arXiv:submit/0873263 [math.QA] 18 Dec 2013 aie(o-unu)cs,teln spaces lens the case, (non-quantum) tative unu esspaces lens quantum esuyuiayeuvlne fsc unu esspaces. lens quantum such j of quantum equivalences triple, on unitary spectral geometries study real spectral we a irreducible of the classify conditions We the of some weakening mrhci n nyif only and if omorphic esuyams elseta rpe on triples t spectral In real manifolds almost type. simplest study homeomorphism we and the type of homotopy L some between difference [15]. are Dyck they Walther by because 1884 interesting in introduced first were spheres, ia pcrmo h aesae hsi rtse oad d towards step first a teardrops. is quantum This these on space. triples base resul spectral this the real Using on fibrations [5]. spectrum Seifert from Dirac are result spaces a lens generalizing quantum teardrops, such all that show ffe cin fcci ruso h pcrlgoer nt on geometry spectral the on groups cyclic of group actions free of a oei re ocp ihcran“og-hoes,whic go-theorems”, “no certain ope with compact [8 cope to of to up sense order algebra the in the in done with was triple commutes spectral only real a algebra not site is This [11]. from eso htif that show We essae,obtsae ffe cin fcci ruson groups cyclic of actions free of spaces orbit spaces, Lens ercl h osrcino h qiain elspectral real equivariant the of construction the recall We ELSETA RPE N3-DIMENSIONAL ON TRIPLES SPECTRAL REAL SU Abstract. n aclt h ia pcrmo h aesaecmn fro coming space base the on spectrum Dirac fibration. quantum the over fibrations calculate Seifert and are spaces lens quantum such esspaces lens esuyuiayeuvlne fsc unu essae.W spaces. lens quantum such len W of if quantum that equivalences triple. on unitary spectral triples study real spectral we real a almost of irreducible conditions the the of some weakening sobtsae ffe cin fcci ruso h spectr the on groups cyclic group of quantum actions the free on of try spaces orbit as q 2,a osrce n[1.Teeseta rpe r give are triples spectral These [11]. in constructed as (2), 1. r OCMUAIELN SPACES. LENS NONCOMMUTATIVE h qiain pcrltil on triple spectral equivariant The NRE IAZADJNJTEVENSELAAR JITSE JAN AND SITARZ ANDRZEJ scpieto coprime is L esuyams elseta rpe nqatmln spaces lens quantum on triples spectral real almost study We q r ( ,r p, scpieto coprime is L and ) q ( r ,r p, p · the , r L and ) ′ q ( ± ≡ SU p, C )aeioopi.As,w hwta all that show we Also, isomorphic. are 1) ∗ q agba orsodn otequantum the to corresponding -algebras 2.Teeseta rpe r ie by given are triples spectral These (2). L mod 1 p q the , 1 ( p, unu esspaces lens quantum )aeioopi.I h commu- the In isomorphic. are 1) p C ∗ or agba orsodn othe to corresponding -algebras L ( r ,r p, ± and ) r ′ ≡ mod 0 ,w aclt the calculate we t, L SU pcsand spaces s ,sneteoppo- the since ], ( teardrops, rpeon triple ,r p, sobtspaces orbit as , lgeome- al odd-dimensional essae and spaces lens q s iei [11]. in like ust classify e (2) xiiigthe exhibiting hwdthat showed h vrquantum over n pcsare spaces ens show e this m ′ equantum he p r home- are ) aos This rators. lo we Also, . i article his fiiga efining , SU by n q (2) 2 A.SITARZANDJ.J.VENSELAAR it was impossible for a suq(2)-equivariant spectral to satisfy all conditions of a real spectral triple [11, Remark 6.6]. Let q denote a real number, 0 This algebra has irreducible finite dimensional representations σl on the 2l+1 1 3 dimensional vector space Vl, labeled by half-integers l = 0, 2 , 1, 2 ,... [26]. Let |l,mi, (where m = −l, −l+1,...l) be a basis for the irreducible Uq(su(2))- qn−q−n module Vl. Define [n] = [n]q = q−q−1 . The representation σl on Vl is given by m σl(k)|l,mi = q |l,mi, (5) σl(f)|l,mi = [l − m]q[l + m + 1]q|l,m + 1i, q σl(e)|l,mi = [l − m + 1]q[l + m]q|l,m − 1i. q We can define two commuting actions of su(2) on the algebra A(SUq(2)), as follows. One is the usual left action ⊲, defined by 1 ∗ − 1 ∗ − 1 ∗ 1 ∗ k ⊲ a = q 2 a, k ⊲ a = q 2 a , k ⊲ b = q 2 b, k ⊲ b = q 2 b , (6) f ⊲ a = 0, f ⊲ a∗ = −qb∗, f ⊲ b = a, f ⊲ b∗ = 0, e ⊲ a = b, e ⊲ a∗ = 0, e ⊲ b = 0, e ⊲ b∗ = −q−1a∗. The other action, commuting with the previous one, is given by applying the order 2 automorphism θ of su(2), which maps e to −f and maps k to QUANTUMLENSSPACES 3 k−1, and the inverse of the antipode to the usual right action. As can be checked, this gives a left representation ·, commuting with (6): 1 ∗ − 1 ∗ 1 ∗ − 1 ∗ k · a = q 2 a, k · a = q 2 a , k · b = q 2 b, k · b = q 2 b , (7) f · a = 0, f · a∗ = qb∗, f · b = 0, f · b∗ = −a, e · a = −b∗, e · a∗ = 0, e · b = q−1a∗, e · b∗ = 0. There is a vector space basis eklm of A(SUq(2)), given by monomials of the form akblb∗m k ∈ Z, k ≥ 0,l,m ∈ N (8) eklm := l ∗m ∗−k (b b a k ∈ Z,k < 0,l,m ∈ N There is a so-called Haar state, which is the unique [33, Theorem 4.2] ∗ bi-invariant linear functional ψ on the C -completion of A(SUq(2)), given by [33, Appendix A1]: 0 when k > 0 or l 6= m ψ(eklm)= 1−q2 ( 1−q2m+2 when k = 0 and l = m . Just as in the commutative (q → 1) case, the GNS-representation space 2 V = L (SUq(2), ψ) with respect to the Haar state ψ has a Peter-Weyl de- composition ∞ V = Vl ⊗ Vl, 2Ml=0 by a suitable analogue of the Peter-Weyl theorem for compact quantum groups [35, Section 6]. We abbreviate a vector |lmi ⊗ |lni ∈ Vl ⊗ Vl by |lmni. On this space we have two commuting left su(2)-actions, λ, ρ, given on the subspace Vl ⊗ Vl by (9) λ := id ⊗ σl, ρ := σl ⊗ id. These two commuting actions are an extension to q 6= 1 of the classical case where we can identify SU(2) = S3 = Spin(4)/Spin(3) = (SU(2) × SU(2))/SU(2), with the action of Spin(4) on SU(2) given by (g, h)x = gxh−1. We recall the definition of a (λ, ρ)-equivariant representation [11, Defini- tion 3.2]: Definition. Let λ and ρ be two commuting representations of a Hopf- algebra H on a vector space V . A representation Π of a ∗-algebra A on V is (λ, ρ)-equivariant if the following compatibility relations hold: (λ(h)Π(a)) v ⊗ w = π h(1) · a λ h(2) v, (10) (ρ(h)Π(a)) v ⊗ w = π h(1) ⊲ a ρ h(2) v, for all h ∈ H, a ∈A and v ∈ V . 