Hilbert C ∗ -modules over noncommutative tori Texas A & M Luef, Franz UC Berkeley 16.07.2012 Luef, Franz UC Berkeley Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert and Banach spaces A sequence gj : j ∈ J in a Hilbert space H is a frame if there exists positive constants A, B such that for all f ∈ H X Akf k2H ≤ |hf , gj i|2 ≤ Bkf k2H . (1) j∈J analysis operator: Cf = {hf , gj i : j ∈ J} Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert and Banach spaces A sequence gj : j ∈ J in a Hilbert space H is a frame if there exists positive constants A, B such that for all f ∈ H X (1) Akf k2H ≤ |hf , gj i|2 ≤ Bkf k2H . j∈J analysis operator: Cf = {hf , gj i : j ∈ J} P synthesis operator: Dc = j∈J cj gj for c = (cj )j∈J Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert and Banach spaces A sequence gj : j ∈ J in a Hilbert space H is a frame if there exists positive constants A, B such that for all f ∈ H X (1) Akf k2H ≤ |hf , gj i|2 ≤ Bkf k2H . j∈J analysis operator: Cf = {hf , gj i : j ∈ J} P synthesis operator: Dc = j∈J cj gj for c = (cj )j∈J P frame operator: Sf = DCf = j∈J hf , gj igj Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert and Banach spaces A sequence gj : j ∈ J in a Hilbert space H is a frame if there exists positive constants A, B such that for all f ∈ H X (1) Akf k2H ≤ |hf , gj i|2 ≤ Bkf k2H . j∈J analysis operator: Cf = {hf , gj i : j ∈ J} P synthesis operator: Dc = j∈J cj gj for c = (cj )j∈J P frame operator: Sf = DCf = j∈J hf , gj igj Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert and Banach spaces A sequence gj : j ∈ J in a Hilbert space H is a frame if there exists positive constants A, B such that for all f ∈ H X (1) Akf k2H ≤ |hf , gj i|2 ≤ Bkf k2H . j∈J analysis operator: Cf = {hf , gj i : j ∈ J} P synthesis operator: Dc = j∈J cj gj for c = (cj )j∈J P frame operator: Sf = DCf = j∈J hf , gj igj Basic fact: {gj : j ∈ J} is a frame for H ⇔, then the frame operator S is a positive invertible operator on H. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert and Banach spaces A sequence gj : j ∈ J in a Hilbert space H is a frame if there exists positive constants A, B such that for all f ∈ H X (1) Akf k2H ≤ |hf , gj i|2 ≤ Bkf k2H . j∈J analysis operator: Cf = {hf , gj i : j ∈ J} P synthesis operator: Dc = j∈J cj gj for c = (cj )j∈J P frame operator: Sf = DCf = j∈J hf , gj igj Basic fact: {gj : j ∈ J} is a frame for H ⇔, then the frame operator S is a positive invertible operator on H. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert and Banach spaces Another useful reformulation of the notion of frames: Let g ⊗ h be the rank one operator defined by (g ⊗ h)f = hf , g ih. Then X A · IH ≤ gj ⊗ gj ≤ B · IH , j∈J the series converges in the strong operator topology. Frames are of relevance because they allow the construction of (non-orthogonal) expansions. If {gj : j ∈ J} is a frame for H, then {S −1 gj : j ∈ J} is a frame with frame bounds B −1 , A−1 f ∈ H and X f = hf , S −1 gj igj . (2) j∈J Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert and Banach spaces Another useful reformulation of the notion of frames: Let g ⊗ h be the rank one operator defined by (g ⊗ h)f = hf , g ih. Then X A · IH ≤ gj ⊗ gj ≤ B · IH , j∈J the series converges in the strong operator topology. Frames are of relevance because they allow the construction of (non-orthogonal) expansions. If {gj : j ∈ J} is a frame for H, then {S −1 gj : j ∈ J} is a frame with frame bounds B −1 , A−1 f ∈ H and X f = hf , S −1 gj igj . (2) j∈J Luef, Franz Hilbert C ∗ -modules over noncommutative tori Banach frames Gröchenig introduced Banach frames: A sequence {gj : j ∈ J} of a Banach space B is called a Banach frame if there exists an associated sequence space Bd (J) and a continuous reconstruction operator R : Bd (J) → B such that for all f ∈ B R((hf , gj i)j∈J )) = f , C −1 kf kB ≤ k(hf , gj i)j∈J )kBd ≤ C kf kB for some constant C ≥ 1. Coorbit spaces provide a natural class of Banach frames for a wide class of function spaces. Banach-Gelfand triples Casazza, Han and Larson introduced the notion of a framing for a Banach space. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Banach frames Gröchenig introduced Banach frames: A sequence {gj : j ∈ J} of a Banach space B is called a Banach frame if there exists an associated sequence space Bd (J) and a continuous reconstruction operator R : Bd (J) → B such that for all f ∈ B R((hf , gj i)j∈J )) = f , C −1 kf kB ≤ k(hf , gj i)j∈J )kBd ≤ C kf kB for some constant C ≥ 1. Coorbit spaces provide a natural class of Banach frames for a wide class of function spaces. Banach-Gelfand triples Casazza, Han and Larson introduced the notion of a framing for a Banach space. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert C ∗ -modules Theory of frames for Hilbert C ∗ -modules due to Frank and Larson. Relevance of Hilbert C ∗ -modules for wavelets was demonstrated by Packer and Rieffel (projective multiresolution analysis). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert C ∗ -modules Theory of frames for Hilbert C ∗ -modules due to Frank and Larson. Relevance of Hilbert C ∗ -modules for wavelets was demonstrated by Packer and Rieffel (projective multiresolution analysis). Construction of equivalence bimodules between noncommutative tori is Gabor analysis. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert C ∗ -modules Theory of frames for Hilbert C ∗ -modules due to Frank and Larson. Relevance of Hilbert C ∗ -modules for wavelets was demonstrated by Packer and Rieffel (projective multiresolution analysis). Construction of equivalence bimodules between noncommutative tori is Gabor analysis. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Left Hilbert C ∗ -modules In the 1970’s Paschke and Rieffel independently generalized the notion of Hilbert spaces to so-called Hilbert C ∗ -modules. Defintion: Let A be a unital C ∗ -algebra. Then a vector space V is a left Hilbert A-module, i.e. (A, g ) 7→ A · g is a map from X × A → A, with a pairing A h., .i such that for all f , g , h ∈ V : Luef, Franz Hilbert C ∗ -modules over noncommutative tori Left Hilbert C ∗ -modules In the 1970’s Paschke and Rieffel independently generalized the notion of Hilbert spaces to so-called Hilbert C ∗ -modules. Defintion: Let A be a unital C ∗ -algebra. Then a vector space V is a left Hilbert A-module, i.e. (A, g ) 7→ A · g is a map from X × A → A, with a pairing A h., .i such that for all f , g , h ∈ V : (a) A hλg + µh, ki = λ A hg , ki + µ A hh, ki for all λ, µ ∈ C; Luef, Franz Hilbert C ∗ -modules over noncommutative tori Left Hilbert C ∗ -modules In the 1970’s Paschke and Rieffel independently generalized the notion of Hilbert spaces to so-called Hilbert C ∗ -modules. Defintion: Let A be a unital C ∗ -algebra. Then a vector space V is a left Hilbert A-module, i.e. (A, g ) 7→ A · g is a map from X × A → A, with a pairing A h., .i such that for all f , g , h ∈ V : (a) A hλg (b) A hA + µh, ki = λ A hg , ki + µ A hh, ki for all λ, µ ∈ C; · f , g i = A · A hf , g i for all A ∈ A; Luef, Franz Hilbert C ∗ -modules over noncommutative tori Left Hilbert C ∗ -modules In the 1970’s Paschke and Rieffel independently generalized the notion of Hilbert spaces to so-called Hilbert C ∗ -modules. Defintion: Let A be a unital C ∗ -algebra. Then a vector space V is a left Hilbert A-module, i.e. (A, g ) 7→ A · g is a map from X × A → A, with a pairing A h., .i such that for all f , g , h ∈ V : (a) A hλg + µh, ki = λ A hg , ki + µ A hh, ki for all λ, µ ∈ C; (b) A hA (c) A hf , g i · f , g i = A · A hf , g i for all A ∈ A; = A hg , f i∗ ; Luef, Franz Hilbert C ∗ -modules over noncommutative tori Left Hilbert C ∗ -modules In the 1970’s Paschke and Rieffel independently generalized the notion of Hilbert spaces to so-called Hilbert C ∗ -modules. Defintion: Let A be a unital C ∗ -algebra. Then a vector space V is a left Hilbert A-module, i.e. (A, g ) 7→ A · g is a map from X × A → A, with a pairing A h., .i such that for all f , g , h ∈ V : (a) A hλg + µh, ki = λ A hg , ki + µ A hh, ki for all λ, µ ∈ C; (b) A hA (c) A hf , g i = A hg , f i∗ ; (d) A hf , f i ≥ 0. · f , g i = A · A hf , g i for all A ∈ A; Luef, Franz Hilbert C ∗ -modules over noncommutative tori Left Hilbert C ∗ -modules In the 1970’s Paschke and Rieffel independently generalized the notion of Hilbert spaces to so-called Hilbert C ∗ -modules. Defintion: Let A be a unital C ∗ -algebra. Then a vector space V is a left Hilbert A-module, i.e. (A, g ) 7→ A · g is a map from X × A → A, with a pairing A h., .i such that for all f , g , h ∈ V : (a) A hλg + µh, ki = λ A hg , ki + µ A hh, ki for all λ, µ ∈ C; (b) A hA (c) A hf , g i = A hg , f i∗ ; (d) A hf , f i ≥ 0. · f , g i = A · A hf , g i for all A ∈ A; (e) V is complete with respect to the norm 1/2 . If the ideal A kf k := kA hf , f ik span{A hf , g i : f , g ∈ V } is dense in A, then V is called full. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Left Hilbert C ∗ -modules In the 1970’s Paschke and Rieffel independently generalized the notion of Hilbert spaces to so-called Hilbert C ∗ -modules. Defintion: Let A be a unital C ∗ -algebra. Then a vector space V is a left Hilbert A-module, i.e. (A, g ) 7→ A · g is a map from X × A → A, with a pairing A h., .i such that for all f , g , h ∈ V : (a) A hλg + µh, ki = λ A hg , ki + µ A hh, ki for all λ, µ ∈ C; (b) A hA (c) A hf , g i = A hg , f i∗ ; (d) A hf , f i ≥ 0. · f , g i = A · A hf , g i for all A ∈ A; (e) V is complete with respect to the norm 1/2 . If the ideal A kf k := kA hf , f ik span{A hf , g i : f , g ∈ V } is dense in A, then V is called full. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Right Hilbert C ∗ -modules Defintion: Let B be a unital C ∗ -algebra. Then a vector space V is a right Hilbert B-module, i.e. (g , B) 7→ g · B is a map from B × V → B, with a pairing h., .iB which is linear in the second variable, such that for all f , g , h ∈ V : (a) hf , λg + µhiB = λhf , g iB + µhf , hiB for all λ, µ ∈ C; Luef, Franz Hilbert C ∗ -modules over noncommutative tori Right Hilbert C ∗ -modules Defintion: Let B be a unital C ∗ -algebra. Then a vector space V is a right Hilbert B-module, i.e. (g , B) 7→ g · B is a map from B × V → B, with a pairing h., .iB which is linear in the second variable, such that for all f , g , h ∈ V : (a) hf , λg + µhiB = λhf , g iB + µhf , hiB for all λ, µ ∈ C; (b) h·f , g · Bi = hf , g iB B for all B ∈ B; Luef, Franz Hilbert C ∗ -modules over noncommutative tori Right Hilbert C ∗ -modules Defintion: Let B be a unital C ∗ -algebra. Then a vector space V is a right Hilbert B-module, i.e. (g , B) 7→ g · B is a map from B × V → B, with a pairing h., .iB which is linear in the second variable, such that for all f , g , h ∈ V : (a) hf , λg + µhiB = λhf , g iB + µhf , hiB for all λ, µ ∈ C; (b) h·f , g · Bi = hf , g iB B for all B ∈ B; (c) hf , g iB = hg , f i∗B ; Luef, Franz Hilbert C ∗ -modules over noncommutative tori Right Hilbert C ∗ -modules Defintion: Let B be a unital C ∗ -algebra. Then a vector space V is a right Hilbert B-module, i.e. (g , B) 7→ g · B is a map from B × V → B, with a pairing h., .iB which is linear in the second variable, such that for all f , g , h ∈ V : (a) hf , λg + µhiB = λhf , g iB + µhf , hiB for all λ, µ ∈ C; (b) h·f , g · Bi = hf , g iB B for all B ∈ B; (c) hf , g iB = hg , f i∗B ; (d) hf , f iB ≥ 0. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Right Hilbert C ∗ -modules Defintion: Let B be a unital C ∗ -algebra. Then a vector space V is a right Hilbert B-module, i.e. (g , B) 7→ g · B is a map from B × V → B, with a pairing h., .iB which is linear in the second variable, such that for all f , g , h ∈ V : (a) hf , λg + µhiB = λhf , g iB + µhf , hiB for all λ, µ ∈ C; (b) h·f , g · Bi = hf , g iB B for all B ∈ B; (c) hf , g iB = hg , f i∗B ; (d) hf , f iB ≥ 0. (e) V is complete with respect to the norm kf kB := khf , f iB k1/2 . If the ideal span{hf , g iB : f , g ∈ V } is dense in B, then V is called full. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Right Hilbert C ∗ -modules Defintion: Let B be a unital C ∗ -algebra. Then a vector space V is a right Hilbert B-module, i.e. (g , B) 7→ g · B is a map from B × V → B, with a pairing h., .iB which is linear in the second variable, such that for all f , g , h ∈ V : (a) hf , λg + µhiB = λhf , g iB + µhf , hiB for all λ, µ ∈ C; (b) h·f , g · Bi = hf , g iB B for all B ∈ B; (c) hf , g iB = hg , f i∗B ; (d) hf , f iB ≥ 0. (e) V is complete with respect to the norm kf kB := khf , f iB k1/2 . If the ideal span{hf , g iB : f , g ∈ V } is dense in B, then V is called full. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Right Hilbert C ∗ -modules Defintion: Let B be a unital C ∗ -algebra. Then a vector space V is a right Hilbert B-module, i.e. (g , B) 7→ g · B is a map from B × V → B, with a pairing h., .iB which is linear in the second variable, such that for all f , g , h ∈ V : (a) hf , λg + µhiB = λhf , g iB + µhf , hiB for all λ, µ ∈ C; (b) h·f , g · Bi = hf , g iB B for all B ∈ B; (c) hf , g iB = hg , f i∗B ; (d) hf , f iB ≥ 0. (e) V is complete with respect to the norm kf kB := khf , f iB k1/2 . If the ideal span{hf , g iB : f , g ∈ V } is dense in B, then V is called full. