Segre variety, conifold, Hopf fibration, and separable multi-qubit states Hoshang Heydari Institute of Quantum Science, Nihon University, 1-8, Kanda-Surugadai, Chiyoda-ku, Tokyo 101- 8308, Japan Abstract We establish relations between Segre variety, conifold, Hopf fibration, and separable sets of pure two-qubit states. Moreover, we investigate the geometry and topology of separable sets of pure multi-qubit states based on a complex multi-projective Segre variety and higher order Hopf fibration. 1 Introduction Quantum entanglement[1, 2] is one of the most interesting features of quan- tum theory. In quantum mechanics, the space of pure states in is an N + 1- dimensional Hilbert space can be described by the complex projective space CPN . For bipartite, pure states, the entanglement of formation can be written in terms of concurrence [3]. The connection between concurrence and geometry is found in a map called Segre embedding, see D. C. Brody and L. P. Hughston [4]. They illustrate this map for a pair of qubits, and point. The Segre embed- ding has also been discussed in [5]. There is also another geometrical description to describe pure state called Hopf fibration. The relation between Hopf fibration and single qubit and two-qubit states is discussed by R. Mosseri and R. Dan- doloff [6]. They have shown that S2 base space of a suitably oriented S3 Hopf fibration is nothing but the Bloch sphere, while the circular fibres represent the qubit overall phase degree of freedom. For two-qubit states, the Hilbert space is a seven-dimensional sphere S7, which also allows for a second Hopf fibration which is entanglement sensitive, with S3 fibres and a S4 base. Moreover, a arXiv:quant-ph/0506043v1 6 Jun 2005 generalization of Hopf fibration to three-qubit state has been presented in Ref. [7], where the Hilbert space of the three-qubit state is the fifteen-dimensional sphere S15, which allows for the third Hopf fibration with S8 as base and S7 as fiber. In this paper we will describe the Segre variety, which is a quadric space in algebraic geometry [8, 9, 10, 11, 12], by giving a complete and explicit formula for it. We will compare the Segre variety with the concurrence of pure, two-qubit states. The vanishing of the concurrence of a pure two-qubit state coincides with the Segre variety. Moreover, we will establish relations between Segre variety, conifold and Hopf fibration. In algebraic geometry, a conifold is a generalization of the notion of a manifold. Unlike manifolds, a conifold can con- tain conical singularities, i.e., points whose neighborhood look like a cone with a certain base. The base is usually a five-dimensional manifold. Conifold are very important in string theory, i.e., in the process of compactification of Calabi-Yau manifolds. A Calabi-Yau manifold is a compact K¨ahler manifold with a vanish- ing first Chern class. A Calabi-Yau manifold can also be defined as a compact Ricci-flat K¨ahler manifold. Finally, we will discuss the geometry and topology of pure multi-qubit states based on some mathematical tools from algebraic ge- ometry and algebraic topology, namely the multi-projective Segre variety and higher-order Hopf fibration. Let us start by denoting a general, pure, composite p quantum system with m subsystems = m(N1,N2,...,Nm)= 1 2 m, N1 Q NQ2 Nm Q Q ···Q consisting of a pure state Ψ = αi ,i ,...,i i1,i2,...,im | i i1=1 i2=1 ··· im=1 1 2 m | i and corresponding Hilbert space as Q = Q Q Q , where the P HP H 1 ⊗P H 2 ⊗···⊗H m dimension of the jth Hilbert space is given by Nj = dim( Qj ). We are going to use this notation throughout this paper, i.e., we denoteH a pure two-qubit p states by (2, 2). Next, let ρQ denotes a density operator acting on Q. Q2 H The density operator ρQ is said to be fully separable, which we will denote by sep ρQ , with respect to the Hilbert space decomposition, if it can be written as sep N m k N ρQ = k=1 pk j=1 ρQj , k=1 pk = 1 for some positive integer N, where pk k are positive real numbers and ρQj denotes a density operator on Hilbert space Pp N P Qj . If ρQ represents a pure state, then the quantum system is fully separable H p sep m if