4 A.SITARZANDJ.J.VENSELAAR In our case, the two commuting actions λ and ρ on the Hilbert space are given by (9). In the case of SUq(2), we need to consider two copies [11] of the Hilbert space: ∞ 2 (11) H = V ⊗ C ≃ V ⊗ V 1 ≃ V 1 ⊕ V 1 ⊗ Vj ⊕ V 1 ⊗ Vj 2 2 j+ 2 j− 2 2Mj=1 For convenience, we rename the spaces on the right hand side as ↑ ↑ ↓ (12) H = H0 ⊕ Hj ⊕Hj , ≥ 2Mj 1 ↑ ↓ with H ≃ V 1 ⊗ Vj and H ≃ V 1 ⊗ Vj . j j+ 2 j j− 2 The definition of the (λ, ρ) action on V in (9) can be amplified to H in several ways. Following [11], we define: Definition. The representations λ and ρ of Uq(su(2)) on V are amplified to λ′ and ρ′ on H, as ′ λ (h) := λ ⊗ σ 1 (∆h)= λ(h(1)) ⊗ σ 1 (h(2)), 2 2 (13) ρ′(h) := ρ(h) ⊗ id. It is easy to check that λ′ and ρ′ commute. We now define an explicit basis of H which is suited to (λ′, ρ′)-equivariance, i.e. basis vectors are eigenvectors for λ′(k) and ρ′(k) and λ′(f), ρ′(f), λ′(e) ′ ± 1 and ρ (e) are ladder operators. We use the shorthand k = k ± 2 . Set 1 1 2 2 −(j+µ)/2 [j − µ] (j−µ)/2 [j + µ] (14) Cjµ := q 1 and Sjµ := q 1 . [2j] 2 [2j] 2 1 1 Using the decomposition of H of (11), we define for j = l + 2 , µ = m − 2 , 1 1 µ = −j,...,j and n = −j + 2 ,...,j − 2 : 1 1 1 1 (15) |jµn ↓i := C |j−µ+ni ⊗ | , − i + S |j−µ−ni ⊗ | , + i, jµ 2 2 jµ 2 2 ± 1 − + − 1 with m = m ± 2 , and |j µ ni ∈ Vj . For j = l − 2 , µ as before and 1 1 n = −j − 2 ,...,j + 2 : 1 1 1 1 (16) |jµn ↑i := −S |j+µ+ni ⊗ | , − i + C |j+µ−ni ⊗ | , + i. jµ 2 2 jµ 2 2 The action of the elements of Uq(su(2)) can now be calculated, and it is easily shown that the |jµn ↑i and |jµn ↓i are joint eigenvectors for λ′(k) and ρ′(k) with eigenvalues qµ for λ′ and qn for ρ′. Summarizing, we have the following definition for the Hilbert space H: QUANTUMLENSSPACES 5 1 3 Definition. Let j = 0, 2 , 1, 2 ,..., and let µ = −j, −j + 1,...,j and n = 1 1 1 −j − 2 , −j + 2 ,...,j + 2 . The Hilbert space H of the real spectral triple (A(SUq(2)), H,D,J) is spanned by basis vectors |jµn ↑i (17) |jµnii := |jµn ↓i 1 with the convention that the |jµn ↓i component is zero when n = ±(j + 2 ) or j = 0. An equivariant representation of the algebra A(SUq(2)) on this Hilbert is given by: Proposition 1.1 ([11]). The following representation Π of A(SUq(2)) on the Hilbert space H with orthonormal basis |jµnii, is equivariant with respect ′ ′ to the (λ , ρ )-action of Uq(su(2)). + + + + − − + + (18a) Π (a) |j,µ,nii = αjµn|j µ n ii + αjµn|j µ n ii + + + − − − + − (18b) Π (b) |j,µ,nii = βjµn|j µ n ii + βjµn|j µ n ii ∗ + + − − − − − − (18c) Π (a ) |j,µ,nii =α ˜jµn|j µ n ii +α ˜jµn|j µ n ii ∗ ˜+ + − + ˜− − − + (18d) Π (b ) |j,µ,nii = βjµn|j µ n ii + βjµn|j µ n ii ± ± where αjµn and βjµn are, up to phase factors depending only on j, the fol- lowing triangular 2 × 2 matrices: 1 3 2 −j− 1 [j+n+ 2 ] 1 1 q 2 0 + (µ+n− )/2 2 [2j+2] (19a) α = q 2 [j + µ + 1] 1 1 jµn − 1 2 1 2 1 j n+ 2 − j+n+ 2 2 [ ] j [ ] q [2j+1][2j+2] q [2j+1] 1 1 − 1 2 1 2 j+1 [j n+ 2 ] 1 [j+n+ 2 ] 1 1 q −q 2 − (µ+n− )/2 2 [2j+1] [2j][2j+1] (19b) α = q 2 [j − µ] 1 jµn − − 1 2 1 j n 2 j+ 2 [ ] 0 q [2j] 1 − 3 2 [j n+ 2 ] 1 1 0 + (µ+n− )/2 2 [2j+2] (19c) β = q 2 [j + µ + 1] 1 1 jµn 1 2 − 1 2 − − j+n+ 2 − 1 j n+ 2 j 1 [ ] 2 [ ] −q [2j+1][2j+2] q [2j+1] 1 1 1 2 − 1 2 − 1 [j+n+ 2 ] j [j n+ 2 ] 1 1 −q 2 −q − (µ+n− )/2 2 [2j+1] [2j][2j+1] (19d) β = q 2 [j − µ] 1 jµn − 1 2 [j+n 2 ] 0 − [2j] and the remaining matrices are the hermitian conjugates ± ∓ † ˜± ∓ † α˜jµn = (αj±µ−n− ) , βjµn = (βj±µ−n+ ) . The Dirac operator of the spectral triple constructed in [11, Section 5] is given by: |jµn ↑i 2j + 3 |jµn ↑i (20) D = 2 . |j′µ′n′ ↓i − 2j′ + 1 |j′µ′n′ ↓i 2 6 A.SITARZANDJ.J.VENSELAAR The reality operator, constructed in [11, Section 6] is given by: |jµn ↑i i2(2j+µ+n)|j, −µ, −n ↑i (21) J = ′ ′ ′ . |j′µ′n′ ↓i i2(2j −µ −n )|j′, −µ′, −n′ ↓i Together, the algebra A(SUq(2)), with representation as given in Propo- sition 1.1, Dirac operator D given by (20), and reality operator J given by (21), satisfy most conditions as stated in [8], with some slight modifica- tions. These modifications are that [Π(x), JΠ(y∗)J †] and [[D, Π(x)], JΠ(y∗)J †] are not exactly 0, but lie in the two-sided ideal Kq in B(H) generated by the compact, positive, trace class operators j (22) Lq : Lq|j,µ,nii = q |j,µ,nii. This means that the first order and reality conditions should be modified appropriately. Also, it is currently unknown whether the Hochschild cycle condition and the Poincar´eduality condition are satisfied. For a discussion of these conditions for SUq(2), see Remark 3.2. 2. Topological quantum lens spaces Commutative lens spaces are defined as the quotient of an odd-dimensional sphere by a free action of a finite cyclic group. We will only consider 3- dimensional lens spaces here. Let p, r1, r2 be integers such that gcd(r1,p)= gcd(r2,p) = 1. 3 2 2 2 We view S as the points (z1, z2) ∈ C such that |z1| + |z2| = 1. We can define an action of the cyclic group Z/pZ on S3 as 2πi 2πi r1 r2 (z1, z2) 7→ (e p z1, e p z2). 3 The lens space L(p; r1, r2) is then defined as the quotient of S by this action of the cyclic group Z/pZ. −1 For 3-dimensional lens spaces we see that L(p; r1, r2)= L(p; 1, r1 r2) by 2πi −1 p r1 −1 multiplying each component by e (where r1 is meant mod p). We −1 will thus write L(p, r) for the lens space L(p; r1, r2), with r = r1 r2. Since S3 ≃ SU(2), it is natural to define quantum lens spaces as quotients of SUq(2) by free actions of finite cyclic groups (the analogy is z1 ↔ a, and z2 ↔ b). We want to consider quotients of the algebra A(SUq(2)) under actions of the group Z/pZ. In order to do this, we need to embed Z/pZ into Aut(SUq(2)). Since there exists only one non-trivial outer automorphism of A(SUq(2)) [19, Proposition 3.1], namely the map of order two induced by ∗ a 7→ a, b 7→ b , we can fit the automorphisms of A(SUq(2)) into the following exact sequence: (23) 0 → Inn(A(SUq(2))) → Aut(A(SUq(2))) → Z/2Z⋉ U(1) → 0, with U(1) acting by multiplying the a and b generators by complex num- bers of modulus 1. To maximize the analogy with the commutative case, we QUANTUMLENSSPACES 7 consider actions of cyclic groups Z/pZ embedded in U(1) as outer automor- phisms. Let g denote a generator of Z/pZ, and let r be an integer relatively prime to p. Let ǫ(r) = (r + 1) mod 2. We have: Proposition 2.1. The following action of Z/pZ on the algebra A(SUq(2)): 2πi (24a) υ(g) ⊲ a = e p a 2πi r (24b) υ(g) ⊲ b = e p b. generates a unitary representation υ of Z/pZ on H given by: 2πi (1+r)µ+(1−r)n+ 1 ǫ(r) (25) υ(g)|j,µ,nii = e p ( 2 )|j,µ,nii. The presence of ǫ(r) guarantees that the expression in brackets is an integer. Proof. Since Z/pZ is a group, we demand that the action of A on H is equivariant with respect to the group viewed as a Hopf algebra, with co- product ∆(g) = g ⊗ g. From this it follows that υ(g)|jµnii = cjµn|jµnii, for some cjµn ∈ C, since otherwise this would not be compatible with the 2πi p equivariance condition υ(g)Π(a)v = Π(g(1)a)υ(g(1))v, where g(1)a is e a, by Proposition 1.1. We can then calculate: 2πi υ(g)Π(a)|jµnii = Π e p a υ(g)|jµnii, 2πi r υ(g)Π(b)|jµnii = Π e p b υ(g)|jµnii. From (18) we see that the action of a and a∗ on H leaves the difference µ−n constant, and b leaves the sum µ + n constant. 