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Hilbert C ∗ -modules In applications one often is in the situation that one has a A0 -module V0 , where A0 is a dense involutive subalgebra of A, and that there is a pairing h., .i0 from V0 × V0 → A0 satisfying all the conditions of a Hilbert C ∗ -module, when positivity is considered with respect to A. Then one can complete V0 to obtain a Hilbert C ∗ -module V over A. The challenge in the construction of Hilbert C ∗ -modules is to find appropriate function spaces V0 . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Hilbert C ∗ -modules In applications one often is in the situation that one has a A0 -module V0 , where A0 is a dense involutive subalgebra of A, and that there is a pairing h., .i0 from V0 × V0 → A0 satisfying all the conditions of a Hilbert C ∗ -module, when positivity is considered with respect to A. Then one can complete V0 to obtain a Hilbert C ∗ -module V over A. The challenge in the construction of Hilbert C ∗ -modules is to find appropriate function spaces V0 . Hilbert C ∗ -modules are also Banach modules for the Hilbert C ∗ -module norm, i.e. A kA · g k ≤ kAk A kg k. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Hilbert C ∗ -modules In applications one often is in the situation that one has a A0 -module V0 , where A0 is a dense involutive subalgebra of A, and that there is a pairing h., .i0 from V0 × V0 → A0 satisfying all the conditions of a Hilbert C ∗ -module, when positivity is considered with respect to A. Then one can complete V0 to obtain a Hilbert C ∗ -module V over A. The challenge in the construction of Hilbert C ∗ -modules is to find appropriate function spaces V0 . Hilbert C ∗ -modules are also Banach modules for the Hilbert C ∗ -module norm, i.e. A kA · g k ≤ kAk A kg k. In some cases the structures of left and right Hilbert C ∗ -modules over two C ∗ -algebras A and B are compatible with each other. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Hilbert C ∗ -modules In applications one often is in the situation that one has a A0 -module V0 , where A0 is a dense involutive subalgebra of A, and that there is a pairing h., .i0 from V0 × V0 → A0 satisfying all the conditions of a Hilbert C ∗ -module, when positivity is considered with respect to A. Then one can complete V0 to obtain a Hilbert C ∗ -module V over A. The challenge in the construction of Hilbert C ∗ -modules is to find appropriate function spaces V0 . Hilbert C ∗ -modules are also Banach modules for the Hilbert C ∗ -module norm, i.e. A kA · g k ≤ kAk A kg k. In some cases the structures of left and right Hilbert C ∗ -modules over two C ∗ -algebras A and B are compatible with each other. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Maps on Hilbert C ∗ -modules Suppose V is a Hilbert A-module. Then a module mapping T : V → V is adjointable, if there is a mapping T ∗ : V → V such that A hTf , g i = A hf , T ∗ g i for all f , g ∈ V . L(V ) denotes the space of all adjointable mappings on V . We define the norm of T by kT k = sup{A kTg k : A kg k ≤ 1}. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Maps on Hilbert C ∗ -modules Suppose V is a Hilbert A-module. Then a module mapping T : V → V is adjointable, if there is a mapping T ∗ : V → V such that A hTf , g i = A hf , T ∗ g i for all f , g ∈ V . L(V ) denotes the space of all adjointable mappings on V . We define the norm of T by kT k = sup{A kTg k : A kg k ≤ 1}. L(V ) is a C ∗ -algebra with respect to this norm. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Maps on Hilbert C ∗ -modules Suppose V is a Hilbert A-module. Then a module mapping T : V → V is adjointable, if there is a mapping T ∗ : V → V such that A hTf , g i = A hf , T ∗ g i for all f , g ∈ V . L(V ) denotes the space of all adjointable mappings on V . We define the norm of T by kT k = sup{A kTg k : A kg k ≤ 1}. L(V ) is a C ∗ -algebra with respect to this norm. rank one operators ΘA g ,h f = A hf , g i · h are adjointable operators. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Maps on Hilbert C ∗ -modules Suppose V is a Hilbert A-module. Then a module mapping T : V → V is adjointable, if there is a mapping T ∗ : V → V such that A hTf , g i = A hf , T ∗ g i for all f , g ∈ V . L(V ) denotes the space of all adjointable mappings on V . We define the norm of T by kT k = sup{A kTg k : A kg k ≤ 1}. L(V ) is a C ∗ -algebra with respect to this norm. rank one operators ΘA g ,h f = A hf , g i · h are adjointable operators. The closed linear subspace of L(V ) spanned by {ΘA g ,h : f , g ∈ V } is the algebra of compact A-module operators. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Maps on Hilbert C ∗ -modules Suppose V is a Hilbert A-module. Then a module mapping T : V → V is adjointable, if there is a mapping T ∗ : V → V such that A hTf , g i = A hf , T ∗ g i for all f , g ∈ V . L(V ) denotes the space of all adjointable mappings on V . We define the norm of T by kT k = sup{A kTg k : A kg k ≤ 1}. L(V ) is a C ∗ -algebra with respect to this norm. rank one operators ΘA g ,h f = A hf , g i · h are adjointable operators. The closed linear subspace of L(V ) spanned by {ΘA g ,h : f , g ∈ V } is the algebra of compact A-module operators. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Properties of rank one Hilbert module operators A A 0 0 ΘA g ,h Θg 0 ,h0 = Θk,h for k = A hg , h ig ∗ A (ΘA g ,h ) = Θh,g Luef, Franz Hilbert C ∗ -modules over noncommutative tori Properties of rank one Hilbert module operators A A 0 0 ΘA g ,h Θg 0 ,h0 = Θk,h for k = A hg , h ig ∗ A (ΘA g ,h ) = Θh,g A For T ∈ L(V ) we have T ΘA g ,h = Θg ,Th . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Properties of rank one Hilbert module operators A A 0 0 ΘA g ,h Θg 0 ,h0 = Θk,h for k = A hg , h ig ∗ A (ΘA g ,h ) = Θh,g A For T ∈ L(V ) we have T ΘA g ,h = Θg ,Th . kΘA g ,h k ≤ A kg kA khk. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Properties of rank one Hilbert module operators A A 0 0 ΘA g ,h Θg 0 ,h0 = Θk,h for k = A hg , h ig ∗ A (ΘA g ,h ) = Θh,g A For T ∈ L(V ) we have T ΘA g ,h = Θg ,Th . kΘA g ,h k ≤ A kg kA khk. Suppose A is unital and A hg , g i = I , then kΘA g ,h k = A khk. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Properties of rank one Hilbert module operators A A 0 0 ΘA g ,h Θg 0 ,h0 = Θk,h for k = A hg , h ig ∗ A (ΘA g ,h ) = Θh,g A For T ∈ L(V ) we have T ΘA g ,h = Θg ,Th . kΘA g ,h k ≤ A kg kA khk. Suppose A is unital and A hg , g i = I , then kΘA g ,h k = A khk. We denote the unit sphere of A V by S(A V ) = {g ∈ A V : A hg , g i = I }. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Properties of rank one Hilbert module operators A A 0 0 ΘA g ,h Θg 0 ,h0 = Θk,h for k = A hg , h ig ∗ A (ΘA g ,h ) = Θh,g A For T ∈ L(V ) we have T ΘA g ,h = Θg ,Th . kΘA g ,h k ≤ A kg kA khk. Suppose A is unital and A hg , g i = I , then kΘA g ,h k = A khk. We denote the unit sphere of A V by S(A V ) = {g ∈ A V : A hg , g i = I }. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert C ∗ -modules For the sake of simplicity I restrict my discussion to the case of finitely generated Hilbert C ∗ -modules. Let A be a unital C ∗ -algebra. A sequence {gj : j = 1, ..., n} in a (left) Hilbert A-module A V is called a standard module frame if there are positive reals C , D such that n X C A hf , f i ≤ A hf , gj iA hgj , f i ≤ D A hf , f i j=1 for each f ∈ A V . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert C ∗ -modules For the sake of simplicity I restrict my discussion to the case of finitely generated Hilbert C ∗ -modules. Let A be a unital C ∗ -algebra. A sequence {gj : j = 1, ..., n} in a (left) Hilbert A-module A V is called a standard module frame if there are positive reals C , D such that n X C A hf , f i ≤ A hf , gj iA hgj , f i ≤ D A hf , f i j=1 for each f ∈ A V . Let A be a unital C ∗ -algebra. A sequence {gj : j = 1, ..., n} in a (left) Hilbert A-module A V is a standard module frame if the reconstruction formula n X f = A hf , gj i · gj for all f ∈ A V . j=1 Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert C ∗ -modules For the sake of simplicity I restrict my discussion to the case of finitely generated Hilbert C ∗ -modules. Let A be a unital C ∗ -algebra. A sequence {gj : j = 1, ..., n} in a (left) Hilbert A-module A V is called a standard module frame if there are positive reals C , D such that n X C A hf , f i ≤ A hf , gj iA hgj , f i ≤ D A hf , f i j=1 for each f ∈ A V . Let A be a unital C ∗ -algebra. A sequence {gj : j = 1, ..., n} in a (left) Hilbert A-module A V is a standard module frame if the reconstruction formula n X f = A hf , gj i · gj for all f ∈ A V . j=1 Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert C ∗ -module The existence of a standard module frame of finite cardinality is equivalent to A V being a projective module, i.e it can be embedded into An as a direct summand. In terms of rank one Hilbert A-module operators the definition of a frame becomes: C A hf , f i ≤ n X A A hΘgj ,gj f , f iA ≤ D A hf , f i. j=1 Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert C ∗ -module The existence of a standard module frame of finite cardinality is equivalent to A V being a projective module, i.e it can be embedded into An as a direct summand. In terms of rank one Hilbert A-module operators the definition of a frame becomes: C A hf , f i ≤ n X A A hΘgj ,gj f , f iA ≤ D A hf , f i. j=1 One has the following: A sequence {gj : j = 1, ..., n} in a (left) A-module A V is a frame if and only if Pn Hilbert A Θ A gj ,gj converges in the strict topology to a j=1 bounded invertible operator in L(A V ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert C ∗ -module The existence of a standard module frame of finite cardinality is equivalent to A V being a projective module, i.e it can be embedded into An as a direct summand. In terms of rank one Hilbert A-module operators the definition of a frame becomes: C A hf , f i ≤ n X A A hΘgj ,gj f , f iA ≤ D A hf , f i. j=1 One has the following: A sequence {gj : j = 1, ..., n} in a (left) A-module A V is a frame if and only if Pn Hilbert A Θ A gj ,gj converges in the strict topology to a j=1 bounded invertible operator in L(A V ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert C ∗ -module Suppose A V is a finitely generated Hilbert A-module. Then any set of generators {gj : j = 1, ..., n} is a standard module frame. The number of the shortest frame gives the number of factors of An into which A V is embeddable. In other P words, the positive module operator S = nj=1 ΘA gj ,gj is invertible and the upper and lower frame bounds are given by kSk2 and kS −1 k−2 . In particular, A V is singly generated if ΘA g ,g is invertible. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert C ∗ -module Suppose A V is a finitely generated Hilbert A-module. Then any set of generators {gj : j = 1, ..., n} is a standard module frame. The number of the shortest frame gives the number of factors of An into which A V is embeddable. In other P words, the positive module operator S = nj=1 ΘA gj ,gj is invertible and the upper and lower frame bounds are given by kSk2 and kS −1 k−2 . In particular, A V is singly generated if ΘA g ,g is invertible. Is there a way to generate finitely generated Hilbert C ∗ -modules? Luef, Franz Hilbert C ∗ -modules over noncommutative tori Frames for Hilbert C ∗ -module Suppose A V is a finitely generated Hilbert A-module. Then any set of generators {gj : j = 1, ..., n} is a standard module frame. The number of the shortest frame gives the number of factors of An into which A V is embeddable. In other P words, the positive module operator S = nj=1 ΘA gj ,gj is invertible and the upper and lower frame bounds are given by kSk2 and kS −1 k−2 . In particular, A V is singly generated if ΘA g ,g is invertible. Is there a way to generate finitely generated Hilbert C ∗ -modules? Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita-Rieffel equivalence Rieffel introduced in 1970’s the notion of strong Morita equivalence for C ∗ -algebras: Definition: Let A and B be C ∗ -algebras. Then an A-B-equivalence bimodule A V B is an A-B-bimodule such that: Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita-Rieffel equivalence Rieffel introduced in 1970’s the notion of strong Morita equivalence for C ∗ -algebras: Definition: Let A and B be C ∗ -algebras. Then an A-B-equivalence bimodule A V B is an A-B-bimodule such that: (a) is a full left Hilbert A-module and a full right Hilbert B-module; AV B Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita-Rieffel equivalence Rieffel introduced in 1970’s the notion of strong Morita equivalence for C ∗ -algebras: Definition: Let A and B be C ∗ -algebras. Then an A-B-equivalence bimodule A V B is an A-B-bimodule such that: (a) is a full left Hilbert A-module and a full right Hilbert B-module; AV B (b) for all f , g ∈ A V B , A ∈ A and B ∈ B we have that hA · f , g iB = hf , A∗ · g iB and A hf · B, g i = A hf , g · B ∗ i; Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita-Rieffel equivalence Rieffel introduced in 1970’s the notion of strong Morita equivalence for C ∗ -algebras: Definition: Let A and B be C ∗ -algebras. Then an A-B-equivalence bimodule A V B is an A-B-bimodule such that: (a) is a full left Hilbert A-module and a full right Hilbert B-module; AV B (b) for all f , g ∈ A V B , A ∈ A and B ∈ B we have that hA · f , g iB = hf , A∗ · g iB and A hf · B, g i = A hf , g · B ∗ i; (c) for all f , g , h ∈ A V B , A hf , g i · h = f · hg , hiB . The C ∗ -algebras A and B are called Morita-Rieffel equivalent if there exists an A − B equivalence bimodule. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita-Rieffel equivalence Rieffel introduced in 1970’s the notion of strong Morita equivalence for C ∗ -algebras: Definition: Let A and B be C ∗ -algebras. Then an A-B-equivalence bimodule A V B is an A-B-bimodule such that: (a) is a full left Hilbert A-module and a full right Hilbert B-module; AV B (b) for all f , g ∈ A V B , A ∈ A and B ∈ B we have that hA · f , g iB = hf , A∗ · g iB and A hf · B, g i = A hf , g · B ∗ i; (c) for all f , g , h ∈ A V B , A hf , g i · h = f · hg , hiB . The C ∗ -algebras A and B are called Morita-Rieffel equivalent if there exists an A − B equivalence bimodule. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita-Rieffel equivalence - Consequences In words, Condition (b) says that A acts by adjointable operators on VB and that B acts by adjointable operators on A V , Condition (c) is an associativity condition between the A-inner product and the B-inner product. ΘB g ,h f = f · hg , hiB is a rank one B-module operator. Consequently, the invertiblity of ΘA g ,g is equivalent to the B invertiblity of Θg ,g . Note that this amounts to ΘB g ,g f = f · hg , g iB , i.e. f and g are “uncoupled”. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita-Rieffel equivalence - Consequences In words, Condition (b) says that A acts by adjointable operators on VB and that B acts by adjointable operators on A V , Condition (c) is an associativity condition between the A-inner product and the B-inner product. ΘB g ,h f = f · hg , hiB is a rank one B-module operator. Consequently, the invertiblity of ΘA g ,g is equivalent to the B invertiblity of Θg ,g . Note that this amounts to ΘB g ,g f = f · hg , g iB , i.e. f and g are “uncoupled”. If A and B are Morita-Rieffel equivalent, then A VB is a projective left A-module, and a projective right B-module. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita-Rieffel equivalence - Consequences In words, Condition (b) says that A acts by adjointable operators on VB and that B acts by adjointable operators on A V , Condition (c) is an associativity condition between the A-inner product and the B-inner product. ΘB g ,h f = f · hg , hiB is a rank one B-module operator. Consequently, the invertiblity of ΘA g ,g is equivalent to the B invertiblity of Θg ,g . Note that this amounts to ΘB g ,g f = f · hg , g iB , i.e. f and g are “uncoupled”. If A and B are Morita-Rieffel equivalent, then A VB is a projective left A-module, and a projective right B-module.There exist g1 , ..., gn ∈ A VB such that n n X X f = hf , g i · g = f · hgj , gj iB A j j j=1 Luef, Franz j=1 Hilbert C ∗ -modules over noncommutative tori Morita-Rieffel equivalence - Consequences In words, Condition (b) says that A acts by adjointable operators on VB and that B acts by adjointable operators on A V , Condition (c) is an associativity condition between the A-inner product and the B-inner product. ΘB g ,h f = f · hg , hiB is a rank one B-module operator. Consequently, the invertiblity of ΘA g ,g is equivalent to the B invertiblity of Θg ,g . Note that this amounts to ΘB g ,g f = f · hg , g iB , i.e. f and g are “uncoupled”. If A and B are Morita-Rieffel equivalent, then A VB is a projective left A-module, and a projective right B-module.There exist g1 , ..., gn ∈ A VB such that n n X X f = hf , g i · g = f · hgj , gj iB A j j j=1 Luef, Franz j=1 Hilbert C ∗ -modules over noncommutative tori Morita-Rieffel equivalence - Consequences The compact A-module operators are isomorphic to B. Suppose A has a normalized trace trA . Then we can introduce a trace trB on B by trB (hf , g iB ) = trA (A hg , f i). trB is not normalized and trB (I ) = dim(A V ), the dimension of the Hilbert A-module A V . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita equivalent spectrally invariant subalgebras Recall that a unital subalgebra A of a unital C ∗ -algebra B with common unit is called spectrally invariant, if for A ∈ A with A−1 ∈ B one actually has that A−1 ∈ A. Theorem: Let A and B be unital C ∗ -algebras that are Morita equivalent via an equivalence bimodule A VB . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita equivalent spectrally invariant subalgebras Recall that a unital subalgebra A of a unital C ∗ -algebra B with common unit is called spectrally invariant, if for A ∈ A with A−1 ∈ B one actually has that A−1 ∈ A. Theorem: Let A and B be unital C ∗ -algebras that are Morita equivalent via an equivalence bimodule A VB .Suppose we have dense ∗-Banach (or Frechet) subalgebras A0 and B0 of A and B respectively containing the identity elements. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita equivalent spectrally invariant subalgebras Recall that a unital subalgebra A of a unital C ∗ -algebra B with common unit is called spectrally invariant, if for A ∈ A with A−1 ∈ B one actually has that A−1 ∈ A. Theorem: Let A and B be unital C ∗ -algebras that are Morita equivalent via an equivalence bimodule A VB .Suppose we have dense ∗-Banach (or Frechet) subalgebras A0 and B0 of A and B respectively containing the identity elements.Furthermore we assume that A0 and B0 are spectrally invariant in A and B respectively. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita equivalent spectrally invariant subalgebras Recall that a unital subalgebra A of a unital C ∗ -algebra B with common unit is called spectrally invariant, if for A ∈ A with A−1 ∈ B one actually has that A−1 ∈ A. Theorem: Let A and B be unital C ∗ -algebras that are Morita equivalent via an equivalence bimodule A VB .Suppose we have dense ∗-Banach (or Frechet) subalgebras A0 and B0 of A and B respectively containing the identity elements.Furthermore we assume that A0 and B0 are spectrally invariant in A and B respectively.Let V0 be a dense subspace of A VB which is closed under the actions of A0 and B0 , and such that the restrictions of the inner products A h., .i and h., .iB have values in A0 and B0 respectively. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita equivalent spectrally invariant subalgebras Recall that a unital subalgebra A of a unital C ∗ -algebra B with common unit is called spectrally invariant, if for A ∈ A with A−1 ∈ B one actually has that A−1 ∈ A. Theorem: Let A and B be unital C ∗ -algebras that are Morita equivalent via an equivalence bimodule A VB .Suppose we have dense ∗-Banach (or Frechet) subalgebras A0 and B0 of A and B respectively containing the identity elements.Furthermore we assume that A0 and B0 are spectrally invariant in A and B respectively.Let V0 be a dense subspace of A VB which is closed under the actions of A0 and B0 , and such that the restrictions of the inner products A h., .i and h., .iB have values in A0 and B0 respectively.Then V0 is a finitely generated projective left A0 -module. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita equivalent spectrally invariant subalgebras Recall that a unital subalgebra A of a unital C ∗ -algebra B with common unit is called spectrally invariant, if for A ∈ A with A−1 ∈ B one actually has that A−1 ∈ A. Theorem: Let A and B be unital C ∗ -algebras that are Morita equivalent via an equivalence bimodule A VB .Suppose we have dense ∗-Banach (or Frechet) subalgebras A0 and B0 of A and B respectively containing the identity elements.Furthermore we assume that A0 and B0 are spectrally invariant in A and B respectively.Let V0 be a dense subspace of A VB which is closed under the actions of A0 and B0 , and such that the restrictions of the inner products A h., .i and h., .iB have values in A0 and B0 respectively.Then V0 is a finitely generated projective left A0 -module.In addition we have that V0 is a finitely generated projective right B0 -module. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita equivalent spectrally invariant subalgebras Recall that a unital subalgebra A of a unital C ∗ -algebra B with common unit is called spectrally invariant, if for A ∈ A with A−1 ∈ B one actually has that A−1 ∈ A. Theorem: Let A and B be unital C ∗ -algebras that are Morita equivalent via an equivalence bimodule A VB .Suppose we have dense ∗-Banach (or Frechet) subalgebras A0 and B0 of A and B respectively containing the identity elements.Furthermore we assume that A0 and B0 are spectrally invariant in A and B respectively.Let V0 be a dense subspace of A VB which is closed under the actions of A0 and B0 , and such that the restrictions of the inner products A h., .i and h., .iB have values in A0 and B0 respectively.Then V0 is a finitely generated projective left A0 -module.In addition we have that V0 is a finitely generated projective right B0 -module. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Stone-von Neumann theorem Rieffel demonstrated that the Stone-von Neumann theorem on the uniqueness of the irreducible representations of the Heisenberg group is equivalent to the following fact: Theorem: The C ∗ -algebra of compact operators on a separable Hilbert space H is Morita-Rieffel equivalent to C. We construct an equivalence bimodule using basic facts and notions from time-frequency analysis. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Stone-von Neumann theorem Rieffel demonstrated that the Stone-von Neumann theorem on the uniqueness of the irreducible representations of the Heisenberg group is equivalent to the following fact: Theorem: The C ∗ -algebra of compact operators on a separable Hilbert space H is Morita-Rieffel equivalent to C. We construct an equivalence bimodule using basic facts and notions from time-frequency analysis. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Schrödinger representation translation Tx f (t) = f (t − x) for x ∈ Rd , modulation bd Mω f (t) = e 2πit·ω f (t) for ω ∈ R time-frequency shift π(x, ω)f (t) = Mω Tx f (t) for bd (x, ω) ∈ Rd × R Mω Tx = e 2πix·ω Tx Mω π(x + y , ω + η) = e −2πix·η π(x, ω)π(y , η) π(x, ω)π(y , η) = e 2πi(y ·ω−x·η) π(y , η)π(x, ω) Luef, Franz Hilbert C ∗ -modules over noncommutative tori Schrödinger representation translation Tx f (t) = f (t − x) for x ∈ Rd , modulation bd Mω f (t) = e 2πit·ω f (t) for ω ∈ R time-frequency shift π(x, ω)f (t) = Mω Tx f (t) for bd (x, ω) ∈ Rd × R Mω Tx = e 2πix·ω Tx Mω π(x + y , ω + η) = e −2πix·η π(x, ω)π(y , η) π(x, ω)π(y , η) = e 2πi(y ·ω−x·η) π(y , η)π(x, ω) csymp (x, ω), (y, η) = c (x, ω), (y , η) c (y , η), (x, ω) c (x, ω), (y , η) = e 2πiy ·ω Luef, Franz Hilbert C ∗ -modules over noncommutative tori Schrödinger representation translation Tx f (t) = f (t − x) for x ∈ Rd , modulation bd Mω f (t) = e 2πit·ω f (t) for ω ∈ R time-frequency shift π(x, ω)f (t) = Mω Tx f (t) for bd (x, ω) ∈ Rd × R Mω Tx = e 2πix·ω Tx Mω π(x + y , ω + η) = e −2πix·η π(x, ω)π(y , η) π(x, ω)π(y , η) = e 2πi(y ·ω−x·η) π(y , η)π(x, ω) csymp (x, ω), (y, η) = c (x, ω), (y , η) c (y , η), (x, ω) c (x, ω), (y , η) = e 2πiy ·ω Short-Time Fourier Transform: Vg f (x, ω) = hf , π(x, ω)g i. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Schrödinger representation translation Tx f (t) = f (t − x) for x ∈ Rd , modulation bd Mω f (t) = e 2πit·ω f (t) for ω ∈ R time-frequency shift π(x, ω)f (t) = Mω Tx f (t) for bd (x, ω) ∈ Rd × R Mω Tx = e 2πix·ω Tx Mω π(x + y , ω + η) = e −2πix·η π(x, ω)π(y , η) π(x, ω)π(y , η) = e 2πi(y ·ω−x·η) π(y , η)π(x, ω) csymp (x, ω), (y, η) = c (x, ω), (y , η) c (y , η), (x, ω) c (x, ω), (y , η) = e 2πiy ·ω Short-Time Fourier Transform: Vg f (x, ω) = hf , π(x, ω)g i. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Integrated Schrödinger Representation 1 (L1 (R2d ), c) is an involutive Banach algebra: ZZ (F ]G )(z) = F (z 0 )G (z − z 0 )c(z, z − z 0 )dz F ∗ (z) = c(z, z)F (z). 2 (x, ω) 7→ Vω Ux is a non-degenerate faithful involutive 1 2d representation RR πL1 of (L (R ), c), πL1 (F ) = F (x, ω)Vω Ux dxdω. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Integrated Schrödinger Representation 1 (L1 (R2d ), c) is an involutive Banach algebra: ZZ (F ]G )(z) = F (z 0 )G (z − z 0 )c(z, z − z 0 )dz F ∗ (z) = c(z, z)F (z). 2 3 (x, ω) 7→ Vω Ux is a non-degenerate faithful involutive 1 2d representation RR πL1 of (L (R ), c), πL1 (F ) = F (x, ω)Vω Ux dxdω. πL1 (F )πL1 (G ) = πL1 (F ]G ) and πL1 (F )∗ = πL1 (F ∗ ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Integrated Schrödinger Representation 1 (L1 (R2d ), c) is an involutive Banach algebra: ZZ (F ]G )(z) = F (z 0 )G (z − z 0 )c(z, z − z 0 )dz F ∗ (z) = c(z, z)F (z). 2 3 4 (x, ω) 7→ Vω Ux is a non-degenerate faithful involutive 1 2d representation RR πL1 of (L (R ), c), πL1 (F ) = F (x, ω)Vω Ux dxdω. πL1 (F )πL1 (G ) = πL1 (F ]G ) and πL1 (F )∗ = πL1 (F ∗ ). RR PF = F (x, ω)Vω Ux dxdω is a projection in (L1 (R2d ), c) if and only if F ]F = F , F = F ∗ . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Integrated Schrödinger Representation 1 (L1 (R2d ), c) is an involutive Banach algebra: ZZ (F ]G )(z) = F (z 0 )G (z − z 0 )c(z, z − z 0 )dz F ∗ (z) = c(z, z)F (z). 