and a free abelian group is decidable in log-space.

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Iterated wreath products Definition The left iterated wreath product , A n ≀ B , of two groups A and B inductively as follows. • A 1 ≀ B = A ≀ B • A n ≀ B = A ≀ ( A n − 1 ≀ B ) Corollary Suppose that • the conjugacy problem in A is log-space decidable, • the conjugacy problem in B is log-space decidable and • the power problem in A and B is computable in log-space. Then the conjugacy problem in A n ≀ B is also log-space decidable.

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Free solvable groups Definition • The n th derived (commutator) subgroup of a group G is G ( n ) = [ G ( n − 1 ) , G ( n − 1 ) ] , where G ( 1 ) = G ′ = [ G , G ] = � [ g , g ′ ] | g , g ′ ∈ G � . • The free solvable group S d , r of degree d and rank r is given by � S d , r = F r F ( d ) r .

The model of computation Wreath products and their normal forms Conjugacy in wreath products Corollaries Conjugacy in free solvable groups Corollary The conjugacy problem in a free solvable group, S d , r , of fixed rank r and degree d is decidable in logarithmic space. Proof. → Z r ≀ S d − 1 , r . • The Magnus embedding is a map φ : S d , r ֒ • The Magnus embedding is a Frattini embedding, i.e., x ∼ S d , r y ⇐ ⇒ φ ( x ) ∼ Z r ≀ S d − 1 , r φ ( y ) . • Iterate the embedding to get Z r ≀ S d − 1 , r ֒ → Z r ≀ � Z r ≀ S d − 2 , r � = Z r 2 ≀ S d − 2 , r ֒ ֒ → → S d , r · · · Z r d − 1 ≀ S 1 , r = Z r d − 1 ≀ Z r . ֒ →