2πi We get the recurrence relations υ(g)|j±µ+n+ii = e p υ(g)|jµnii and 2πi r υ(g)|j±µ+n−ii = e p υ(g)|jµnii. Solving these recurrence relations, we 2πi ((µ+n)+r(µ−n))+c see that υ(g)|jµnii = e p , where c is any constant. In order to have gp = 1, we see that c = ǫ(r) plus an additional integer, which can be set to zero. Proposition 2.2. The above defined action on the SUq(2) extends to its C∗-algebra and is free in the sense of Ellwood [16]. Proof. First of all, we can easily translate the action of Z/pZ to the right coaction of C(Z/pZ) on the SUq(2) algebra: ∆Rx = (g⊲x) ⊗ δg. Z Z g∈X/p Recall that that the freeness of a coaction of a Hopf algebra H on a C∗- algebra A means (for a right coaction) that the spans of (A ⊗ id)∆R(A) and ∆R(A)(A ⊗ id) are dense in A ⊗ H for a minimal tensor product. 8 A.SITARZANDJ.J.VENSELAAR For the action, which defines lens spaces the freeness is easy to verify. Consider the identity: (aa∗ + bb∗)r = 1, which could be rewritten, using the commutation relations as: ar(a∗)r + bP (a, a∗, b, b∗) = 1, where P is some polynomial in the generators. Therefore, p−1 r ∗ r ∗ ∗ 2πi rk ∆R(a ) ((a ) ⊗ 1)+∆R(b) (P (a, a , b, b ) ⊗ 1) = e p ⊗ δk = 1 ⊗ f, Xk=0 where f is the function on Z/pZ: 2πi rk f(k)= e p . As for r > 0 and p relatively prime the function f generates the algebra C(Z/pZ) this finishes the proof. Observe that if we replace r by r − p the action on the generators does not change. If we take p even then r is necessarily odd, and ǫ(r) is 0. If r is even, then p is necessarily odd, and r − p is odd, and on the Hilbert space the actions determined by (p, r) and (p, r − p) are equivalent since: − 2πi ((1+(r−p))µ+(1−(r−p))n) 2πip(n µ) − πiǫ(r) 2πi (1+r)µ+(1−r)n+ 1 ǫ(r)) e p = e p e p e p ( 2 − πiǫ(r) = −e p ρ(g), 1 Z Z where we have used that n − µ ∈ 2 \ . Since p − πiǫ(r) (26) −e p = 1, for odd p, the actions are equivalent. For this reason, we can always take r odd, and ǫ(r) = 0. Let K = 0, 1, 2,...,p − 1. We define HK as the 2πiK eigensubspace of H for the action of g with eigenvalue e p . Further, let Lq(p, r) be the subalgebra of A(SUq(2)) which is invariant under the action of Z/pZ. Proposition 2.3. For each K = 0, 1, 2,...,p − 1 and each x ∈ Lq(p, r), Π(x)HK ⊂HK. Proof. From (8) we know that A(SUq(2)) is generated as a vector space by the elements akblb∗m and blb∗ma∗n for k,l,m,n ∈ N. Since p and r are taken to be relatively prime, we see from (24a) and (24b) that L(p, r) is generated by elements akblb∗m with k + r(l − m) ≡ 0 mod p and blb∗ma∗n with r(l − m) − n ≡ 0 mod p. From (18), we can calculate that the action of akblb∗m on µ and n is given by 1 1 1 1 1 1 µ 7→ µ + k + l − m, n 7→ n + k − l + m. 2 2 2 2 2 2 QUANTUMLENSSPACES 9 2πi K Then if υ(g)|j,µ,nii = e p |j,µ,nii, we have (1+r)µ+(1−r)n ≡ K mod p and it then follows that 1 1 1 1 1 1 (1 + r)(µ + k + l − m)+(1 − r)(n + k − l + m)= 2 2 2 2 2 2 = (1+ r)µ + (1 − r)n + k + r(l − m) ≡ K mod p. Proposition 2.4. The equivariant real structure J, as given in (21) satis- fies: (27) JHK = HK′ , where K + K′ ≡ 0 mod p. Proof. We have J|jµnii = cjµn|j − µ − nii with cjµn a complex number, and from (1 + r)µ + (1 − r)n = K mod p, it follows that (1 + r) · (−µ) + (1 − r) · (−n)= −K mod p. 3. Geometry of quantum lens spaces We now turn to the construction of the almost real spectral triple of the quantum lens space. As stated at the end of Section 1, we modify some conditions of a real spectral triple, exactly as in [11], i.e. the real structure and the first order condition only hold up to the compact operators of positive trace class defined in (22). Furthermore, it is unknown if the Hochschild cycle condition and the Poincar´eduality are satisfied for SUq(2). This means that also for Lq(p, r) we do not know if they are satisfied. We call a structure, satisfying all conditions of [8], with the modification of the first order condition, and the removal of the Hochschild cycle condition and Poincar´eduality a almost real spectral triple. Proposition 3.1. Let Lq(p, r), q ∈ (0, 1), be the quantum lens space as defined above. Then for any K = 0, 1,...p − 1, the Hilbert space HK ⊕HK′ , where K + K′ ≡ 0 mod p, the reality structure J and the Dirac operator D taken as the restrictions of J and D from the A(SUq(2)) spectral geometry constitute a spectral geometry over the quantum lens space Lq(p, r). Proof. Almost all the usual conditions for a real spectral triple are easily seen to carry over from the SUq(2) case. For completeness, we list them here, and give short arguments why they carry over. • The compact resolvent and dimension growth follow from the fact that for j big enough, there always exist µ and n such that there are vectors |j,µ,nii ∈ HK, and thus also in HK′ , hence the dimension growth is satisfied. The compact resolvent condition also follows from this, and the fact that the dimension of the kernel of D still is finite dimensional. 