2 3 4 (x, ω) 7→ Vω Ux is a non-degenerate faithful involutive 1 2d representation RR πL1 of (L (R ), c), πL1 (F ) = F (x, ω)Vω Ux dxdω. πL1 (F )πL1 (G ) = πL1 (F ]G ) and πL1 (F )∗ = πL1 (F ∗ ). RR PF = F (x, ω)Vω Ux dxdω is a projection in (L1 (R2d ), c) if and only if F ]F = F , F = F ∗ .A natural choice for F is Vg g . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Integrated Schrödinger Representation 1 (L1 (R2d ), c) is an involutive Banach algebra: ZZ (F ]G )(z) = F (z 0 )G (z − z 0 )c(z, z − z 0 )dz F ∗ (z) = c(z, z)F (z). 2 3 4 (x, ω) 7→ Vω Ux is a non-degenerate faithful involutive 1 2d representation RR πL1 of (L (R ), c), πL1 (F ) = F (x, ω)Vω Ux dxdω. πL1 (F )πL1 (G ) = πL1 (F ]G ) and πL1 (F )∗ = πL1 (F ∗ ). RR PF = F (x, ω)Vω Ux dxdω is a projection in (L1 (R2d ), c) if and only if F ]F = F , F = F ∗ .A natural choice for F is Vg g . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Feichtinger’s algebra Definition: Feichtinger’s algebra S0 (Rd ) = {f ∈ L2 (Rd ) : kf kS0 = RR 1 d R2d |hf , π(x, ω)g i|dxdω < ∞}, also denoted by M (R ). S0 (Rd ) is invariant under time-frequency shifts: kπ(u, η)g kS0 ≤ kg kS0 . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Feichtinger’s algebra Definition: Feichtinger’s algebra S0 (Rd ) = {f ∈ L2 (Rd ) : kf kS0 = RR 1 d R2d |hf , π(x, ω)g i|dxdω < ∞}, also denoted by M (R ). S0 (Rd ) is invariant under time-frequency shifts: kπ(u, η)g kS0 ≤ kg kS0 . S0 (Rd ) is a Banach algebra w.r.t point-wise multiplication. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Feichtinger’s algebra Definition: Feichtinger’s algebra S0 (Rd ) = {f ∈ L2 (Rd ) : kf kS0 = RR 1 d R2d |hf , π(x, ω)g i|dxdω < ∞}, also denoted by M (R ). S0 (Rd ) is invariant under time-frequency shifts: kπ(u, η)g kS0 ≤ kg kS0 . S0 (Rd ) is a Banach algebra w.r.t point-wise multiplication. S0 (R2d )RR is invariant under symplectic Fourier transform Fb (b z ) = R2d F (z)e 2πiΩ(bz ,z) dz. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Feichtinger’s algebra Definition: Feichtinger’s algebra S0 (Rd ) = {f ∈ L2 (Rd ) : kf kS0 = RR 1 d R2d |hf , π(x, ω)g i|dxdω < ∞}, also denoted by M (R ). S0 (Rd ) is invariant under time-frequency shifts: kπ(u, η)g kS0 ≤ kg kS0 . S0 (Rd ) is a Banach algebra w.r.t point-wise multiplication. S0 (R2d )RR is invariant under symplectic Fourier transform Fb (b z ) = R2d F (z)e 2πiΩ(bz ,z) dz. If f , g ∈ S0 (Rd ), then Vg f ∈ S0 (R2d ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Feichtinger’s algebra Definition: Feichtinger’s algebra S0 (Rd ) = {f ∈ L2 (Rd ) : kf kS0 = RR 1 d R2d |hf , π(x, ω)g i|dxdω < ∞}, also denoted by M (R ). S0 (Rd ) is invariant under time-frequency shifts: kπ(u, η)g kS0 ≤ kg kS0 . S0 (Rd ) is a Banach algebra w.r.t point-wise multiplication. S0 (R2d )RR is invariant under symplectic Fourier transform Fb (b z ) = R2d F (z)e 2πiΩ(bz ,z) dz. If f , g ∈ S0 (Rd ), then Vg f ∈ S0 (R2d ). Suppose f1 , f2 , g1 , g2 ∈ S0 (Rd ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Feichtinger’s algebra Definition: Feichtinger’s algebra S0 (Rd ) = {f ∈ L2 (Rd ) : kf kS0 = RR 1 d R2d |hf , π(x, ω)g i|dxdω < ∞}, also denoted by M (R ). S0 (Rd ) is invariant under time-frequency shifts: kπ(u, η)g kS0 ≤ kg kS0 . S0 (Rd ) is a Banach algebra w.r.t point-wise multiplication. S0 (R2d )RR is invariant under symplectic Fourier transform Fb (b z ) = R2d F (z)e 2πiΩ(bz ,z) dz. If f , g ∈ S0 (Rd ), then Vg f ∈ S0 (R2d ). Suppose f1 , f2 , g1 , g2 ∈ S0 (Rd ). (Vg1\ f1 · Vg2 f2 )(b z ) = (Vf2 f1 · Vg2 g1 )(b z) Luef, Franz Hilbert C ∗ -modules over noncommutative tori Feichtinger’s algebra Definition: Feichtinger’s algebra S0 (Rd ) = {f ∈ L2 (Rd ) : kf kS0 = RR 1 d R2d |hf , π(x, ω)g i|dxdω < ∞}, also denoted by M (R ). S0 (Rd ) is invariant under time-frequency shifts: kπ(u, η)g kS0 ≤ kg kS0 . S0 (Rd ) is a Banach algebra w.r.t point-wise multiplication. S0 (R2d )RR is invariant under symplectic Fourier transform Fb (b z ) = R2d F (z)e 2πiΩ(bz ,z) dz. If f , g ∈ S0 (Rd ), then Vg f ∈ S0 (R2d ). Suppose f1 , f2 , g1 , g2 ∈ S0 (Rd ). (Vg1\ f1 · Vg2 f2 )(b z ) = (Vf2 f1 · Vg2 g1 )(b z) Luef, Franz Hilbert C ∗ -modules over noncommutative tori Facts Consider G(g , R2d ) and the associated analysis and synthesis mapping ZZ f 7→ (Vg f (z)), A 7→ A(z)π(z)gdz Luef, Franz Hilbert C ∗ -modules over noncommutative tori Facts Consider G(g , R2d ) and the associated analysis and synthesis mapping ZZ f 7→ (Vg f (z)), A 7→ A(z)π(z)gdz For f , g in S0 (Rd ) and F in S0 (R2d ) these two mappings are well-defined. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Facts Consider G(g , R2d ) and the associated analysis and synthesis mapping ZZ f 7→ (Vg f (z)), A 7→ A(z)π(z)gdz For f , g in S0 (Rd ) and F in S0 (R2d ) these two mappings are well-defined. Consider the pseudodifferential operator ZZ H= A(z)π(z)dz R2d for A ∈ S0 (R2d ). Then H is trace-class. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Facts Consider G(g , R2d ) and the associated analysis and synthesis mapping ZZ f 7→ (Vg f (z)), A 7→ A(z)π(z)gdz For f , g in S0 (Rd ) and F in S0 (R2d ) these two mappings are well-defined. Consider the pseudodifferential operator ZZ H= A(z)π(z)dz R2d for A ∈ S0 (R2d ). Then H is trace-class. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Different Perspective Left action of K(L2 (Rd )) on S0 (Rd ): For F ∈ S0 (R2d ) and g ∈ S0 (Rd ) ZZ π(F ) · g = F (z)π(z)gdz Luef, Franz Hilbert C ∗ -modules over noncommutative tori Different Perspective Left action of K(L2 (Rd )) on S0 (Rd ): For F ∈ S0 (R2d ) and g ∈ S0 (Rd ) ZZ π(F ) · g = F (z)π(z)gdz Inner product with values in K(L2 (Rd )) Luef, Franz Hilbert C ∗ -modules over noncommutative tori Different Perspective Left action of K(L2 (Rd )) on S0 (Rd ): For F ∈ S0 (R2d ) and g ∈ S0 (Rd ) ZZ π(F ) · g = F (z)π(z)gdz Inner product with values in K(L2 (Rd )) ZZ hf , g i = Vg f (z)π(z) K for f , g ∈ S0 (Rd ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Different Perspective Left action of K(L2 (Rd )) on S0 (Rd ): For F ∈ S0 (R2d ) and g ∈ S0 (Rd ) ZZ π(F ) · g = F (z)π(z)gdz Inner product with values in K(L2 (Rd )) ZZ hf , g i = Vg f (z)π(z) K for f , g ∈ S0 (Rd ). K hπ(A) · f , g i = π(A)K hf , g i, i.e. the right hand side amounts to A]Vg f . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Different Perspective Left action of K(L2 (Rd )) on S0 (Rd ): For F ∈ S0 (R2d ) and g ∈ S0 (Rd ) ZZ π(F ) · g = F (z)π(z)gdz Inner product with values in K(L2 (Rd )) ZZ hf , g i = Vg f (z)π(z) K for f , g ∈ S0 (Rd ). K hπ(A) · f , g i = π(A)K hf , g i, i.e. the right hand side amounts to A]Vg f . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Hilbert K-module Theorem: S0 (Rd ) becomes a full left Hilbert K-module K V when completed with respect to the norm K kf k = kK hf , f ik1/2 for f ∈ S0 (Rd ) The proof relies on the properties of S0 (Rd ) and L1 (R2d , c). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Hilbert K-module Theorem: S0 (Rd ) becomes a full left Hilbert K-module K V when completed with respect to the norm K kf k = kK hf , f ik1/2 for f ∈ S0 (Rd ) The proof relies on the properties of S0 (Rd ) and L1 (R2d , c). Rank-one K V -module operators are just the rank-one operators on L2 (Rd ) Θg ,h f = hf , g i · h. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Hilbert K-module Theorem: S0 (Rd ) becomes a full left Hilbert K-module K V when completed with respect to the norm K kf k = kK hf , f ik1/2 for f ∈ S0 (Rd ) The proof relies on the properties of S0 (Rd ) and L1 (R2d , c). Rank-one K V -module operators are just the rank-one operators on L2 (Rd ) Θg ,h f = hf , g i · h. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Equivalence bimodule between K and C Since we need a right Hilbert C-module, we take the conjugate Hilbert space of L2 (Rd ) with inner product hf , g iC = hg , f iL2 (Rd ) . K hf , g i Luef, Franz · h = f · hg , hiC Hilbert C ∗ -modules over noncommutative tori Equivalence bimodule between K and C Since we need a right Hilbert C-module, we take the conjugate Hilbert space of L2 (Rd ) with inner product hf , g iC = hg , f iL2 (Rd ) . K hf , g i · h = f · hg , hiC Equivalent to hVg f , Vk hiL2 (R2d ) = hf , kiL2 (Rd ) hh, g iL2 (Rd ) Luef, Franz Hilbert C ∗ -modules over noncommutative tori Equivalence bimodule between K and C Since we need a right Hilbert C-module, we take the conjugate Hilbert space of L2 (Rd ) with inner product hf , g iC = hg , f iL2 (Rd ) . K hf , g i · h = f · hg , hiC Equivalent to hVg f , Vk hiL2 (R2d ) = hf , kiL2 (Rd ) hh, g iL2 (Rd ) Theorem: S0 (Rd ) is an equivalence bimodule between K and C. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Equivalence bimodule between K and C Since we need a right Hilbert C-module, we take the conjugate Hilbert space of L2 (Rd ) with inner product hf , g iC = hg , f iL2 (Rd ) . K hf , g i · h = f · hg , hiC Equivalent to hVg f , Vk hiL2 (R2d ) = hf , kiL2 (Rd ) hh, g iL2 (Rd ) Theorem: S0 (Rd ) is an equivalence bimodule between K and C. Morita-Rieffel equivalent C ∗ -algebras have equivalent categories of representations. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Equivalence bimodule between K and C Since we need a right Hilbert C-module, we take the conjugate Hilbert space of L2 (Rd ) with inner product hf , g iC = hg , f iL2 (Rd ) . K hf , g i · h = f · hg , hiC Equivalent to hVg f , Vk hiL2 (R2d ) = hf , kiL2 (Rd ) hh, g iL2 (Rd ) Theorem: S0 (Rd ) is an equivalence bimodule between K and C. Morita-Rieffel equivalent C ∗ -algebras have equivalent categories of representations. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Noncommutative torus Let Θ = (θj,k ) be a real skew-symmetric matrix and let U1 , ..., U2d be unitary operators. Then the noncommutative torus AΘ is defined as the universal C ∗ -algebra generated by Uj ’s satisfying Uk Uj = e 2πiθj,k Uj Uk . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Noncommutative torus Let Θ = (θj,k ) be a real skew-symmetric matrix and let U1 , ..., U2d be unitary operators. Then the noncommutative torus AΘ is defined as the universal C ∗ -algebra generated by Uj ’s satisfying Uk Uj = e 2πiθj,k Uj Uk . For the connection between noncommutative tori and Gabor analysis it is more useful to realize AΘ as the group C ∗ -algebra of a lattice Λ in R2d . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Noncommutative torus Let Θ = (θj,k ) be a real skew-symmetric matrix and let U1 , ..., U2d be unitary operators. Then the noncommutative torus AΘ is defined as the universal C ∗ -algebra generated by Uj ’s satisfying Uk Uj = e 2πiθj,k Uj Uk . For the connection between noncommutative tori and Gabor analysis it is more useful to realize AΘ as the group C ∗ -algebra of a lattice Λ in R2d . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Twisted group algebras Let Λ be a lattice in R2d and c a continuous 2-cocycle with values in T. Then the twisted group algebra `1 (Λ, c) is `1 (Λ) with twisted convolution \ as multiplication and ∗ as involution. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Twisted group algebras Let Λ be a lattice in R2d and c a continuous 2-cocycle with values in T. Then the twisted group algebra `1 (Λ, c) is `1 (Λ) with twisted convolution \ as multiplication and ∗ as involution. Twisted convolution of a and b is defined by X a\b(λ) = a(µ)b(λ − µ)c(µ, λ − µ) for λ, µ ∈ Λ, µ∈Λ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Twisted group algebras Let Λ be a lattice in R2d and c a continuous 2-cocycle with values in T. Then the twisted group algebra `1 (Λ, c) is `1 (Λ) with twisted convolution \ as multiplication and ∗ as involution. Twisted convolution of a and b is defined by X a\b(λ) = a(µ)b(λ − µ)c(µ, λ − µ) for λ, µ ∈ Λ, µ∈Λ Twisted involution of a given by a∗ (λ) = c(λ, λ)a(−λ) for λ ∈ Λ. C ∗ (Λ, c) is the enveloping C ∗ -algebra of `1 (Λ, c). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Twisted group algebras Let Λ be a lattice in R2d and c a continuous 2-cocycle with values in T. Then the twisted group algebra `1 (Λ, c) is `1 (Λ) with twisted convolution \ as multiplication and ∗ as involution. Twisted convolution of a and b is defined by X a\b(λ) = a(µ)b(λ − µ)c(µ, λ − µ) for λ, µ ∈ Λ, µ∈Λ Twisted involution of a given by a∗ (λ) = c(λ, λ)a(−λ) for λ ∈ Λ. C ∗ (Λ, c) is the enveloping C ∗ -algebra of `1 (Λ, c). Pick a basis {v1 , ..., v2d } for Λ. Then {Uvj } together with (θvj ,vk ) provides a description of the noncommutative torus. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Twisted group algebras Let Λ be a lattice in R2d and c a continuous 2-cocycle with values in T. Then the twisted group algebra `1 (Λ, c) is `1 (Λ) with twisted convolution \ as multiplication and ∗ as involution. Twisted convolution of a and b is defined by X a\b(λ) = a(µ)b(λ − µ)c(µ, λ − µ) for λ, µ ∈ Λ, µ∈Λ Twisted involution of a given by a∗ (λ) = c(λ, λ)a(−λ) for λ ∈ Λ. C ∗ (Λ, c) is the enveloping C ∗ -algebra of `1 (Λ, c). Pick a basis {v1 , ..., v2d } for Λ. Then {Uvj } together with (θvj ,vk ) provides a description of the noncommutative torus. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Representations of noncommutative torus The mapping of λ 7→ π(λ) is a projective representation of Λ, which gives a non-degenerate involutive representation of `1 (Λ, c) by X πΛ (a) := a(λ)π(λ) for a = (a(λ)) ∈ `1 (Λ). λ∈Λ πΛ (a\b) = πΛ (a)πΛ (b) Luef, Franz Hilbert C ∗ -modules over noncommutative tori Representations of noncommutative torus The mapping of λ 7→ π(λ) is a projective representation of Λ, which gives a non-degenerate involutive representation of `1 (Λ, c) by X πΛ (a) := a(λ)π(λ) for a = (a(λ)) ∈ `1 (Λ). λ∈Λ πΛ (a\b) = πΛ (a)πΛ (b) πΛ (a∗ ) = πΛ (a)∗ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Representations of noncommutative torus The mapping of λ 7→ π(λ) is a projective representation of Λ, which gives a non-degenerate involutive representation of `1 (Λ, c) by X πΛ (a) := a(λ)π(λ) for a = (a(λ)) ∈ `1 (Λ). λ∈Λ πΛ (a\b) = πΛ (a)πΛ (b) πΛ (a∗ ) = πΛ (a)∗ This involutive representation of `1 (Λ, c) is faithful, Luef, Franz Hilbert C ∗ -modules over noncommutative tori Representations of noncommutative torus The mapping of λ 7→ π(λ) is a projective representation of Λ, which gives a non-degenerate involutive representation of `1 (Λ, c) by X πΛ (a) := a(λ)π(λ) for a = (a(λ)) ∈ `1 (Λ). λ∈Λ πΛ (a\b) = πΛ (a)πΛ (b) πΛ (a∗ ) = πΛ (a)∗ This involutive representation of `1 (Λ, c) is faithful, πΛ (a) = 0 implies a = 0. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Representations of noncommutative torus The mapping of λ 7→ π(λ) is a projective representation of Λ, which gives a non-degenerate involutive representation of `1 (Λ, c) by X πΛ (a) := a(λ)π(λ) for a = (a(λ)) ∈ `1 (Λ). λ∈Λ πΛ (a\b) = πΛ (a)πΛ (b) πΛ (a∗ ) = πΛ (a)∗ This involutive representation of `1 (Λ, c) is faithful, πΛ (a) = 0 implies a = 0. Denote the image of the map a → 7 πΛ (a) for a ∈ `1s (Λ) by A1s (Λ, c). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Representations of noncommutative torus The mapping of λ 7→ π(λ) is a projective representation of Λ, which gives a non-degenerate involutive representation of `1 (Λ, c) by X πΛ (a) := a(λ)π(λ) for a = (a(λ)) ∈ `1 (Λ). λ∈Λ πΛ (a\b) = πΛ (a)πΛ (b) πΛ (a∗ ) = πΛ (a)∗ This involutive representation of `1 (Λ, c) is faithful, πΛ (a) = 0 implies a = 0. Denote the image of the map a → 7 πΛ (a) for a ∈ `1s (Λ) by A1s (Λ, c). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Spectrally invariant subalgebras of C ∗ (Λ, c) Recall that a unital subalgebra A of a unital C ∗ -algebra B with common unit is called spectrally invariant, if for A ∈ A with A−1 ∈ B one actually has that A−1 ∈ A. X A1s (Λ, c) = {A ∈ B(L2 (R)) : A = a(λ)π(λ), kak`1s < ∞} λ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Spectrally invariant subalgebras of C ∗ (Λ, c) Recall that a unital subalgebra A of a unital C ∗ -algebra B with common unit is called spectrally invariant, if for A ∈ A with A−1 ∈ B one actually has that A−1 ∈ A. X A1s (Λ, c) = {A ∈ B(L2 (R)) : A = a(λ)π(λ), kak`1s < ∞} λ smooth noncommutative torus Luef, Franz A∞ (Λ, c) = T 1 s≥0 As (Λ, c) Hilbert C ∗ -modules over noncommutative tori Spectrally invariant subalgebras of C ∗ (Λ, c) Recall that a unital subalgebra A of a unital C ∗ -algebra B with common unit is called spectrally invariant, if for A ∈ A with A−1 ∈ B one actually has that A−1 ∈ A. X A1s (Λ, c) = {A ∈ B(L2 (R)) : A = a(λ)π(λ), kak`1s < ∞} λ smooth noncommutative torus A∞ (Λ, c) = T 1 s≥0 As (Λ, c) Theorem: Let Λ be a lattice in R2d . Then A1s (Λ, c) and A∞ (Λ, c) are spectrally invariant subalgebras of the noncommutative torus C ∗ (Λ, c). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Spectrally invariant subalgebras of C ∗ (Λ, c) Recall that a unital subalgebra A of a unital C ∗ -algebra B with common unit is called spectrally invariant, if for A ∈ A with A−1 ∈ B one actually has that A−1 ∈ A. X A1s (Λ, c) = {A ∈ B(L2 (R)) : A = a(λ)π(λ), kak`1s < ∞} λ smooth noncommutative torus A∞ (Λ, c) = T 1 s≥0 As (Λ, c) Theorem: Let Λ be a lattice in R2d . Then A1s (Λ, c) and A∞ (Λ, c) are spectrally invariant subalgebras of the noncommutative torus C ∗ (Λ, c). The statement for A∞ (Λ, c) was obtained by Connes in 1980 and the one for A1s (Λ, c) is due to Gröchenig-Leinert and independently by Rosenberg. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Spectrally invariant subalgebras of C ∗ (Λ, c) Recall that a unital subalgebra A of a unital C ∗ -algebra B with common unit is called spectrally invariant, if for A ∈ A with A−1 ∈ B one actually has that A−1 ∈ A. X A1s (Λ, c) = {A ∈ B(L2 (R)) : A = a(λ)π(λ), kak`1s < ∞} λ smooth noncommutative torus A∞ (Λ, c) = T 1 s≥0 As (Λ, c) Theorem: Let Λ be a lattice in R2d . Then A1s (Λ, c) and A∞ (Λ, c) are spectrally invariant subalgebras of the noncommutative torus C ∗ (Λ, c). The statement for A∞ (Λ, c) was obtained by Connes in 1980 and the one for A1s (Λ, c) is due to Gröchenig-Leinert and independently by Rosenberg. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Equivalence bimodule between noncommutative tori Quote from Rieffel’s seminal paper from 1988: We need suitable spaces of functions for the construction of equivalence bimodules between noncommutative tori. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Equivalence bimodule between noncommutative tori Quote from Rieffel’s seminal paper from 1988: We need suitable spaces of functions for the construction of equivalence bimodules between noncommutative tori. In the present context this means we need a space of functions on Rd which behaves well under both the Fourier transform and restriction to subgroups. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Equivalence bimodule between noncommutative tori Quote from Rieffel’s seminal paper from 1988: We need suitable spaces of functions for the construction of equivalence bimodules between noncommutative tori. In the present context this means we need a space of functions on Rd which behaves well under both the Fourier transform and restriction to subgroups. As suggested by Weil, the appropriate space is the space S(Rd ) of Schwartz functions. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Equivalence bimodule between noncommutative tori Quote from Rieffel’s seminal paper from 1988: We need suitable spaces of functions for the construction of equivalence bimodules between noncommutative tori. In the present context this means we need a space of functions on Rd which behaves well under both the Fourier transform and restriction to subgroups. As suggested by Weil, the appropriate space is the space S(Rd ) of Schwartz functions. Feichtinger constructed a Banach algebra in 1980, S0 (Rd ), that turned out to be a very important substitute of the Schwartz space. Shortly, after he introduced the class of modulation spaces Msp,q (Rd ) and it turned out that S0 (Rd ) is the modulation space M 1 (Rd ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Equivalence bimodule between noncommutative tori Quote from Rieffel’s seminal paper from 1988: We need suitable spaces of functions for the construction of equivalence bimodules between noncommutative tori. In the present context this means we need a space of functions on Rd which behaves well under both the Fourier transform and restriction to subgroups. As suggested by Weil, the appropriate space is the space S(Rd ) of Schwartz functions. Feichtinger constructed a Banach algebra in 1980, S0 (Rd ), that turned out to be a very important substitute of the Schwartz space. Shortly, after he introduced the class of modulation spaces Msp,q (Rd ) and it turned out that S0 (Rd ) is the modulation space M 1 (Rd ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Symplectic Fourier transform b d are of the form Observe that the characters of Rd × R 0) 2πiΩ(z,z 0 bd , z 7→ χs (z) = e for some z ∈ Rd × R where the standard symplectic form Ω of z = (x, ω) and z 0 = (y , η) is defined by Luef, Franz Hilbert C ∗ -modules over noncommutative tori Symplectic Fourier transform b d are of the form Observe that the characters of Rd × R 0) 2πiΩ(z,z 0 bd , z 7→ χs (z) = e for some z ∈ Rd × R where the standard symplectic form Ω of z = (x, ω) and z 0 = (y , η) is defined by Ω(z, z 0 ) = y · ω − x · η. b d is R b d × Rd . Consequently, the dual group of Rd × R Luef, Franz Hilbert C ∗ -modules over noncommutative tori Symplectic Fourier transform b d are of the form Observe that the characters of Rd × R 0) 2πiΩ(z,z 0 bd , z 7→ χs (z) = e for some z ∈ Rd × R where the standard symplectic form Ω of z = (x, ω) and z 0 = (y , η) is defined by Ω(z, z 0 ) = y · ω − x · η. b d is R b d × Rd . Consequently, the dual group of Rd × R In terms of the Euclidean inner product h., .i on R2d the symplectic form Ω can be expressed as follows: pause 0 Id Ω(z, z 0 ) = hJz, z 0 i for J = . −Id 0 Luef, Franz Hilbert C ∗ -modules over noncommutative tori Symplectic Fourier transform b d are of the form Observe that the characters of Rd × R 0) 2πiΩ(z,z 0 bd , z 7→ χs (z) = e for some z ∈ Rd × R where the standard symplectic form Ω of z = (x, ω) and z 0 = (y , η) is defined by Ω(z, z 0 ) = y · ω − x · η. b d is R b d × Rd . Consequently, the dual group of Rd × R In terms of the Euclidean inner product h., .i on R2d the symplectic form Ω can be expressed as follows: pause 0 Id Ω(z, z 0 ) = hJz, z 0 i for J = . −Id 0 Λ◦ = {z ∈ R2d : e 2πiΩ(λ,z) = 1 for all λ ∈ Λ}. The relation between Λ⊥ and the adjoint lattice Λ◦ is Λ◦ = JΛ⊥ . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Symplectic Fourier transform b d are of the form Observe that the characters of Rd × R 0) 2πiΩ(z,z 0 bd , z 7→ χs (z) = e for some z ∈ Rd × R where the standard symplectic form Ω of z = (x, ω) and z 0 = (y , η) is defined by Ω(z, z 0 ) = y · ω − x · η. b d is R b d × Rd . Consequently, the dual group of Rd × R In terms of the Euclidean inner product h., .i on R2d the symplectic form Ω can be expressed as follows: pause 0 Id Ω(z, z 0 ) = hJz, z 0 i for J = . −Id 0 Λ◦ = {z ∈ R2d : e 2πiΩ(λ,z) = 1 for all λ ∈ Λ}. The relation between Λ⊥ and the adjoint lattice Λ◦ is Λ◦ = JΛ⊥ . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Feichtinger’s algebra - Properties Definition: Weighted versions of Feichtinger’s algebra M 1 (Rd ) = {f ∈ L2 (Rd ) : kf kMs1 = RRs 2 2 s/2 dxdω < ∞} R2d |hπ(x, ω)g i|(1 + |x| + |ω| ) If f , g ∈ Ms1 (Rd ), then Vg f ∈ Ms1 (R2d ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Feichtinger’s algebra - Properties Definition: Weighted versions of Feichtinger’s algebra M 1 (Rd ) = {f ∈ L2 (Rd ) : kf kMs1 = RRs 2 2 s/2 dxdω < ∞} R2d |hπ(x, ω)g i|(1 + |x| + |ω| ) If f , g ∈ Ms1 (Rd ), then Vg f ∈ Ms1 (R2d ). M 1 (R2d ) is invariant symplectic Fourier RRunder the2πiΩ(b z ,z) dz. b transform: F (b z ) = R2d F (z)e Luef, Franz Hilbert C ∗ -modules over noncommutative tori Feichtinger’s algebra - Properties Definition: Weighted versions of Feichtinger’s algebra M 1 (Rd ) = {f ∈ L2 (Rd ) : kf kMs1 = RRs 2 2 s/2 dxdω < ∞} R2d |hπ(x, ω)g i|(1 + |x| + |ω| ) If f , g ∈ Ms1 (Rd ), then Vg f ∈ Ms1 (R2d ). M 1 (R2d ) is invariant symplectic Fourier RRunder the2πiΩ(b z ,z) dz. b transform: F (b z ) = R2d F (z)e (Vg1\ f1 · Vg2 f2 )(z) = (Vf2 f1 · Vg2 g1 )(z) Luef, Franz Hilbert C ∗ -modules over noncommutative tori Feichtinger’s algebra - Properties Definition: Weighted versions of Feichtinger’s algebra M 1 (Rd ) = {f ∈ L2 (Rd ) : kf kMs1 = RRs 2 2 s/2 dxdω < ∞} R2d |hπ(x, ω)g i|(1 + |x| + |ω| ) If f , g ∈ Ms1 (Rd ), then Vg f ∈ Ms1 (R2d ). M 1 (R2d ) is invariant symplectic Fourier RRunder the2πiΩ(b z ,z) dz. b transform: F (b z ) = R2d F (z)e (Vg1\ f1 · Vg2 f2 )(z) = (Vf2 f1 · Vg2 g1 )(z) Luef, Franz Hilbert C ∗ -modules over noncommutative tori Janssen representation Poisson summation formula for the symplectic Fourier transform: X X F (λ) = vol(Λ)−1 Fb (λ◦ ). λ∈Λ λ◦ ∈Λ◦ Fundamental Identity of Gabor analysis: Luef, Franz Hilbert C ∗ -modules over noncommutative tori Janssen representation Poisson summation formula for the symplectic Fourier transform: X X F (λ) = vol(Λ)−1 Fb (λ◦ ). λ∈Λ λ◦ ∈Λ◦ Fundamental Identity of Gabor analysis: X X hf , π(λ◦ )kihπ(λ◦ )h, g i. hf , π(λ)g ihπ(λ)h, ki = vol(Λ)−1 λ◦ ∈Λ◦ λ∈Λ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Janssen representation Poisson summation formula for the symplectic Fourier transform: X X F (λ) = vol(Λ)−1 Fb (λ◦ ). λ∈Λ λ◦ ∈Λ◦ Fundamental Identity of Gabor analysis: X X hf , π(λ◦ )kihπ(λ◦ )h, g i. hf , π(λ)g ihπ(λ)h, ki = vol(Λ)−1 λ◦ ∈Λ◦ λ∈Λ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Janssen representation Theorem: For f , g , h ∈ Ms1 (Rd ) or in S(Rd ) we have that X X hf , π(λ)g ihπ(λ)h = vol(Λ)−1 hh, π(λ◦ )g iπ(λ◦ )f , λ◦ ∈Λ◦ λ∈Λ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Rieffel’s Theorem A deep result of Rieffel relates these different noncommutative tori. Theorem: C ∗ (Λ, c) and C ∗ (Λ◦ , c) are Morita-Rieffel equivalent. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Rieffel’s Theorem A deep result of Rieffel relates these different noncommutative tori. Theorem: C ∗ (Λ, c) and C ∗ (Λ◦ , c) are Morita-Rieffel equivalent. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Gabor frames – traditional point of view Let G(g , Λ) = {π(λ)g : λ ∈ Λ} be a Gabor system. analysis operator: Cg f = (hf , π(λ)g i)λ∈Λ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Gabor frames – traditional point of view Let G(g , Λ) = {π(λ)g : λ ∈ Λ} be a Gabor system. analysis operator: Cg f = (hf , π(λ)g i)λ∈Λ P synthesis operator: Dg a = λ∈Λ a(λ)π(λ)f Luef, Franz Hilbert C ∗ -modules over noncommutative tori Gabor frames – traditional point of view Let G(g , Λ) = {π(λ)g : λ ∈ Λ} be a Gabor system. analysis operator: Cg f = (hf , π(λ)g i)λ∈Λ P synthesis operator: Dg a = λ∈Λ a(λ)π(λ)f P frame operator: Sg ,Λ f = λ∈Λ hf , π(λ)g iπ(λ)g Luef, Franz Hilbert C ∗ -modules over noncommutative tori Gabor frames – traditional point of view Let G(g , Λ) = {π(λ)g : λ ∈ Λ} be a Gabor system. analysis operator: Cg f = (hf , π(λ)g i)λ∈Λ P synthesis operator: Dg a = λ∈Λ a(λ)π(λ)f P frame operator: Sg ,Λ f = λ∈Λ hf , π(λ)g iπ(λ)g G(g , Λ) is a frame for L2 (Rd ) if Sg ,Λ is invertible on L2 (Rd ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Gabor frames – traditional point of view Let G(g , Λ) = {π(λ)g : λ ∈ Λ} be a Gabor system. analysis operator: Cg f = (hf , π(λ)g i)λ∈Λ P synthesis operator: Dg a = λ∈Λ a(λ)π(λ)f P frame operator: Sg ,Λ f = λ∈Λ hf , π(λ)g iπ(λ)g G(g , Λ) is a frame for L2 (Rd ) if Sg ,Λ is invertible on L2 (Rd ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Gabor frames – traditional point of view Let G(g , Λ) = {π(λ)g : λ ∈ Λ} be a Gabor system. analysis operator: Cg f = (hf , π(λ)g i)λ∈Λ P synthesis operator: Dg a = λ∈Λ a(λ)π(λ)f P frame operator: Sg ,Λ f = λ∈Λ hf , π(λ)g iπ(λ)g G(g , Λ) is a frame for L2 (Rd ) if Sg ,Λ is invertible on L2 (Rd ). X Akf k22 ≤ |hf , π(λ)g i|2 ≤ Bkf k22 λ∈Λ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Gabor frames – traditional point of view Let G(g , Λ) = {π(λ)g : λ ∈ Λ} be a Gabor system. analysis operator: Cg f = (hf , π(λ)g i)λ∈Λ P synthesis operator: Dg a = λ∈Λ a(λ)π(λ)f P frame operator: Sg ,Λ f = λ∈Λ hf , π(λ)g iπ(λ)g G(g , Λ) is a frame for L2 (Rd ) if Sg ,Λ is invertible on L2 (Rd ). X Akf k22 ≤ |hf , π(λ)g i|2 ≤ Bkf k22 λ∈Λ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Discrete reconstruction f = S −1 Sf = X hf , π(λ)S −1 g iπ(λ)g λ∈Λ = SS −1 f = X hf , π(λ)g iπ(λ)S −1 g λ∈Λ = S −1/2 SS −1/2 f = X hf , π(λ)S −1/2 g iπ(λ)S −1/2 g . λ∈Λ canonical dual atom g̃ := (SgΛ,g )−1 g canonical tight atom h0 := (SgΛ,g )−1/2 g Luef, Franz Hilbert C ∗ -modules over noncommutative tori Observation Left action of A1s (Λ, c) on Ms1 (Rd ) by X Da g = πΛ (a)·g = a(λ)π(λ) g for a ∈ `1s (Λ), g ∈ Ms1 (Rd ) λ∈Λ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Observation Left action of A1s (Λ, c) on Ms1 (Rd ) by X Da g = πΛ (a)·g = a(λ)π(λ) g for a ∈ `1s (Λ), g ∈ Ms1 (Rd ) λ∈Λ Λ hf , g i = X hf , π(λ)g iπ(λ) λ∈Λ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Observation Left action of A1s (Λ, c) on Ms1 (Rd ) by X Da g = πΛ (a)·g = a(λ)π(λ) g for a ∈ `1s (Λ), g ∈ Ms1 (Rd ) λ∈Λ Λ hf , g i = X hf , π(λ)g iπ(λ) λ∈Λ For f , g ∈ Ms1 (Rd ) define P Λ hf , g i = πΛ (Vg f ) = λ∈Λ hf , π(λ)g iπ(λ) Luef, Franz Hilbert C ∗ -modules over noncommutative tori Observation Left action of A1s (Λ, c) on Ms1 (Rd ) by X Da g = πΛ (a)·g = a(λ)π(λ) g for a ∈ `1s (Λ), g ∈ Ms1 (Rd ) λ∈Λ Λ hf , g i = X hf , π(λ)g iπ(λ) λ∈Λ For f , g ∈ Ms1 (Rd ) define P Λ hf , g i = πΛ (Vg f ) = λ∈Λ hf , π(λ)g iπ(λ) Λ hf , g i = Λ hg , f i∗ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Observation Left action of A1s (Λ, c) on Ms1 (Rd ) by X Da g = πΛ (a)·g = a(λ)π(λ) g for a ∈ `1s (Λ), g ∈ Ms1 (Rd ) λ∈Λ Λ hf , g i = X hf , π(λ)g iπ(λ) λ∈Λ For f , g ∈ Ms1 (Rd ) define P Λ hf , g i = πΛ (Vg f ) = λ∈Λ hf , π(λ)g iπ(λ) Λ hf , g i = Λ hg , f i∗ Λ hπΛ (a)f , g i = πΛ (a) Λ hf , g i Luef, Franz Hilbert C ∗ -modules over noncommutative tori Observation Left action of A1s (Λ, c) on Ms1 (Rd ) by X Da g = πΛ (a)·g = a(λ)π(λ) g for a ∈ `1s (Λ), g ∈ Ms1 (Rd ) λ∈Λ Λ hf , g i = X hf , π(λ)g iπ(λ) λ∈Λ For f , g ∈ Ms1 (Rd ) define P Λ hf , g i = πΛ (Vg f ) = λ∈Λ hf , π(λ)g iπ(λ) Λ hf , g i = Λ hg , f i∗ Λ hπΛ (a)f , g i Λ hf , f i = πΛ (a) Λ hf , g i is positive in C ∗ (Λ, c), acutally in A1s (Λ, c) Luef, Franz Hilbert C ∗ -modules over noncommutative tori Observation Left action of A1s (Λ, c) on Ms1 (Rd ) by X Da g = πΛ (a)·g = a(λ)π(λ) g for a ∈ `1s (Λ), g ∈ Ms1 (Rd ) λ∈Λ Λ hf , g i = X hf , π(λ)g iπ(λ) λ∈Λ For f , g ∈ Ms1 (Rd ) define P Λ hf , g i = πΛ (Vg f ) = λ∈Λ hf , π(λ)g iπ(λ) Λ hf , g i = Λ hg , f i∗ Λ hπΛ (a)f , g i Λ hf , f i = πΛ (a) Λ hf , g i is positive in C ∗ (Λ, c), acutally in A1s (Λ, c) Luef, Franz Hilbert C ∗ -modules over noncommutative tori Observation Left action of A1s (Λ, c) on Ms1 (Rd ) by X Da g = πΛ (a)·g = a(λ)π(λ) g for a ∈ `1s (Λ), g ∈ Ms1 (Rd ) λ∈Λ Λ hf , g i = X hf , π(λ)g iπ(λ) λ∈Λ For f , g ∈ Ms1 (Rd ) define P Λ hf , g i = πΛ (Vg f ) = λ∈Λ hf , π(λ)g iπ(λ) Λ hf , g i = Λ hg , f i∗ Λ hπΛ (a)f , g i Λ hf , f i = πΛ (a) Λ hf , g i is positive in C ∗ (Λ, c), acutally in A1s (Λ, c) Luef, Franz Hilbert C ∗ -modules over noncommutative tori Hilbert C ∗ (Λ, c)-module Theorem: Ms1 (Rd ) becomes a full left Hilbert C ∗ (Λ, c)-module Λ V w.r.t to right action on Ms1 (Rd ) and the inner product h., .iΛ when 1/2 completed w.r.t. kf kΛ = khf , f iΛ kop . There is an analogous result for the opposite C ∗ -algebra of C ∗ (Λ◦ , c), i.e. C ∗ (Λ◦ , c). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Hilbert C ∗ (Λ, c)-module Theorem: Ms1 (Rd ) becomes a full left Hilbert C ∗ (Λ, c)-module Λ V w.r.t to right action on Ms1 (Rd ) and the inner product h., .iΛ when 1/2 completed w.r.t. kf kΛ = khf , f iΛ kop . There is an analogous result for the opposite C ∗ -algebra of C ∗ (Λ◦ , c), i.e. C ∗ (Λ◦ , c). What is the appropriate right action of C ∗ (Λ◦ , c) on Ms1 (Rd )? Luef, Franz Hilbert C ∗ -modules over noncommutative tori Hilbert C ∗ (Λ, c)-module Theorem: Ms1 (Rd ) becomes a full left Hilbert C ∗ (Λ, c)-module Λ V w.r.t to right action on Ms1 (Rd ) and the inner product h., .iΛ when 1/2 completed w.r.t. kf kΛ = khf , f iΛ kop . There is an analogous result for the opposite C ∗ -algebra of C ∗ (Λ◦ , c), i.e. C ∗ (Λ◦ , c). What is the appropriate right action of C ∗ (Λ◦ , c) on Ms1 (Rd )? Luef, Franz Hilbert C ∗ -modules over noncommutative tori Hilbert C ∗ (Λ◦ , c)-module FIGA = vol(Λ)−1 X hπ(λ◦ )∗ f , kihπ(λ◦ )h, g i λ◦ ∈Λ◦ D E X = vol(Λ)−1 hπ(λ◦ )h, g iπ(λ◦ )∗ f , k λ◦ ∈Λ◦ g · πΛ◦ (b) = X π(λ◦ )∗ g b(λ◦ ) b ∈ `1s (Λ◦ ), g ∈ Ms1 (R) λ◦ ∈Λ◦ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Hilbert C ∗ (Λ◦ , c)-module FIGA = vol(Λ)−1 X hπ(λ◦ )∗ f , kihπ(λ◦ )h, g i λ◦ ∈Λ◦ D E X = vol(Λ)−1 hπ(λ◦ )h, g iπ(λ◦ )∗ f , k λ◦ ∈Λ◦ g · πΛ◦ (b) = X π(λ◦ )∗ g b(λ◦ ) b ∈ `1s (Λ◦ ), g ∈ Ms1 (R) λ◦ ∈Λ◦ hf , g iΛ◦ = X π(λ◦ )∗ hg , π(λ◦ )f i f , g ∈ Ms1 (Rd ) Luef, Franz Hilbert C ∗ -modules over noncommutative tori Hilbert C ∗ (Λ◦ , c)-module FIGA = vol(Λ)−1 X hπ(λ◦ )∗ f , kihπ(λ◦ )h, g i λ◦ ∈Λ◦ D E X = vol(Λ)−1 hπ(λ◦ )h, g iπ(λ◦ )∗ f , k λ◦ ∈Λ◦ g · πΛ◦ (b) = X π(λ◦ )∗ g b(λ◦ ) b ∈ `1s (Λ◦ ), g ∈ Ms1 (R) λ◦ ∈Λ◦ hf , g iΛ◦ = X π(λ◦ )∗ hg , π(λ◦ )f i f , g ∈ Ms1 (Rd ) Theorem: Ms1 (Rd ) becomes a full right Hilbert C ∗ (Λ◦ , c)-module VΛ◦ w.r.t to right action on Ms1 (Rd ) and the inner product h., .iΛ◦ 1/2 when completed w.r.t. kf kΛ◦ = khf , f iΛ◦ kop . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Hilbert C ∗ (Λ◦ , c)-module FIGA = vol(Λ)−1 X hπ(λ◦ )∗ f , kihπ(λ◦ )h, g i λ◦ ∈Λ◦ D E X = vol(Λ)−1 hπ(λ◦ )h, g iπ(λ◦ )∗ f , k λ◦ ∈Λ◦ g · πΛ◦ (b) = X π(λ◦ )∗ g b(λ◦ ) b ∈ `1s (Λ◦ ), g ∈ Ms1 (R) λ◦ ∈Λ◦ hf , g iΛ◦ = X π(λ◦ )∗ hg , π(λ◦ )f i f , g ∈ Ms1 (Rd ) Theorem: Ms1 (Rd ) becomes a full right Hilbert C ∗ (Λ◦ , c)-module VΛ◦ w.r.t to right action on Ms1 (Rd ) and the inner product h., .iΛ◦ 1/2 when completed w.r.t. kf kΛ◦ = khf , f iΛ◦ kop . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Adjointable C ∗ (Λ, c)-module operators By definition T is adjointable if there exists a T ? such that ? Λ hf , Tg i = Λ hf , T g i X hTf , π(λ)g iπ(λ) = λ∈Λ X hf , π(λ)T ? g iπ(λ) λ∈Λ T ? exists only if T π(λ) = π(λ)T for all λ ∈ Λ, such an operator T is Λ-invariant. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Adjointable C ∗ (Λ, c)-module operators By definition T is adjointable if there exists a T ? such that ? Λ hf , Tg i = Λ hf , T g i X hTf , π(λ)g iπ(λ) = λ∈Λ X hf , π(λ)T ? g iπ(λ) λ∈Λ T ? exists only if T π(λ) = π(λ)T for all λ ∈ Λ, such an operator T is Λ-invariant. Therefore, T ? is the adjoint T ∗ of a Λ-invariant operator. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Adjointable C ∗ (Λ, c)-module operators By definition T is adjointable if there exists a T ? such that ? Λ hf , Tg i = Λ hf , T g i X hTf , π(λ)g iπ(λ) = λ∈Λ X hf , π(λ)T ? g iπ(λ) λ∈Λ T ? exists only if T π(λ) = π(λ)T for all λ ∈ Λ, such an operator T is Λ-invariant. Therefore, T ? is the adjoint T ∗ of a Λ-invariant operator. Most important example, ΘΛg ,h f = Λ hf , g i · h and its adjoint is ΘΛh,g . ΘΛg ,h are known as Gabor frame-type operator Luef, Franz Hilbert C ∗ -modules over noncommutative tori Adjointable C ∗ (Λ, c)-module operators By definition T is adjointable if there exists a T ? such that ? Λ hf , Tg i = Λ hf , T g i X hTf , π(λ)g iπ(λ) = λ∈Λ X hf , π(λ)T ? g iπ(λ) λ∈Λ T ? exists only if T π(λ) = π(λ)T for all λ ∈ Λ, such an operator T is Λ-invariant. Therefore, T ? is the adjoint T ∗ of a Λ-invariant operator. Most important example, ΘΛg ,h f = Λ hf , g i · h and its adjoint is ΘΛh,g . ΘΛg ,h are known as Gabor frame-type operator Luef, Franz Hilbert C ∗ -modules over noncommutative tori Invertible adjointable C ∗ (Λ, c)-module operators If ΘΛg ,g is invertible, then G (g , Λ) is a Gabor frame for L2 (Rd ). Equivalently, Λ V is singly generated. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Invertible adjointable C ∗ (Λ, c)-module operators If ΘΛg ,g is invertible, then G (g , Λ) is a Gabor frame for L2 (Rd ). Equivalently, Λ V is singly generated. There are various criteria about the invertibility of the Gabor frame operator. For g ∈ M 1 (Rd ) the Gabor frame operator ΘΛg ,g is invertible if and only if the analysis operator CgΛ is one-to-one from M 1 (Rd )0 to `∞ (Λ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Invertible adjointable C ∗ (Λ, c)-module operators If ΘΛg ,g is invertible, then G (g , Λ) is a Gabor frame for L2 (Rd ). Equivalently, Λ V is singly generated. There are various criteria about the invertibility of the Gabor frame operator. For g ∈ M 1 (Rd ) the Gabor frame operator ΘΛg ,g is invertible if and only if the analysis operator CgΛ is one-to-one from M 1 (Rd )0 to `∞ (Λ). For g ∈ M 1 (Rd ) the Gabor frame operator ΘΛg ,g is invertible if and only if the synthesis operator DgΛ is one-to-one from `∞ (Λ) to M 1 (Rd )0 . For g ∈ M 1 (RdP ) the Gabor frame operator has as Weyl symbol σg ,Λ = λ∈Λ Tλ W (g , g ), where W (g , g ) is the Wigner distribution. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Invertible adjointable C ∗ (Λ, c)-module operators If ΘΛg ,g is invertible, then G (g , Λ) is a Gabor frame for L2 (Rd ). Equivalently, Λ V is singly generated. There are various criteria about the invertibility of the Gabor frame operator. For g ∈ M 1 (Rd ) the Gabor frame operator ΘΛg ,g is invertible if and only if the analysis operator CgΛ is one-to-one from M 1 (Rd )0 to `∞ (Λ). For g ∈ M 1 (Rd ) the Gabor frame operator ΘΛg ,g is invertible if and only if the synthesis operator DgΛ is one-to-one from `∞ (Λ) to M 1 (Rd )0 . For g ∈ M 1 (RdP ) the Gabor frame operator has as Weyl symbol σg ,Λ = λ∈Λ Tλ W (g , g ), where W (g , g ) is the Wigner distribution. Then ΘΛg ,g is invertible if and only if σg ,Λ is invertible in M ∞,1 (R2d ) with respect to twisted convolution. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Invertible adjointable C ∗ (Λ, c)-module operators If ΘΛg ,g is invertible, then G (g , Λ) is a Gabor frame for L2 (Rd ). Equivalently, Λ V is singly generated. There are various criteria about the invertibility of the Gabor frame operator. For g ∈ M 1 (Rd ) the Gabor frame operator ΘΛg ,g is invertible if and only if the analysis operator CgΛ is one-to-one from M 1 (Rd )0 to `∞ (Λ). For g ∈ M 1 (Rd ) the Gabor frame operator ΘΛg ,g is invertible if and only if the synthesis operator DgΛ is one-to-one from `∞ (Λ) to M 1 (Rd )0 . For g ∈ M 1 (RdP ) the Gabor frame operator has as Weyl symbol σg ,Λ = λ∈Λ Tλ W (g , g ), where W (g , g ) is the Wigner distribution. Then ΘΛg ,g is invertible if and only if σg ,Λ is invertible in M ∞,1 (R2d ) with respect to twisted convolution. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita equivalent C ∗ -algebras in Gabor analysis Therefore V0 is an equivalence bimodule between A0 and B0 . If we are in the situation of the preceding theorem, then we call the algebras A0 and B0 Morita-Rieffel equivalent. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita equivalent C ∗ -algebras in Gabor analysis Therefore V0 is an equivalence bimodule between A0 and B0 . If we are in the situation of the preceding theorem, then we call the algebras A0 and B0 Morita-Rieffel equivalent. The inner products Λ h., .i and h., .iΛ◦ satisfy Rieffel’s associativity condition Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita equivalent C ∗ -algebras in Gabor analysis Therefore V0 is an equivalence bimodule between A0 and B0 . If we are in the situation of the preceding theorem, then we call the algebras A0 and B0 Morita-Rieffel equivalent. The inner products Λ h., .i and h., .iΛ◦ satisfy Rieffel’s associativity condition Λ hf , g i Luef, Franz · h = f hg , hiΛ◦ Hilbert C ∗ -modules over noncommutative tori Morita equivalent C ∗ -algebras in Gabor analysis Therefore V0 is an equivalence bimodule between A0 and B0 . If we are in the situation of the preceding theorem, then we call the algebras A0 and B0 Morita-Rieffel equivalent. The inner products Λ h., .i and h., .iΛ◦ satisfy Rieffel’s associativity condition Λ hf , g i · h = f hg , hiΛ◦ rank-one operators on the Hilbert C ∗ (Λ, c)-module: ΘΛf ,g h = Λ hf , g i · h = SgΛ,h f are Gabor frame-type operators and the associativity condition is the Janssen representation for these Gabor frame-type operators. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita equivalent C ∗ -algebras in Gabor analysis Therefore V0 is an equivalence bimodule between A0 and B0 . If we are in the situation of the preceding theorem, then we call the algebras A0 and B0 Morita-Rieffel equivalent. The inner products Λ h., .i and h., .iΛ◦ satisfy Rieffel’s associativity condition Λ hf , g i · h = f hg , hiΛ◦ rank-one operators on the Hilbert C ∗ (Λ, c)-module: ΘΛf ,g h = Λ hf , g i · h = SgΛ,h f are Gabor frame-type operators and the associativity condition is the Janssen representation for these Gabor frame-type operators. Theorem: Ms1 (Rd ) is an equivalence bimodule between C ∗ (Λ, c) and C ∗ (Λ◦ , c). Moreover Ms1 (Rd ) is an equivalence bimodule between A1s (Λ, c) and A1s (Λ◦ , c). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita equivalent C ∗ -algebras in Gabor analysis Therefore V0 is an equivalence bimodule between A0 and B0 . If we are in the situation of the preceding theorem, then we call the algebras A0 and B0 Morita-Rieffel equivalent. The inner products Λ h., .i and h., .iΛ◦ satisfy Rieffel’s associativity condition Λ hf , g i · h = f hg , hiΛ◦ rank-one operators on the Hilbert C ∗ (Λ, c)-module: ΘΛf ,g h = Λ hf , g i · h = SgΛ,h f are Gabor frame-type operators and the associativity condition is the Janssen representation for these Gabor frame-type operators. Theorem: Ms1 (Rd ) is an equivalence bimodule between C ∗ (Λ, c) and C ∗ (Λ◦ , c). Moreover Ms1 (Rd ) is an equivalence bimodule between A1s (Λ, c) and A1s (Λ◦ , c). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projective modules over noncommutative tori Theorem: Ms1 (Rd ) is a finitely generated projective right A1 (Λ◦ , c),i.e. there exist g1 , ..., gn in Ms1 (Rd ) f = n X Λ hf , gi i i=1 Luef, Franz · gi = n X f hgi , gi iΛ◦ . i=1 Hilbert C ∗ -modules over noncommutative tori Projective modules over noncommutative tori Theorem: Ms1 (Rd ) is a finitely generated projective right A1 (Λ◦ , c),i.e. there exist g1 , ..., gn in Ms1 (Rd ) f = n X Λ hf , gi i · gi = i=1 Λ hf , f i n X f hgi , gi iΛ◦ . i=1 = n X Λ hgi , f iΛ hf , gi i i=1 for all f in Ms1 (Rd ) (or S(Rd )). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projective modules over noncommutative tori Theorem: Ms1 (Rd ) is a finitely generated projective right A1 (Λ◦ , c),i.e. there exist g1 , ..., gn in Ms1 (Rd ) f = n X Λ hf , gi i · gi = i=1 Λ hf , f i n X f hgi , gi iΛ◦ . i=1 = n X Λ hgi , f iΛ hf , gi i i=1 for all f in Ms1 (Rd ) (or S(Rd )).In particular kf k22 = n X X |hf , π(λ)gi i|2 . i=1 λ∈Λ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projective modules over noncommutative tori Theorem: Ms1 (Rd ) is a finitely generated projective right A1 (Λ◦ , c),i.e. there exist g1 , ..., gn in Ms1 (Rd ) f = n X Λ hf , gi i · gi = i=1 Λ hf , f i n X f hgi , gi iΛ◦ . i=1 = n X Λ hgi , f iΛ hf , gi i i=1 for all f in Ms1 (Rd ) (or S(Rd )).In particular kf k22 = n X X |hf , π(λ)gi i|2 . i=1 λ∈Λ Luef, Franz Hilbert C ∗ -modules over noncommutative tori Generalizations b , where G is a All our results hold for Λ a lattice in G × G locally compact abelian group. There exist weighted variants of our statements: Luef, Franz Hilbert C ∗ -modules over noncommutative tori Generalizations b , where G is a All our results hold for Λ a lattice in G × G locally compact abelian group. There exist weighted variants of our statements: v is submultiplicative, i.e. v (x + y , ω + η) ≤ v (x, ω)v (y , η) for all (x, ω), (y , η) ∈ R2d . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Generalizations b , where G is a All our results hold for Λ a lattice in G × G locally compact abelian group. There exist weighted variants of our statements: v is submultiplicative, i.e. v (x + y , ω + η) ≤ v (x, ω)v (y , η) for all (x, ω), (y , η) ∈ R2d . v (x, ω) ≥ 1 and v (−x, −ω) = v (x, ω) for all (x, ω) ∈ R2d . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Generalizations b , where G is a All our results hold for Λ a lattice in G × G locally compact abelian group. There exist weighted variants of our statements: v is submultiplicative, i.e. v (x + y , ω + η) ≤ v (x, ω)v (y , η) for all (x, ω), (y , η) ∈ R2d . v (x, ω) ≥ 1 and v (−x, −ω) = v (x, ω) for all (x, ω) ∈ R2d . (`1v (Λ), c) is a Banach algebra with continuous involution. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Generalizations b , where G is a All our results hold for Λ a lattice in G × G locally compact abelian group. There exist weighted variants of our statements: v is submultiplicative, i.e. v (x + y , ω + η) ≤ v (x, ω)v (y , η) for all (x, ω), (y , η) ∈ R2d . v (x, ω) ≥ 1 and v (−x, −ω) = v (x, ω) for all (x, ω) ∈ R2d . (`1v (Λ), c) is a Banach algebra with continuous involution. Gröchenig: A1v (Λ, c) is spectrally invariant in C ∗ (Λ, c) if and only if v is a GRS-weight, i.e. lim v (nλ)1/n = 1 for all λ ∈ Λ. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Generalizations b , where G is a All our results hold for Λ a lattice in G × G locally compact abelian group. There exist weighted variants of our statements: v is submultiplicative, i.e. v (x + y , ω + η) ≤ v (x, ω)v (y , η) for all (x, ω), (y , η) ∈ R2d . v (x, ω) ≥ 1 and v (−x, −ω) = v (x, ω) for all (x, ω) ∈ R2d . (`1v (Λ), c) is a Banach algebra with continuous involution. Gröchenig: A1v (Λ, c) is spectrally invariant in C ∗ (Λ, c) if and only if v is a GRS-weight, i.e. lim v (nλ)1/n = 1 for all λ ∈ Λ. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in a C ∗ -algebra Lemma: • Let g be in A VB . Then Pg := A hg , g i is a projection in A if and only if g hg , g iB = g . • Any element g in the unit sphere S(VB ) of the Hilbert B-module VB , i.e. of all g ∈ VB such that hg , g iB = IB , gives a projection Pg in A. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in a C ∗ -algebra Lemma: • Let g be in A VB . Then Pg := A hg , g i is a projection in A if and only if g hg , g iB = g . • Any element g in the unit sphere S(VB ) of the Hilbert B-module VB , i.e. of all g ∈ VB such that hg , g iB = IB , gives a projection Pg in A. −1/2 In particular g0 = g hg , g iB is in S(VB ), therefore Pg0 is a (canonical) projection in A. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in a C ∗ -algebra Lemma: • Let g be in A VB . Then Pg := A hg , g i is a projection in A if and only if g hg , g iB = g . • Any element g in the unit sphere S(VB ) of the Hilbert B-module VB , i.e. of all g ∈ VB such that hg , g iB = IB , gives a projection Pg in A. −1/2 In particular g0 = g hg , g iB is in S(VB ), therefore Pg0 is a (canonical) projection in A. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in noncommutative tori Recall p is called a projection if p = p ∗ = p 2 . Theorem: Let G(g , Λ) be a Gabor system on L2 (Rd ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in noncommutative tori Recall p is called a projection if p = p ∗ = p 2 . Theorem: Let G(g , Λ) be a Gabor system on L2 (Rd ).Then pg = Λ hg , g i is a projection in C ∗ (Λ, c) if and only if one of the following condition holds: Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in noncommutative tori Recall p is called a projection if p = p ∗ = p 2 . Theorem: Let G(g , Λ) be a Gabor system on L2 (Rd ).Then pg = Λ hg , g i is a projection in C ∗ (Λ, c) if and only if one of the following condition holds: G(g , Λ) is a tight Gabor frame for L2 (Rd ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in noncommutative tori Recall p is called a projection if p = p ∗ = p 2 . Theorem: Let G(g , Λ) be a Gabor system on L2 (Rd ).Then pg = Λ hg , g i is a projection in C ∗ (Λ, c) if and only if one of the following condition holds: G(g , Λ) is a tight Gabor frame for L2 (Rd ). G(g , Λ◦ ) is an orthogonal system. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in noncommutative tori Recall p is called a projection if p = p ∗ = p 2 . Theorem: Let G(g , Λ) be a Gabor system on L2 (Rd ).Then pg = Λ hg , g i is a projection in C ∗ (Λ, c) if and only if one of the following condition holds: G(g , Λ) is a tight Gabor frame for L2 (Rd ). G(g , Λ◦ ) is an orthogonal system. hg , g iΛ◦ = I . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in noncommutative tori Recall p is called a projection if p = p ∗ = p 2 . Theorem: Let G(g , Λ) be a Gabor system on L2 (Rd ).Then pg = Λ hg , g i is a projection in C ∗ (Λ, c) if and only if one of the following condition holds: G(g , Λ) is a tight Gabor frame for L2 (Rd ). G(g , Λ◦ ) is an orthogonal system. hg , g iΛ◦ = I . hg , π(λ◦ )g i = vol(Λ)δλ◦ ,0 for all λ◦ ∈ Λ◦ . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in noncommutative tori Recall p is called a projection if p = p ∗ = p 2 . Theorem: Let G(g , Λ) be a Gabor system on L2 (Rd ).Then pg = Λ hg , g i is a projection in C ∗ (Λ, c) if and only if one of the following condition holds: G(g , Λ) is a tight Gabor frame for L2 (Rd ). G(g , Λ◦ ) is an orthogonal system. hg , g iΛ◦ = I . hg , π(λ◦ )g i = vol(Λ)δλ◦ ,0 for all λ◦ ∈ Λ◦ . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in noncommutative tori Recall p is called a projection if p = p ∗ = p 2 . Theorem: Let G(g , Λ) be a Gabor system on L2 (Rd ).Then pg = Λ hg , g i is a projection in C ∗ (Λ, c) if and only if one of the following condition holds: G(g , Λ) is a tight Gabor frame for L2 (Rd ). G(g , Λ◦ ) is an orthogonal system. hg , g iΛ◦ = I . hg , π(λ◦ )g i = vol(Λ)δλ◦ ,0 for all λ◦ ∈ Λ◦ . In particular the canonical tight Gabor atom h0 := (SgΛ,g )−1/2 g yields a projection ph0 in C ∗ (Λ, c). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in noncommutative tori Recall p is called a projection if p = p ∗ = p 2 . Theorem: Let G(g , Λ) be a Gabor system on L2 (Rd ).Then pg = Λ hg , g i is a projection in C ∗ (Λ, c) if and only if one of the following condition holds: G(g , Λ) is a tight Gabor frame for L2 (Rd ). G(g , Λ◦ ) is an orthogonal system. hg , g iΛ◦ = I . hg , π(λ◦ )g i = vol(Λ)δλ◦ ,0 for all λ◦ ∈ Λ◦ . In particular the canonical tight Gabor atom h0 := (SgΛ,g )−1/2 g yields a projection ph0 in C ∗ (Λ, c). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in noncommutative tori – continued Lema: Let g be in Λ VΛ◦ . Then Pg := Λ hg , g i is a projection in C ∗ (Λ, c) if and only if g hg , g iΛ◦ = g . If g ∈ Ms1 (R) or S (R), then Pg gives a projection in A1s (Λ, c) or A∞ (Λ, c), respectively. First we assume that g hg , g iΛ◦ = g for some g in Λ VΛ◦ . Then we have that Pg2 = Λ hg , g iΛ hg , g i = Λ Λ hg , g ig , g = Λ hg hg , g iΛ◦ , g i = Λ hg , g i and Pg∗ = Pg . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in noncommutative tori – continued Lema: Let g be in Λ VΛ◦ . Then Pg := Λ hg , g i is a projection in C ∗ (Λ, c) if and only if g hg , g iΛ◦ = g . If g ∈ Ms1 (R) or S (R), then Pg gives a projection in A1s (Λ, c) or A∞ (Λ, c), respectively. First we assume that g hg , g iΛ◦ = g for some g in Λ VΛ◦ . Then we have that Pg2 = Λ hg , g iΛ hg , g i = Λ Λ hg , g ig , g = Λ hg hg , g iΛ◦ , g i = Λ hg , g i and Pg∗ = Pg . Now we suppose that Λ hg , g i is a projection in C ∗ (Λ, c). Then an elementary computation yields the assertion: Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in noncommutative tori – continued Lema: Let g be in Λ VΛ◦ . Then Pg := Λ hg , g i is a projection in C ∗ (Λ, c) if and only if g hg , g iΛ◦ = g . If g ∈ Ms1 (R) or S (R), then Pg gives a projection in A1s (Λ, c) or A∞ (Λ, c), respectively. First we assume that g hg , g iΛ◦ = g for some g in Λ VΛ◦ . Then we have that Pg2 = Λ hg , g iΛ hg , g i = Λ Λ hg , g ig , g = Λ hg hg , g iΛ◦ , g i = Λ hg , g i and Pg∗ = Pg . Now we suppose that Λ hg , g i is a projection in C ∗ (Λ, c). Then an elementary computation yields the assertion: Λ g hg , g iΛ◦ − g , g hg , g iΛ◦ − g = 0. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Projections in noncommutative tori – continued Lema: Let g be in Λ VΛ◦ . Then Pg := Λ hg , g i is a projection in C ∗ (Λ, c) if and only if g hg , g iΛ◦ = g . If g ∈ Ms1 (R) or S (R), then Pg gives a projection in A1s (Λ, c) or A∞ (Λ, c), respectively. First we assume that g hg , g iΛ◦ = g for some g in Λ VΛ◦ . Then we have that Pg2 = Λ hg , g iΛ hg , g i = Λ Λ hg , g ig , g = Λ hg hg , g iΛ◦ , g i = Λ hg , g i and Pg∗ = Pg . Now we suppose that Λ hg , g i is a projection in C ∗ (Λ, c). Then an elementary computation yields the assertion: Λ g hg , g iΛ◦ − g , g hg , g iΛ◦ − g = 0. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Consequences The unit sphere of the Hilbert C ∗ (Λ◦ , c)-module VΛ◦ is defined by S(VΛ◦ ) = {g ∈ VΛ◦ : hg , g iΛ◦ = I }, which is the set of all tight Gabor frames. 2 1 g1 (t) = (2)1/4 e −πt a Gaussian, g2 (t) = ( π2 )1/2 cosh(πt) the hyperbolic secant and g3 (t) = e −π|t| the two-sided exponential. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Consequences The unit sphere of the Hilbert C ∗ (Λ◦ , c)-module VΛ◦ is defined by S(VΛ◦ ) = {g ∈ VΛ◦ : hg , g iΛ◦ = I }, which is the set of all tight Gabor frames. 2 1 the g1 (t) = (2)1/4 e −πt a Gaussian, g2 (t) = ( π2 )1/2 cosh(πt) hyperbolic secant and g3 (t) = e −π|t| the two-sided exponential. Theorem: Let Λ = αZ × βZ. Then pgi = Λ hgi , gi i is a projection in C ∗ (Λ, c) if and only if αβ < 1. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Consequences The unit sphere of the Hilbert C ∗ (Λ◦ , c)-module VΛ◦ is defined by S(VΛ◦ ) = {g ∈ VΛ◦ : hg , g iΛ◦ = I }, which is the set of all tight Gabor frames. 2 1 the g1 (t) = (2)1/4 e −πt a Gaussian, g2 (t) = ( π2 )1/2 cosh(πt) hyperbolic secant and g3 (t) = e −π|t| the two-sided exponential. Theorem: Let Λ = αZ × βZ. Then pgi = Λ hgi , gi i is a projection in C ∗ (Λ, c) if and only if αβ < 1. The case of the Gaussian g1 is known as Boca’s projection. In Manin’s work pg1 = Λ hg1 , g1 i appear as quantum theta functions. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Consequences The unit sphere of the Hilbert C ∗ (Λ◦ , c)-module VΛ◦ is defined by S(VΛ◦ ) = {g ∈ VΛ◦ : hg , g iΛ◦ = I }, which is the set of all tight Gabor frames. 2 1 the g1 (t) = (2)1/4 e −πt a Gaussian, g2 (t) = ( π2 )1/2 cosh(πt) hyperbolic secant and g3 (t) = e −π|t| the two-sided exponential. Theorem: Let Λ = αZ × βZ. Then pgi = Λ hgi , gi i is a projection in C ∗ (Λ, c) if and only if αβ < 1. The case of the Gaussian g1 is known as Boca’s projection. In Manin’s work pg1 = Λ hg1 , g1 i appear as quantum theta functions. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Extensions Feichtinger-Kaiblinger showed in particular that the set of Gabor frames G(g0 , Λ) with g0 ∈ Ms1 (Rd ) is open, i.e. there exists a ε > 0 such that for kg − g0 kMs1 < ε then G(g , Λ) is also a Gabor frame. Consequently, we have an open set of projections Λ hg , g i close to Λ hg0 , g0 i. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Extensions Feichtinger-Kaiblinger showed in particular that the set of Gabor frames G(g0 , Λ) with g0 ∈ Ms1 (Rd ) is open, i.e. there exists a ε > 0 such that for kg − g0 kMs1 < ε then G(g , Λ) is also a Gabor frame. Consequently, we have an open set of projections Λ hg , g i close to Λ hg0 , g0 i. Actually, Feichtinger and Kaiblinger proved that one can also vary the lattice. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Extensions Feichtinger-Kaiblinger showed in particular that the set of Gabor frames G(g0 , Λ) with g0 ∈ Ms1 (Rd ) is open, i.e. there exists a ε > 0 such that for kg − g0 kMs1 < ε then G(g , Λ) is also a Gabor frame. Consequently, we have an open set of projections Λ hg , g i close to Λ hg0 , g0 i. Actually, Feichtinger and Kaiblinger proved that one can also vary the lattice. Therefore, a formulation in our setting amounts to incorporate the fact that noncommutative tori are also give rise to a continuous field of C ∗ -algebras. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Extensions Feichtinger-Kaiblinger showed in particular that the set of Gabor frames G(g0 , Λ) with g0 ∈ Ms1 (Rd ) is open, i.e. there exists a ε > 0 such that for kg − g0 kMs1 < ε then G(g , Λ) is also a Gabor frame. Consequently, we have an open set of projections Λ hg , g i close to Λ hg0 , g0 i. Actually, Feichtinger and Kaiblinger proved that one can also vary the lattice. Therefore, a formulation in our setting amounts to incorporate the fact that noncommutative tori are also give rise to a continuous field of C ∗ -algebras. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Extensions Feichtinger-Kaiblinger showed in particular that the set of Gabor frames G(g0 , Λ) with g0 ∈ Ms1 (Rd ) is open, i.e. there exists a ε > 0 such that for kg − g0 kMs1 < ε then G(g , Λ) is also a Gabor frame. Consequently, we have an open set of projections Λ hg , g i close to Λ hg0 , g0 i. Actually, Feichtinger and Kaiblinger proved that one can also vary the lattice. Therefore, a formulation in our setting amounts to incorporate the fact that noncommutative tori are also give rise to a continuous field of C ∗ -algebras. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Extensions Rieffel demonstrated that viewed as quantum metric spaces one is dealing with a continuous field of quantum metric spaces. Another problem fitting into this framework is the approximation of continuous Gabor frames by finite-dimensional Gabor frames. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Extensions Rieffel demonstrated that viewed as quantum metric spaces one is dealing with a continuous field of quantum metric spaces. Another problem fitting into this framework is the approximation of continuous Gabor frames by finite-dimensional Gabor frames. In other words one is looking to control the changes of projective modules over “close” noncommutative tori. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Extensions Rieffel demonstrated that viewed as quantum metric spaces one is dealing with a continuous field of quantum metric spaces. Another problem fitting into this framework is the approximation of continuous Gabor frames by finite-dimensional Gabor frames. In other words one is looking to control the changes of projective modules over “close” noncommutative tori. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Rotation algebras Theorem: L Let α be an irrational number. Then C ∗ (αZ, R/Z) and C ∗ (Z, R/αZ) are Morita-Rieffel equivalent. C ∗ (αZ, R/Z) is the C ∗ -algebra for Z acting on the circle by 2πα, and it is the completion of the ∗-algebra Cc (T × Z) where the convolution product is given by X F ∗ G (t, n) = F (m, t)G (n − m, e −2πimθ t) m∈Z and involution is given by F ∗ (n, t) = F (−n, −e −2πiθ )t Luef, Franz Hilbert C ∗ -modules over noncommutative tori Rotation algebras Theorem: L Let α be an irrational number. Then C ∗ (αZ, R/Z) and C ∗ (Z, R/αZ) are Morita-Rieffel equivalent. C ∗ (αZ, R/Z) is the C ∗ -algebra for Z acting on the circle by 2πα, and it is the completion of the ∗-algebra Cc (T × Z) where the convolution product is given by X F ∗ G (t, n) = F (m, t)G (n − m, e −2πimθ t) m∈Z and involution is given by F ∗ (n, t) = F (−n, −e −2πiθ )t C ∗ (Z, R/αZ) is the C ∗ -algebra for Z acting on the circle by 2π/α and it is isomorphic to C ∗ (α−1 Z, R/Z). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Rotation algebras Theorem: L Let α be an irrational number. Then C ∗ (αZ, R/Z) and C ∗ (Z, R/αZ) are Morita-Rieffel equivalent. C ∗ (αZ, R/Z) is the C ∗ -algebra for Z acting on the circle by 2πα, and it is the completion of the ∗-algebra Cc (T × Z) where the convolution product is given by X F ∗ G (t, n) = F (m, t)G (n − m, e −2πimθ t) m∈Z and involution is given by F ∗ (n, t) = F (−n, −e −2πiθ )t C ∗ (Z, R/αZ) is the C ∗ -algebra for Z acting on the circle by 2π/α and it is isomorphic to C ∗ (α−1 Z, R/Z). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita-Rieffel equivalence for rotation algebras Let A = C ∗ (αZ, R/Z) and B = C ∗ (Z, R/αZ). Then we define on Cc (R) the structure of a right pre A Hilbert module and a left pre B Hilbert module. For f , g ∈ Cc (R), F ∈ C ∗ (αZ, R/Z) and G ∈ C ∗ (Z, R/αZ) we define: Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita-Rieffel equivalence for rotation algebras Let A = C ∗ (αZ, R/Z) and B = C ∗ (Z, R/αZ). Then we define on Cc (R) the structure of a right pre A Hilbert module and a left pre B Hilbert module. For f , g ∈ Cc (R), F ∈ C ∗ (αZ, R/Z) and G ∈ C ∗ (Z, R/αZ) we define: X (f F )(t) = f (t − nα)F (nα, t − nα) n∈Z hf , g iA (mα, t) = X f (r − n)g (r − n + mα) n∈Z Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita-Rieffel equivalence for rotation algebras Let A = C ∗ (αZ, R/Z) and B = C ∗ (Z, R/αZ). Then we define on Cc (R) the structure of a right pre A Hilbert module and a left pre B Hilbert module. For f , g ∈ Cc (R), F ∈ C ∗ (αZ, R/Z) and G ∈ C ∗ (Z, R/αZ) we define: X (f F )(t) = f (t − nα)F (nα, t − nα) n∈Z hf , g iA (mα, t) = X f (r − n)g (r − n + mα) n∈Z (G f )(t) = X G (n, t)f (t − n) n∈Z B hf , g i(m, t) = X f (t − nα)g (t − nα − m). n∈Z Luef, Franz Hilbert C ∗ -modules over noncommutative tori Morita-Rieffel equivalence for rotation algebras Let A = C ∗ (αZ, R/Z) and B = C ∗ (Z, R/αZ). Then we define on Cc (R) the structure of a right pre A Hilbert module and a left pre B Hilbert module. For f , g ∈ Cc (R), F ∈ C ∗ (αZ, R/Z) and G ∈ C ∗ (Z, R/αZ) we define: X (f F )(t) = f (t − nα)F (nα, t − nα) n∈Z hf , g iA (mα, t) = X f (r − n)g (r − n + mα) n∈Z (G f )(t) = X G (n, t)f (t − n) n∈Z B hf , g i(m, t) = X f (t − nα)g (t − nα − m). n∈Z Luef, Franz Hilbert C ∗ -modules over noncommutative tori Wiener amalgam spaces In his work on Generalized Harmonic Analysis and Tauberian Theorems Norbert Wiener introduced the space W (L∞ , `1 ). Later H.-G. Feichtinger generalized Wiener’s space in his work on function spaces which allows one to measure global and local information W (B1 , B2 ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Wiener amalgam spaces In his work on Generalized Harmonic Analysis and Tauberian Theorems Norbert Wiener introduced the space W (L∞ , `1 ). Later H.-G. Feichtinger generalized Wiener’s space in his work on function spaces which allows one to measure global and local information W (B1 , B2 ). These spaces found applications in approximation theory, sampling theory, symbol classes of pseudo-differential operators, stochastic processes, etc. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Wiener amalgam spaces In his work on Generalized Harmonic Analysis and Tauberian Theorems Norbert Wiener introduced the space W (L∞ , `1 ). Later H.-G. Feichtinger generalized Wiener’s space in his work on function spaces which allows one to measure global and local information W (B1 , B2 ). These spaces found applications in approximation theory, sampling theory, symbol classes of pseudo-differential operators, stochastic processes, etc. We restrict our discussion to the case of weighted analogs of W (L∞ , `1 ). Luef, Franz Hilbert C ∗ -modules over noncommutative tori Wiener amalgam spaces In his work on Generalized Harmonic Analysis and Tauberian Theorems Norbert Wiener introduced the space W (L∞ , `1 ). Later H.-G. Feichtinger generalized Wiener’s space in his work on function spaces which allows one to measure global and local information W (B1 , B2 ). These spaces found applications in approximation theory, sampling theory, symbol classes of pseudo-differential operators, stochastic processes, etc. We restrict our discussion to the case of weighted analogs of W (L∞ , `1 ). Let v be a weight on Z such that: • v (k) = v (−k), • v (k + l) ≤ v (k)v (l), • limn→∞ v (kn)1/n = 1. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Wiener amalgam spaces In his work on Generalized Harmonic Analysis and Tauberian Theorems Norbert Wiener introduced the space W (L∞ , `1 ). Later H.-G. Feichtinger generalized Wiener’s space in his work on function spaces which allows one to measure global and local information W (B1 , B2 ). These spaces found applications in approximation theory, sampling theory, symbol classes of pseudo-differential operators, stochastic processes, etc. We restrict our discussion to the case of weighted analogs of W (L∞ , `1 ). Let v be a weight on Z such that: • v (k) = v (−k), • v (k + l) ≤ v (k)v (l), • limn→∞ v (kn)1/n = 1. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Wiener amalgam spaces - continued For α > 0 we set Iα = [0, α) and denote by χα the characteristic function of Iα . A function f belongs to W (L∞ , `1v ) if X kf kW (L∞ ,`1v ) = kf · Tkα χα kL∞ v (k) < ∞. k∈Z Luef, Franz Hilbert C ∗ -modules over noncommutative tori Wiener amalgam spaces - continued For α > 0 we set Iα = [0, α) and denote by χα the characteristic function of Iα . A function f belongs to W (L∞ , `1v ) if X kf kW (L∞ ,`1v ) = kf · Tkα χα kL∞ v (k) < ∞. k∈Z The definition is independent of α and we have the following embeddings: S ⊂ W (L∞ , `1v ) ⊂ W (L∞ , `1 ) ⊂ L2 . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Wiener amalgam spaces - continued For α > 0 we set Iα = [0, α) and denote by χα the characteristic function of Iα . A function f belongs to W (L∞ , `1v ) if X kf kW (L∞ ,`1v ) = kf · Tkα χα kL∞ v (k) < ∞. k∈Z The definition is independent of α and we have the following embeddings: S ⊂ W (L∞ , `1v ) ⊂ W (L∞ , `1 ) ⊂ L2 . Luef, Franz Hilbert C ∗ -modules over noncommutative tori Walnut representation of Gabor frame operator Recall B = C ∗ (Z, R/αZ). Lemma: Let g , h be in W (L∞ , `1v ). Then hg , hiB is in L∞ (R). Theorem: If f , g , h are in W (L∞ , `1v ), then X hg , hiB f = hf , Tαk Ml g iTαk Ml h =: Sg ,h f . k,n∈Z In time-frequency analysis the representation in the last Theorem is called the Walnut representation of a Gabor frame operator. Luef, Franz Hilbert C ∗ -modules over noncommutative tori Walnut representation of Gabor frame operator Recall B = C ∗ (Z, R/αZ). Lemma: Let g , h be in W (L∞ , `1v ). Then hg , hiB is in L∞ (R). Theorem: If f , g , h are in W (L∞ , `1v ), then X hg , hiB f = hf , Tαk Ml g iTαk Ml h =: Sg ,h f . k,n∈Z In time-frequency analysis the representation in the last Theorem is called the Walnut representation of a Gabor frame operator. Luef, Franz Hilbert C ∗ -modules over noncommutative tori