M ay 2 00 2 A simple way to detect the state transition caused by the nondiagonal abelian Berry phase

In a nondegenerate system, the abelian Berry’s phase will never cause transitions among the Hamiltonian’s eigenstate. However, in a degenerate system, it is well known that the state transition can be caused by the non-abelian Berry phase. Actually, in a such a system, the phase factor is not always nonabelian. Even in the case that the phase is a nondiagonal matrix, it still can be an abelian phase and can cause the state transitions. We then propose a simple scheme to detect such an effect of the Berry phase to the degenerate states by using the optical entangled pair state. In this scheme, we need not control any external parameters during the quantum state evolution. ∗email: [email protected] †email: [email protected] ‡email: [email protected] §email: [email protected] ∗∗email: [email protected] 1 Geometric phase [1–3] plays an important role in quantum interferometry and many other disciplines. Many observable quantum phyonemena are related to the geometric phase shift. The geometric phase to the degenerate states is qualitatively different from the one to the nondegenerate states, because it can cause the detectable effects which never happen to the nondegenerate states. For example, to the degenerate states one can obtain the nonabelian geometric phase [4–8]. Such a non-abelian Berry phase can cause transitions among the orthogonal states belonged the same eigenvalue of the Hamiltonian therefore could be useful in quantum computation [9]. Due to the non-abelian property, if we alternate two cyclic adiabatic operations, the states evolve differently. In this paper we will show that, to the degenerate states, there are abelian phase factors which are nondiagonal matrices. This is to say, even the Abelian Berry phase is qualitatively different from the one to the nondegenerate states, because it can cause the transitions among the eigenstates after a cyclic adiabatic evolution. We also give a feasible scheme to detect this type of nontrivil effect caused by the Berry phase to the degenerate states. The experimental testing of geometric phase is an interesting and important topic. For a two level system, geometric phase is equal to half of the solid angle subtended by the area in the Bloch sphere enclosed by the closed evolution loop of the eigenstate. Most of the times, after certain time evolution path, there are both dynamic and geometric phase shift, this causes extra treatment to prove the effects of the geometric phase. Recently, applications of the geometric phase is extended the geometric quantum computation [9,11–16], i.e., making the quantum gate through the geometric phase shift. In particular, several of them are based on the idea of using the Berry phase [9,14,15] shift to the degenerate states. Different from the case to the nondegenerate states, to a degenerate system, the Berry phase after a cyclic adiabatic evolution may cause the bit flip or the conditional bit flip between two states. This is rather significant because of its obvious fault tolerance property. For the nonabelian holonomic quantum computation, the computational space C is always an eigen space(spanned by a number of degenerate eigen states). The first clear experimental proposal to produce the controllable Berry phase ship for the degenerate 2 system was raised by Duan et al [15]. The proposal suggests using trapped ions for the test. However, in this scheme, we will have to exactly control the external parameters so that the eigenvalues of the instantaneous Hamiltonian keeps to be zero. In this paper, we propose a simple scheme to test the states transition caused by Berry phase to the degenerate. Our scheme could be more robust because it requires little external parameters control in the whole process. Lets first see what is the non-trivil effect caused by the arbitrary Berry phase to the degenerate states. Suppose orthonormal states {|ψ′ i〉} spans a degenrate space of Hamiltonian H , i.e., all of the states are the eigenstates of H with the same eigenvalue. If we change the Hamiltonian adiabatically and cyclically, there could be a non-trivil time evolution of the states in the degenerate subspace. There is no dynamical contribution to the time evolution because the dynamical phase is all zero. This time evlution operator in general can be a matrix with nonzero off diagonal elements. Since there is no dynamical contribution, we name this time evolution operator as the geometric phase factor. Suppose after certain cyclic adiabatic process the Berry phase factor is U , i.e. {|ψ′ i(t)|〉} = U{ψ′ i(0)〉}. (1) Since U is unitary, in principle we can always diagonalize it by another unitary matrix S in the same degenerate space so that S†US is diagonalized. This is to say, in the basis {|ψi〉} = S{|ψ′ i〉}, the time evolution operator is diagonal. Any state |ψi〉 = S|ψ′ i〉 must be also an eigenstate of H due to the degeneracy property. And, all states of {ψi} have the same eigenvalues with that of {ψ′ i}. Consequently, if any adiabatical process causes a diagonal time evolution(a phase shift) in certain basis {|ψi〉}, the phase factor must be a nondiagonal matrix in another basis {|ψ′ i}. This type of time evolution is in general regarded as nonabelian and may cause transitions among the eigenstates, even though all states keep on being the eigenstates of of the instantaneous Hamiltonian during the whole process. This transition never happens by the same operation to the nondegenerate eigenstates. However, in certain case, the evolution operator can be abelian although it is not diagonal. And as we 3 will show in the following, such a nontrivil abelian phase can also cause state transitions. Consider the following entangled state for the spin half system |Φ±〉 = 1 √ 2 (|0A1B〉 ± |1A0B〉), (2) subscripts A and B represent for the qubit A or qubit B respectively, 0 is spin down and 1 is spin up. Denoting |ψ1〉 = |0A1B〉 and |ψ2 = |1A0B〉. Obviously state |ψ1〉 and |ψ2〉 are dengenrate in a Z-direction external field. Now we consider how to produce an abelian and nondiagonal matrix phase factor to the two level degenerate states. We can first spatially separate spin A from spin B. Obviously, we can also acquire a non-trivil Berry phase in the basis |Φ±〉 which is linearly superposed by state |ψ1〉 and |ψ2〉. We will show that we can have a non-trivil effect if it is observed in the |Φ±〉 basis through the abelian phase in the basis |ψ1,2〉. Consider another set of states |Φ±〉, each of which is a linear superposed state of |ψ1〉 and |ψ2〉. We can first spatially separate qubit A and B. Then rotate the local field of qubit B so that qubit B evolve a closed loop. After the evolution, the Berry phase for 0A and 1A are opposite and the value is propotional to solid angle substended by the evolution loop at the degenerate point of the Bloch sphere for spin B. Explicitly, we have Γ0,1 = ±Ω/2 (3) where Γ0,1 is the Berry phase shift for state |0B〉 or |1B〉 and Ω is the value of solid angle substended by the evoultion loop. For example, if we rotate the field adiabatically around the Z ′-axis for a loop. Then we have |Γ0,1| = π cos θ, θ is the angle between Z and Z ′. Without loss of generality, we use Γ to denote the Berry phase shift for state |0B〉. At time T when the evolution loop is completed we have |Φ(T )〉 = 1 √ 2 (e|ψ1〉 ± e|ψ2〉). (4) Note that here no dynamical phase is involved because state |ψ1〉 and state |ψ2〉 are degenerate in the whole process. Although the phase is diagonal in the |ψ1,2〉 basis, it is not 4 diagonal in the |Φ±〉 basis. In such a non-trivil basis, the phase factor is given as