10 A.SITARZANDJ.J.VENSELAAR • The commutation relations between J and D and the sign of J 2 are trivially the same as for SUq(2). • The algebra satisfies the regularity condition: [D, a] is a bounded operator on H for all a ∈A and both a and [D, a] belong to the do- ∞ k main of smoothness k=1 Dom(δ ) of the derivation δ, with δ(T )= [|D|, T ]. This condition is satisfied since Lq(p, r) is a subalgebra of T the algebra A(SUq(2)) which satisfies this condition by [32, Propo- sition 2.1]. The opposite algebra and first order condition are weakened as described in [11]: the opposite algebra condition [a, Jb∗J −1] = 0 and the first order ∗ −1 condition [[D, a], Jb J ] = 0 should hold up to compact operators in Kq as defined in (22). We should check whether this condition still holds for the lens space. First, observe that the Lq operators are unchanged, since they only see the j component, which does not play a role in the υ(g) action. In fact, due to [11, Proposition 7.2] for the generators a, a∗, b, b∗, the o 2 commutation relations are given by [π(x),π (y)] = LqA, with A a bounded operator depending on x and y. Since for any a, b in a ring R, we have that if [a, b] ∈ I, with I an ideal of R, we have [am, bn] ∈ I by repeatedly applying [a, b] ∈I, the generators of Lq(p, r) as described in Proposition 2.3 also satisfy this condition. Remark 3.2. The Hochschild cycle condition, the analogue of having a nowhere vanishing volume form, is problematic in the same way as it is for SUq(2) alone. Certainly, as shown in [27, Proposition 1.2], for SUq(2) there are no nontrivial Hochschild cycles n-cycles for n > 1. However, the original Hochschild cycle condition does not require the cycle to be nontrivial. It remains yet to be verified whether a (possibly modified) version of the Hochschild cycle condition holds for SUq(2) and the quantum lens spaces. The “dimension drop” phenomenon alone can be addressed by using twisted Hochschild homology as in [20, Theorem 1.2], where an analogue of the volume form, dA was constructed in twisted Hochschild homology for SUq(2). However, the relevant setup for the that would be this of modular Fredholm modules and modular spectral triples [1, 30]. 3.0.1. Irreducibility. Though any of the above listed spectral geometries for the quantum lens spaces is admissible from the point of view of the non- commutative axiomatic approach, not all correspond to spin structures on commutative lens spaces. We demand that our spectral triple be irreducible as the analogue of a connected manifold in commutative geometry. If we use [18, Definition 11.2] or [25, Definition 2.1], all real spectral triples above are irreducible, since the J operator interchanges the HK and HK′ spaces. If we use the definition of [9, Remark 6 on p.163], only irreducibility with respect to the algebra action and Dirac operator are demanded. None of the above real spectral triples are then irreducible, however the cases where K = K′ = 0 and K = K′ = p/2 if p is even can be made irreducible by QUANTUMLENSSPACES 11 dropping one of the two copies of the Hilbert space, and setting the real spectral triple to be (Lq(p, r), HK ,D,J). Since JHK ⊂ HK in this case, this is a well-defined spectral triple, and irreducible. For even p we obtain two possible spin structures, for odd p just one, just as in the commutative case [17]. Theorem 3.3. The quantum lens space Lq(p, r) admits one irreducible al- most real spectral triple if p is odd, and two if p is even. The spectral geometries are given in the two cases by • p mod2=1: (Lq(p, r), H0,D,J). • p = 2P : (Lq(p, r), H0,D,J) and (Lq(p, r), HP ,D,J). with D the Dirac operator as described in (20) and J the operator given in (21). The spectrum of the Dirac operator. Let us recall that the spectrum of the Dirac operator over A(SUq(2)) (with appropriate normalization) is given by: 3 1 D|j, µ, n, ↑i = 2j + |j, µ, n, ↑i with j = 0, ,... 2 2 with multiplicity: (2j + 1)(2j + 2) 1 1 D|j, µ, n, ↓i = − 2j + |j, µ, n, ↓i with j = , 1,... 2 2 with multiplicity: (2j + 1)2j Note that the spectrum is symmetric. Since in the construction of the spec- tral geometries on quantum lens spaces we keep the Dirac operator of SUq(2) (and only restrict the Hilbert space), the spectrum remains unchanged, only the multiplicities differ. We shall reduce the problem of computing these multiplicities to a number theoretic problem of solving congruence relations. Proposition 3.4. The eigenvalues of D belonging to an irreducible almost real spectral triple as in Theorem 3.3 with Hilbert space HK (K = 0 or K = 1 3 1 + − 2 p) are 2j + 2 and −2j − 2 , with respective multiplicity N (j) and N (j), (either of them could be 0, which means that the value is not present in the spectrum) where N ±(j) denotes the number of solutions to the equation: (28) (1 + r)µ + (1 − r)n ≡ K mod p 1 1 with −j ≤ µ ≤ j and −(j ± 2 ) ≤ n ≤ j ± 2 . The exact calculation of the number of eigenvalues is a tedious task. The solution depends heavily on the properties of (1+r) and (1−r), in particular on the greatest common divisor of 1 ± r and p. Although in each case the explicit solutions for µ and n can be easily found, calculating the number of solutions for a given j is rather difficult for an abstract choice of r and p. For illustration, we show here some pictures of what the spectra for small p looks like. We represent the basis vectors |jµn ↑i, |jµn ↓i of the Hilbert space H of the spectral triple on SUq(2) as defined in Proposition 1.1 as a 12 A.SITARZANDJ.J.VENSELAAR lattice, with µ on the horizontal axis and n on the vertical axis. We project the j coordinate away, since it does not play a role for determining whether a vector lies in HK, only in confining the possible µ and n. We illustrate this by drawing lines through the allowed values (in the form of rectangles around the origin) for j = 3 and in the ↑ part of H. We draw a circle through the origin µ = 0,n = 0. The stars ( ) and diamonds( ) represent basis vectors of the HK Hilbert space of the lens space (Lq(p, r), HK ,D,J). The circles ( ) represent basis vectors of the Hilbert space of SUq(2) which are not part of the lens space. The stars are the allowed values for integer valued j, the diamonds are the allowed values for half-integer valued j. Figure 1. H0 for Figure 2. H1 for p = 2, r = 1 p = 2, r = 1 Figure 3. H0 for Figure 4. H0 for p = 5, r = 1 p = 5, r = −3 Example: the L(p,p−1) lens space. This corresponds to the choice r = p−1 or equivalently r = −1 (then ǫ(r) = 0). Proposition 3.5. The spectrum of the Dirac operator on L(p,p − 1) is: QUANTUMLENSSPACES 13 Figure 5. H0 for Figure 6. H0 for p = 7, r = −5 p = 7, r = 3 • for p ≡ 1 mod 2: 1 2kp + 2l + 2 , with multiplicity: 2(2k + 1)(kp + l) k = 0, 1,... l = 0, 1,...,p − 1 (2k + 1)p + 2l + 1 , with multiplicity: (2k + 2)((2k + 1)p + 2l) 2 k = 0, 1,... l = 0, 1,...,p − 1 λD = 3 −2kp − 2l − , with multiplicity: 2(2k + 1)(kp + 2l + 1) 2 k = 0, 1,... l = 0, 1,...,p − 1 1 −(2k + 1)p − 2l + 2 , with multiplicity: (2k + 2)((2k + 1)p + 2l + 2) k = 0, 1,... l = 0, 1,...,p − 1 1 Note that the λD = 2 eigenvalue (k = l = 0) has multiplicity 0. • for p = 2P and K = 0 we have 1 2kp + 2l + 2 with multiplicity: 2(2k + 1)(kp + l) k = 0, 1,... l = 0, 1,...,p − 1 λD = 3 −2kp − 2l − with multiplicity: 2(2k + 1)(kp + l + 1) 2 k = 0, 1,... l = 0, 1,...,p − 1 1 Again, the λD = 2 eigenvalue has multiplicity 0 (k = l = 0). • for p = 2P and and K = P : 1 (2k + 1)P + 2l + 2 with multiplicity: (2k + 2)((2k + 1)p + 2l) k = 0, 1,... l = 0, 1,...,P − 1 λD = 3 −(2k + 1)P − 2l − with multiplicity: (2k + 2)((2k + 1)p + 2l + 2) 2 k = 0, 1,... l = 0, 1,...,P − 1 Remark 3.6. The result is the same as the result for the commutative case as described in [2, Theorem 5], when we set T = 1. 