ρQ can be written as ρQ = j=1 ρQj , where ρQj is a density operator on Q . If a state is not separable, then it is said to be entangled state. H j N 2 Complex projective variety In this section we will review basic definition of complex projective variety. Let n f1,f2,...,fq be continuous functions K K, where K is field of real R{ or complex} number C. Then we define real−→ (complex) space as the set of simultaneous zeroes of the functions n K(f1,f2,...,fq)= (z1,z2,...,zn) K : fi(z1,z2,...,zn)=0 1 i q . V { ∈ ∀ ≤ ≤ (1)} These real (complex) spaces become a topological spaces by giving them the n induced topology from K . Now, if all fi are polynomial functions in coordinate functions, then the real (complex) space is called a real (complex) affine variety. A complex projective space CPn which is defined to be the set of lines through the origin in Cn+1, that is, CPn = (Cn+1 0)/ , where is an equivalence relation define by (x ,...,x ) (y ,...,y− ∼) λ ∼ C 0 such that 1 n+1 ∼ 1 n+1 ⇔ ∃ ∈ − λxi = yi 0 i n. For n = 1 we have a one dimensional complex manifold CP1 which∀ is≤ very≤ important one, since as a real manifold it is homeomorphic to the 2-sphere S2. Moreover every complex compact manifold can be embedded in some CPn. In particular, we can embed a product of two projective spaces into a third one. Let f1,f2,...,fq be a set of homogeneous polynomials in the { } n+1 coordinates α1, α2,...,αn+1 of C . Then the projective variety is defined to be the subset{ } n (f1,f2,...,fq)= [α1,...,αn+1] CP : fi(α1,...,αn+1)=0 1 i q . V { ∈ ∀ ≤ ≤ (2)} n+1 We can view the complex affine variety C(f1,f2,...,fq) C as complex V ⊂ n cone over projective variety (f1,f2,...,fq). We can also view CP as a quo- tient of the unit 2n + 1 sphereV in Cn+1 under the action of U(1) = S1, that is CPn = S2n+1/U(1) = S2n+1/S1, since every line in Cn+1 intersects the unit sphere in a circle. 3 Hopf fibration and two- and three-qubit states p For a pure one-qubit state 1(2) with Ψ = α1 1 + α2 2 , where α1, α2 C, and α 2 + α 2 = 1, we canQ parameterize| i this state| i as | i ∈ | 1| | 2| ϑ ϕ χ α1 cos( 2 )exp(i( 2 + 2 )) = ϑ ϕ χ (3) α2 cos( )exp(i( )) 2 2 − 2 where ϑ [0, π], ϕ [0, 2π] and χ [0, 2π]. The Hilbert space Q of a single qubit is∈ the unit 3-dimensional∈ sphere∈ S3 R4 = C2. But sinceH quantum mechanics is U(1) projective, the projective⊂ Hilbert space is defined up to a phase exp (iϕ), so we have CP1 = S3/U(1) = S3/S1 = S2. Now, the first Hopf S1 map, S3 / S2 as an S1 fibration over a base space S2. For a pure two- qubit state p(2, 2) with Ψ = α 1, 1 +α 1, 2 +α 2, 1 +α 2, 2 , where Q2 | i 1,1| i 1,2| i 2,1| i 2,2| i α , α , α , α C and 2 α 2 = 1. The normalization condition iden- 1,1 1,2 2,1 2,2 ∈ k,l | k,l| tifies the Hilbert space to be the seven dimensional sphere S7 R8 = C4 Q P and the projective HilbertH space to be CP3 = S7/U(1). Thus we can⊂ parame- S3 terized the sphere S7 as a S3 fiber over S4, that is S7 / S4 which is called the Hopf second fibration. This Hopf map is entanglement sensitive and the separable states satisfy α1,1α2,2 = α1,2α2,1, see Ref. [6]. 4 Segre variety for a general bipartite state and concurrence p N1−1 For given general pure bipartite state 2(N1,N2) we want make CP CPN2−1 into a projective variety by itsQ Segre embedding which we construct× as follows. Let (α1, α2,...,αN1 ) and (α1, α2,...,αN2 ) be two points defined on CPN1−1 and CPN2−1, respectively, then the Segre map : CPN1−1 CPN2−1 CPN1N2−1 (4) SN1,N2 × −→ ((α ,...,α ), (α ,...,α )) (α ,...,α ,...,α ,...,α ) 1 N1 1 N2 7−→ 1,1 1,N1 N1,1 N1,N2 is well defined. Next, let αi,j be the homogeneous coordinate function on CPN1N2−1. Then the image of the Segre embedding is an intersection of a family of quadric hypersurfaces in CPN1N2−1, that is Im( ) = < α α α α >= (α α α α ) . (5) SN1,N2 i,k j,l − i,l j,k V i,k j,l − i,l j,k This quadric space is the space of separable states and it coincides with the p definition of general concurrence ( 2(N1,N2)) of a pure bipartite state [13, 14] because C Q 1 2 N1 N2 ( p(N ,N )) = α α α α 2 , (6) C Q2 1 2 N | i,k j,l − i,l j,k| j,i=1 X l,kX=1 where is a somewhat arbitrary normalization constant.