14 A.SITARZANDJ.J.VENSELAAR Proof. Since r = −1, (1 + r)µ = 0 and so we only need to consider n in the analysis below. The multiplicity of the µ factor for a given j is 2j +1. When we consider the standard spin structure K = 0, we see that 2n ≡ 0 mod p. Consider first the case p odd. 2n ≡ 0 mod p means that either n is integer valued, and n = kp for some k ∈ Z, or n is half integer valued and 1 Z n = 2 p + kp for some k ∈ . Since the spectrum is symmetric, we only need to consider the ↑ part. In case n is integer valued, j must be half-integer valued. When j There is a second case which we can easily derive from the calculations above. If we look at equation (28), we see that in addition to the r = −1 case, the r = +1 case should easily be calculable. Classically, the L(p,p − 1) and L(p, 1) lens spaces are isometric, the orientation is just reversed, so one expects the spin structures to be the same. We will show that this is the case for the quantum lens spaces in the next section. 4. Unitary equivalences It is known [29] that two commutative lens spaces L(p, r) and L(p′, r′) are homeomorphic if and only if p = p′ and r′r ≡±1 mod p, or r′ ±r ≡ 0 mod p. It is also known that two 3-dimensional lens spaces are homeomorphic if and only if they are (Laplace) isospectral, see [24]. That these concepts are related for lens spaces can be intuitively under- stood by looking at diagrams as in Section 3.0.1. There we see that the diagram of L(p, r) is the same as the diagram of L(p, r′), r′ = ±r±1, up QUANTUMLENSSPACES 15 to rotation, mirroring and interchanging the stars and diamonds. For ex- ample, 3 ≡ −(−5)−1 mod 7, and we see in Figures 5 and 6 that they are the same if we rotate Figure 5 one quarter clockwise and flip the stars and diamonds. The L(5, 1) and L(5, −3) lens space diagrams of Figures 3 and 4 are very different however, and not related by mirroring and rotations by quarter-turns. In [24], using results on these type of lattices from [36], it is then shown that these type of lattice isomorphism induce an isomorphism of the algebra of smooth functions. In the noncommutative case when q 6= 1, the lens spaces Lq(p, r) and ′ ′ Lq(p, r ) are shown to be unitary equivalent when r = ±r , if we take care of the subtleties of equivariant representations. Theorem 4.1. When q ∈ (0, 1), the quantum lens spaces given by (Lq(p, r), ′ ′ HK,D,J) and (Lq(p, r ), HK ,D,J) are unitary equivalent if r ≡−r mod p. The unitary equivalence is implemented by the order order-two automor- ∗ phism σ(a) = a, σ(b) = −b of A(SUq(2)) and the action U on the Hilbert + − space given by U|jµn ↑i = −cjµn|j nµ ↓i and U|jµn ↓i = cjµn|j nµ ↑i, with again cjµn a complex number of norm 1. Proof. To show that the this map is a unitary equivalence, we first study the equivalence of Hilbert spaces. If v ∈ HK with K = 0 or K = p/2, we have (1 + r)µ + (1 − r)n ≡ 0 mod p or p/2 respectively. If we take r′ = −r, we see that (1 + r′)µ + (1 − r′)n = (1 − r)µ +(1+ r)n and we see that if we ′ interchange µ and n, v ∈ HK is mapped to a vector v ∈ HK in the K = 0 or K = p/2 subspace of the υ action for r′ = −r. ′ To define a compatible action of the algebra on HK, we see from Propo- sition 1.1 that we need to interchange b and −b∗. Now to show that this indeed gives a unitary equivalence on the algebra, we need to show that the U −1(Π′(−b))U = Π(b∗), and U −1(Π′(a))U = Π(a). However, since U interchanges the ↑ and ↓ part, Π′ is not the representation as given in Proposition 1.1, because then U −1Π′(b)U is not (ρ′, λ′)-equivariant as de- scribed in Section 1. We need to demand that Π′ is equivariant with re- spect to a different su(2)-action (ρ′′, λ′′), such that U −1ρ′′(g)U = ρ′(g) and U −1λ′′(g)U = λ′(g). Remark 4.2. For the other type of equivalences, i.e. the r → r−1 case, we do not know if they give rise to unitary equivalences in the q 6= 1 case. For isomorphisms of Lq(p, r) coming from the automorphisms of A(SUq(2)), as classified by (23), we can show that they do not give rise to isomorphisms −1 between Lq(p, r) and Lq(p, r ). Take for example an element of the form ∗ l ∗m −1 a b b ∈ Lq(p, r), with r(l − m) − 1 ≡ 0 mod p, i.e. l − m ≡ r mod p. We have r−1(l − m) ≡ (r−1)2 6≡ 1 mod p if r−1 6≡ r. This means that if r−1 6≡ r, −1 the identity automorphism is not a map from Lq(p, r) to Lq(p, r ). Also, we have r−1(m − l) ≡ r−1 · (−r−1) 6≡ 1 mod p if r−1 6≡ −r, hence the ∗ −1 automorphism a → a, b → b is not a map from Lq(p, r) to Lq(p, r ). 16 A.SITARZANDJ.J.VENSELAAR Hence the automorphisms of A(SUq(2)) do not give homomorphisms from −1 −1 Lq(p, r) to Lq(p, r ) if r 6≡ ±r. The same argument holds for Lq(p, r) −1 and Lq(p, −r ). We do not know that there are isomorphisms of Lq(p, r) not compatible with automorphisms of A(SUq(2)). 5. Quantum lens spaces as Seifert fibrations In [5], it was shown that certain quantum lens spaces, namely the lens spaces Lq(p; 1,p) could be viewed as being Seifert fibrations, i.e. princi- pal U(1)-comodule algebras over the quantum teardrops WPq(1,p). More precisely, they showed the existence of such a fibration for the coordinate algebras of both quantum spaces. The Lq(p; 1,p) lens spaces do not fit into the framework described above, as we only study lens spaces were the coac- tion of the finite group action is free, which is not the case for lens spaces of the form Lq(p, r) with p, r not coprime. In this section, we show that the situation for the coordinate algebras O(Lq(p, r)) resemble that of the commutative q = 1 case: the algebras O(Lq(p, 1)) are principal U(1)-comodule algebras over the quantum 2-sphere 2 WP Sq ≃ q(1, 1), while for r 6= 1, O(Lq(p, r)) does not admit such a fibration over any quantum teardrop. On the other hand, the situation on the C∗-algebra level is quite different. Due to the isomorphism C(SUq(2)) ≃ C(SU0(2)) for 0 ≤ q < 1, at the level of C∗-algebras, as found by Woronowicz [34, Theorem A2.2], we can show that the algebra C(Lq(p, r)) is a U(1)-principal comodule algebra over the quantum teardrop WPq(1, r). 3 The teardrop orbifold WP(r1, r2) can be defined as the quotient of S := 2 2 2 1 {(z1, z2) ∈ C : |z1| + |z2| = 1} by the following action of S := {t ∈ C : |r|2 = 1}: r1 r2 t · (z1, z2) = (t z1,t z2). Of course, the teardrop WP(n,n) for any n > 0 is just the sphere S2, but the the quotients where r1 6= r2 are not manifolds anymore, but orbifolds (in fact, what is usually called a “bad” orbifold, meaning that it there doesn’t exist a finite covering by a simply connected manifold). The quantum teardrop can be defined similarly, as the subalgebra of SUq(2) invariant under a suitable action of U(1): t · (α, β) = (tr1 α, tr2 β), ∗ with α and β the generators of the SUq(2) C -algebra. The invariant subal- gebra is of course the algebra generated as a vector space by basis vectors of k l ∗ m the form α β (β ) , such that r1k + r2(l − m) = 0. In [5, Theorem 2.1], it was shown that this algebra is generated by elements a and b, which satisfy QUANTUMLENSSPACES 17 the relations: r2−1 a = a∗ bb∗ = q2r1r2 ar1 (1 − q2ma) m=0 Y r2 ab = q−2r2 ba b∗b = ar1 (1 − q−2ma). m=1 Y It is not hard to see that a = ββ∗ and b = αr2 (β∗)r1 satisfy these relations, and that these elements generate the invariant subalgebra. In [5] it was also ∗ shown that on the C -algebra level, the quantum teardrops WPq(1, r) and WPq(r1, r) are isomorphic. It is convenient to switch to coactions for this section. A continuous coaction ρ for a Hopf algebra H acting on a C∗-algebra A is a map ρ : A → A ⊗ H that has the following properties: • ρ is injective • ρ is a comodule structure: (1 ⊗ ∆) ◦ ρ = (ρ ⊗ 1) ◦ ρ, where ∆ is the coproduct of H. • Podle´scondition: ρ(A)(1 ⊗ H)= A⊗ H. This coaction can be used to define principal H-comodule algebras, which can be seen as a a generalization of the concept of a principal fiber bundle to noncommutative geometry [21],[6], [31]. Let A be a C∗-algebra, with a coaction ρ : A → A ⊗ H by a Hopf algebra H. Denote by B the coinvariant part of A, i.e. the part where ρ(h)a = a⊗1.This is a quantum principal fibration, or principal H-comodule algebra if: ′ ′ • The canonical map can : A ⊗B A → A ⊗ H : a ⊗ a 7→ aρ(a ) is a bijection. • The map B ⊗A → A : b⊗a 7→ ba splits as a left B-module and and a right H-comodule map. This is also called equivariant projectivity. In the present case with H = U(1) we will call them quantum Seifert fibra- tions for their obvious analogue to Seifert fibrations in commutative geom- etry. Because of the results of [10] and [5], a right H-comodule algebra A is principal if and only if there exists a strong connection, i.e. there exists a map ω : H → A ⊗ H such that: (29a) ω(1) = 1 ⊗ 1 (29b) µ ◦ ω = η ◦ ǫ (29c) (ω ⊗ id) ◦ ∆ = (id ⊗ ρ) ◦ ω (29d) (S ⊗ ω) ◦ ∆ = (σ ⊗ id) ◦ (ρ ⊗ id) ◦ ω, where S,η,ǫ, ∆ are the antipode, unit, counit and comultiplication of the Hopf algebra H, and σ : A ⊗ H → H ⊗ A is the flip. 18 A.SITARZANDJ.J.VENSELAAR A strong connection in the case where H = U(1) is particularly nice to work with, because of the following lemma: Lemma 5.1. An algebra A, with continuous coaction ρ : A → A⊗U(1), has ′ ′ a strong connection if there exists elements i ai ⊗ bi, and i bi ⊗ ai such ′ ′ −1 ′ ′ −1 that aibi = biai = 1, ρ(ai)= ai ⊗ u , ρ(ai)= ai ⊗ u , ρ(bi)= bi ⊗ u, and ρ(b′ )= b′ ⊗ u. P P P i i P The proof of this lemma is a slight generalization of a construction done in the proof of [5, Theorem 3.3]. Proof. We proceed by induction. Define a strong connection ω by: ω(1) = 1 ⊗ 1 n n−1 ω(u )= aiω(u )bi Xi −n ′ −n+1 ′ ω(u )= biω(u )ai, Xi for each n ≥ 1. Condition (29a) is immediate. For n = 1, we see that because ′ ′ aibi = biai =1= η(ǫ(u)), condition (29b) is satisfied. Conditions (29c) and (29d) are also obvious by the definition. PNow supposeP ω(un−1) satisfies conditions (29b)–(29d). Then n n−1 µ(ω(u )) = µ( aiω(u )bi) = 1, Xi because µ(ω(un−1)) = 1. We also see that n n−1 (id ⊗ ρ)(ω(u )) = (id ⊗ ρ) aiω(u )bi Xi n−1 n = aiω(u )bi ⊗ u Xi = ω(un) ⊗ un. The same argument also works for condition (29d), and the u−n case. Theorem 5.2. The coordinate algebra O(Lq(p, r)) is a quantum principal U(1)-fibration over the quantum teardrop O(WPq(1, r)) when 0 < q ≤ 1, if and only if r = 1. k l ∗ m Proof. For basis vector α β (β ) ∈ Lq(p, r), we have k + r(l − m)= np for some n ∈ Z. Define the coaction ρ to be ρ(αkβl(β∗)m) = αkβl(β∗)m ⊗ un, with n the number above. Clearly WPq(1, r) is the coinvariant subalgebra under this coaction. QUANTUMLENSSPACES 19 Starting with the identities (α∗α + q2β∗β)p = 1, and (αα∗ + β∗β)p = 1, we see that by working out the parenthesis, we get the identities p ∗p p ∗p−p1 p−p1 ∗p1 p1 (30a) α α + cp1 α α β β = 1. p =1 X1 p p ∗p ′ p−p1 ∗p−p1 ∗p1 p1 (30b) α α + cp1 α α β β = 1. pX1=1 ′ R where the cp1 , cp1 ∈ are calculated by working out the identity. We can then define a strong connection by setting ω(1) = 1 ⊗ 1 and defining it inductively as: p n ∗ p n−1 p −p1(p−p1) ∗ p−p1 ∗ p1 n−1 p−p1 p1 ω(u ) = (α ) ω(u )α + cp1 q (α ) (β ) ω(u )α β . pX1=1 If r = 1, we have ρ(αp−p1 βp1 ) = αp−p1 βp1 ⊗ u, so this clearly satisfies the conditions for i ai ⊗ bi stated above, and so defines a strong connection by Lemma 5.1 For r 6= 1, butP coprime to p, it is impossible to define a strong connection this way, since the (ββ∗)p part of the (30a) can never be split into two parts a1a2 such that ρ(a2) = a2 ⊗ u. In fact, we can show that the canonical map can : A ⊗B A → A ⊗ H, used in the definition of a quantum principal fibration, cannot be surjective if we restrict ourselves to coordinate algebra O(Lq(p, r)), with r 6= 1 and p, r coprime. To show this, we adapt the proof of [5, Theorem 3.2] to the case of lens spaces. A basis for O(Lq(p, r)) ⊗ O(Lq(p, r)) is given by ′ ′ ′ αkβlβ∗m ⊗ αk βl β∗m , where k, k′ ∈ Z and l, l′,m,m′ ∈ N, with the usual convention that for k < 0: αk := α∗−k, and k + r(l − m) ≡ 0 mod p, and k′ + r(l′ − m′) ≡ 0 mod p, or equivalently, k′ + r(l′ − m′) = np, for some n ∈ Z The image of can of these basis element is then of the form ′ ′ ′ αkαk βl+l β∗m+m ⊗ un, hence every element in the image of can must be a linear combination of these. We proceed to show that 1 ⊗ u cannot lie in this image. An easy calculation shows that in order to have 1 in the image, we must have k = −k′, otherwise we always end up with some factor of α or α∗ in the image. Similarly, we can assume that l+l′ = m+m′. We also need k′+r(l′−m′)= p. Because r is coprime to p, we see that k′ 6= 0. However, every finite sum adding up to 1 is always going to contain a term k = k′ = 0, because of the decomposition αα∗ + ββ∗ = 1 (or alternatively α∗α + q2β∗β = 1). Hence 1 ⊗ u does not lie in the image of can, and so can is not bijective, and O(Lq(p, r)) is not a quantum principal fibration over O(WPq(1, r)). 20 A.SITARZANDJ.J.VENSELAAR If we go through the assumptions of the theorem, and the proof of the only if part of the theorem, we can deduce the following: Theorem 5.3. If q = 0, the algebra O(Lq(p, r)) is a quantum principal fibration over O(WPq(1, r)). Proof. For q = 0, we have for n> 0 the map define by: ω(1) = 1 ⊗ 1 ω(un) = (α∗)pω(un)αp ω(u−n)= αpω(un−1)(α∗)p + βp(α∗)p(r−1)ω(u−n+1)(α∗)p(r−1)βp defines a strong connection. It is immediately obvious that for un, n> 0 this defines a strong connection, because α∗α = 1 for q = 0. For n< 0, observe that we can rewrite the identity (αα∗ + ββ∗)p = 1 to (αα∗)p + βp(β∗)p = 1, because α∗β = β∗α = 0 when q = 0. Also (αα∗)2 = αα∗ and (ββ∗)2 = ββ∗, this mean we can write: (αα∗)p = αα∗ = αα∗αα∗ = α2(α∗)2 + αββ∗α∗ = α3(α∗)3 + α2ββ∗(α∗)2 + α(ββ∗)2α∗, and so on. Becaue α∗α = 1, we have ββ∗ = βα∗αβ∗. This mean that we can write p−1 αα∗ = αp(α∗)p + αp−p1 βp(α∗)p(r+1)−p1 αp(r+1)−p1 (β∗)p(α∗)p1 , pX1=1 which means that p−1 αp(α∗)p + αp−p1 βp(α∗)p(r+1)−p1 αp(r+1)−p1 (β∗)p(α∗)p1 p =1 X1 +βp(α∗)p(r−1)(α∗)p(r−1)βp = 1, and we can easily find a decomposition into a degree −1 and a degree +1 part for the coaction. Thus it satisfies the conditions of Lemma 5.1, and defines a strong connection. Of course, because in the continuous case C(SUq(2)) ≃ C(SU0(2)) for 0 ≤ q < 1, by [34, Theorem A2.2], this leads to the corollary: Corollary 5.4. The quantum lens space C(Lq(p, r)) is quantum principal fiber bundle over quantum teardrop C(WPq(1, l)). 6. Towards real spectral triples on quantum teardrops The coaction for which the quantum teardrop is the coinvariant part can easily be translated into an action of a Hopf algebra, with which we can con- struct an equivariant real spectral triple on the quantum teardrop WPq(1, r), coming from the real spectral triple on the quantum lens space Lq(p, r). QUANTUMLENSSPACES 21 This allows us to calculate for example the Dirac spectrum of the quantum 2 teardrops, and compare them to the spectrum of the quantum sphere Sq . Intuitively, the quantum teardrop WPq(1, r) can be understood to be the limit p →∞ of the quantum lens space Lq(p, r). Of course, in the construction below does not demonstrate a spectral triple over a quantum teardrop as, in particular, it cannot be repeated in the q = 1 limit. This is because classical teardrops are not manifolds and therefore the Dirac operator and the spectral triple for them make sense only when considered over the covering space. It remains open whether the orbifold-type singularities disappear when one considers q-deformed objects as some studies suggest (see [5]). In contrast to [13,14] we cannot claim that the spectral triple over the quantum Lens space is projectable. We can define an action ⊲ of u(1) on Lq(p, r) (or indeed, SUq(2)) as: (31) g ⊲ αkβlβ∗m = (k + r(l − m))αkβlβ∗.m The algebra of the quantum teardrop is kernel of this action. The coaction corresponding to this action was shown to admit a strong connection on Lq(p, r), which in the light of [3] is equivalent to proving that the coaction is free. The action ρ on the Hilbert space is then seen to be, using the equivariance conditions: (32) ρ(g)|jµnii = (µ + n + r(µ − n)+ c)|jµnii, where c is a constant. Compatibility with the real structure means that Jρ(g)J −1 = ρ(Sg)∗ = −ρ(g)∗, which leads to c = 0. It follows from the proof of Proposition 2.4, that J maps the eigenspace Hk := {v ∈H : ρ(g)v = kv} to H−k, hence using irreducibility with respect to the Dirac operator and the algebra action, we see that H0 is the Hilbert space corresponding to a spin structure in commutative geometry. 1 N Because µ + n ≡ µ − n ≡ 2 mod , this leads to the conclusion that there are no equivariant real spectral triples on WPq(1, r) if r is even. The action ρ obviously commutes with the Dirac operator of SUq(2). Thus we have: Proposition 6.1. If r is odd, the data (WPq(1, r), H0,D,J) defines a almost real (odd) spectral triple on the quantum teardrop WPq(1, r). Note that even though teardrops are even-dimensional and so is the metric dimension of the Dirac operator - there is not Z2-grading. The Dirac spectrum of this spectral triple can easily be calculated: Proposition 6.2. The positive part of the Dirac spectrum of the quantum teardrop WPq(1, r), with the spectral triple defined above, is given by (33) 2k + 3 k ∈ Nwith multiplicity 2k when r = 1 λ = 2 D 3 N ((2k + 1)l + p + 2 k ∈ ,p = 0, 1,...,l − 1with multiplicity 2k when r = 2l + 1 22 A.SITARZANDJ.J.VENSELAAR 1 Proof. For each j ∈ {0, 2 , 1,..., }, we have to count the number of solutions to the equation (µ + n + r(µ − n)) = 0. If r = 1, we see that µ = 0. This mean that j must be integer valued. The multiplicity is just the allowed number of n, which is 2j + 2 for the ↑ part and 2j for the ↓ part. The 3 spectrum is thus {±( 2 + 2k), k ≥ 0}, with multiplicity 2k + 2. For r> 1, the computation is a bit more involved. We see that for r> 1, 1+r 1 n = 1−r µ, which, since r = 2l + 1 is odd, can be written a n = (−1+ l )µ. 1 The smallest µ that satisfies this and gives a valid n is given by µ = 2 l, so 1 3 the smallest positive eigenvalue of the Dirac operator is given by 2 2 l + 2 = 3 l + 2 where r = 2l + 1. The multiplicity of this eigenvalue is then 2. The 3 multiplicity jumps to 4 at j = 2 l, etc., which gives the spectrum stated in the proposition. It appears the spectrum of WPq(1, 1) thus constructed is not the same 2 WP 2 as that of the S , even though q(1, 1) ≃ Sq . This is in contrast to the θ-deformed case, where in [14, Section 7.1] a similar computation was done 2 3 for the spectrum for S as an U(1) invariant subspace of Sθ . References [1] Adam Rennie, Andrzej Sitarz, and Makoto Yamashita, Twisted cyclic cohomology and modular Fredholm modules, SIGMA Symmetry Integrability Geom. Methods Appl. 9 (2013), Paper 51, 15. [2] Christian B¨ar, The Dirac operator on homogeneous spaces and its spectrum on 3- dimensional lens spaces, Archiv der Mathematik 59 (1992), no. 1, 65–79. [3] Paul F. Baum, Kenny De Commer, and Piotr M. Hajac, Free actions of compact quantum group on unital c*-algebras (2013), available at 1304.2812. [4] Tomasz Brzezi´nski, On the smoothness of the noncommutative pillow and quantum teardrops (2013), available at 1311.4758. [5] Tomasz Brzezi´nski and Simon A. Fairfax, Quantum teardrops, Comm. Math. Phys. 316 (2012), no. 1, 151–170. MR2989456 [6] Tomasz Brzezi´nski and Piotr M. Hajac, The Chern-Galois character, C. R. Math. Acad. Sci. Paris 338 (2004), no. 2, 113–116. MR2038278 [7] Tomasz Brzezi´nski and Andrzej Sitarz, Spectral triples on quantum teardrops and quantum pillows. In progress. [8] Alain Connes, Noncommutative geometry and reality, Journal of Mathematical Physics 36 (1995), no. 11, 6194–6231. MR1355905 (96g:58014) [9] , Gravity coupled with matter and the foundation of non-commutative ge- ometry, Communications in Mathematical Physics 182 (1996), no. 1, 155–176. MR1441908 (98f:58024) [10] Ludwik Dabrowski, Harald Grosse, and Piotr M. Hajac, Strong connections and ‘ Chern-Connes pairing in the Hopf-Galois theory, Comm. Math. Phys. 220 (2001), no. 2, 301–331. MR1844628 (2002g:58007) [11] Ludwik D¸abrowski, Giovanni Landi, Andrzej Sitarz, Walter van Suijlekom, and Joseph C. V´arilly, The Dirac operator on SUq(2), Communications in Mathemati- cal Physics 259 (2005), no. 3, 729–759. MR2174423 (2006h:58034) [12] Ludwik Dabrowski and Andrzej Sitarz, Dirac operator on the standard Podle´squan- ‘ tum sphere, Noncommutative geometry and quantum groups (Warsaw, 2001), 2003, pp. 49–58. MR2024421 (2005g:58056) QUANTUMLENSSPACES 23 [13] , Noncommutative circle bundles and new Dirac operators, Comm. Math. Phys. 318 (2013), no. 1, 111–130. MR3017065 [14] Ludwik Dabrowski, Andrzej Sitarz, and Alessandro Zucca, Dirac operator on non- ‘ commutative principal circle bundles (2013), available at 1305.6185. [15] Walther Dyck, On the “Analysis situs” of three-dimensional spaces, Report of the Fifty-fourth Meeting of the British Association for the Advancement of Science: Held at Montreal in August and September 1884, 1885. [16] David A. Ellwood, A new characterisation of principal actions, J. Funct. Anal. 173 (2000), no. 1, 49–60. MR1760277 (2001c:46126) [17] Annick Franc, Spin structures and Killing spinors on lens spaces, J. Geom. Phys. 4 (1987), no. 3, 277–287. MR957015 (90e:57047) [18] Jos´eM. Gracia-Bond´ıa, Joseph C. V´arilly, and H´ector Figueroa, Elements of noncom- mutative geometry, Birkh¨auser Advanced Texts: Basel Textbooks, Birkh¨auser Boston Inc., Boston, MA, 2001. MR1789831 (2001h:58038) [19] Tom Hadfield and Ulrich Kr¨ahmer, Twisted homology of quantum SL(2), K-Theory 34 (2005), no. 4, 327–360. MR2242563 (2007j:58009) [20] , Twisted homology of quantum SL(2) - Part II, Journal of K-Theory 6 (2010), no. 1, 69–98. MR2672153 (2011i:58008) [21] Piotr M. Hajac, Strong connections on quantum principal bundles, Comm. Math. Phys. 182 (1996), no. 3, 579–617. MR1461943 (98e:58022) [22] Jeong Hee Hong and Wojciech Szyma´nski, Quantum spheres and projective spaces as graph algebras, Comm. Math. Phys. 232 (2002), no. 1, 157–188. MR1942860 (2003i:46080) [23] , Quantum lens spaces and graph algebras, Pacific Journal of Mathematics 211 (2003), no. 2, 249–263. MR2015735 (2004g:46074) [24] Akira Ikeda and Yoshihiko Yamamoto, On the spectra of 3-dimensional lens spaces, Osaka J. Math. 16 (1979), no. 2, 447–469. MR539600 (80e:58042) [25] Bruno Iochum, Thomas Sch¨ucker, and Christoph Stephan, On a classification of irre- ducible almost commutative geometries, J. Math. Phys. 45 (2004), no. 12, 5003–5041. MR2105233 (2005j:58038) [26] Anatoli Klimyk and Konrad Schm¨udgen, Quantum groups and their representa- tions, Texts and Monographs in Physics, Springer-Verlag, Berlin, 1997. MR1492989 (99f:17017) [27] Tetsuya Masuda, Yoshiomi Nakagami, and Junsei Watanabe, Noncommutative differ- ential geometry on the quantum SU(2). I. An algebraic viewpoint, K-Theory 4 (1990), no. 2, 157–180. MR1081658 (91j:58016) [28] Piotr Olczykowski and Andrzej Sitarz, Real spectral triples over noncommutative Bieberbach manifolds, J. Geom. Phys. 73 (2013), 91–103. MR3090104 [29] Kurt Reidemeister, Homotopieringe und Linsenr¨aume, Abh. Math. Sem. Univ. Ham- burg 11 (1935), no. 1, 102–109. MR3069647 [30] Adam Rennie and Roger Senior, The resolvent cocycle in twisted cyclic cohomology and a local index formula for the Podles sphere (2011), available at 1111.5862. [31] Hans-J¨urgen Schneider, Principal homogeneous spaces for arbitrary Hopf algebras, Israel J. Math. 72 (1990), no. 1-2, 167–195. Hopf algebras. MR1098988 (92a:16047) [32] Walter van Suijlekom, Ludwik D¸abrowski, Giovanni Landi, Andrzej Sitarz, and Joseph C. V´arilly, The local index formula for SUq (2), K-Theory 35 (2005), no. 3-4, 375–394 (2006). MR2240238 (2007f:58035) [33] Stanis law L. Woronowicz, Compact matrix pseudogroups, Communications in Math- ematical Physics 111 (1987), no. 4, 613–665. MR901157 (88m:46079) [34] , Twisted SU(2) group. An example of a noncommutative differential calcu- lus, Kyoto University. Research Institute for Mathematical Sciences. Publications 23 (1987), no. 1, 117–181. MR890482 (88h:46130) 24 A.SITARZANDJ.J.VENSELAAR [35] , Compact quantum groups, Sym´etries quantiques (Les Houches, 1995), 1998, pp. 845–884. MR1616348 (99m:46164) [36] Yoshihiko Yamamoto, On the number of lattice points in the square x + y ≤ u with a certain congruence condition, Osaka J. Math. 17 (1980), no. 1, 9–21. MR558314 (81c:10062) Institute of Physics, Jagiellonian University, Reymonta 4, 30-059 Krak´ow, Poland, Mathematics Department, Mail Code 253-37, Caltech, 1200 E. California Blvd. Pasadena, CA 91125, USA
Arxiv:Submit/0873263 [Math.QA] 18 Dec 2013 Aie(O-Unu)Cs,Teln Spaces Lens the Case, (Non-Quantum) Tative Unu Esspaces Lens Quantum Esuyuiayeuvlne Fsc Unu Esspaces
![Arxiv:Submit/0873263 [Math.QA] 18 Dec 2013 Aie(O-Unu)Cs,Teln Spaces Lens the Case, (Non-Quantum) Tative Unu Esspaces Lens Quantum Esuyuiayeuvlne Fsc Unu Esspaces](http://data.docslib.org/img/89f6cef669d60e8eaf35dd5d9004b870-1.webp)