• Abstract and Figures
• Public Full-text

Preparing Dicke states in a spin ensemble using phase estimation Yang Wang QuTech, Delft University of Technology, P.O. Box 5046, 2600 GA Delft, The Netherlands Barbara M. Terhal QuTech, Delft University of Technology, P.O. Box 5046, 2600 GA Delft, The Netherlands and JARA Institute for Quantum Information, Forschungszentrum Juelich, D-52425 Juelich, Germany (Dated: September 13, 2021) We present a Dicke state preparation scheme which uses global control of N spin qubits: our scheme is based on the standard phase estimation algorithm, which estimates the eigenvalue of a unitary operator. The scheme prepares a Dicke state non-deterministically by collectively coupling the spins to an ancilla qubit via a ZZ-interaction, using dlog2 Ne + 1 ancilla qubit measurements. The preparation of such Dicke states can be useful if the spins in the ensemble are used for magnetic sensing: we discuss a possible realization using an ensemble of electronic spins located at diamond nitrogen-vacancy centers coupled to a single superconducting flux qubit. We also analyze the effect of noise and limitations in our scheme. I. INTRODUCTION preparing Dicke states which are based on the full ad- dressability of the qubits [25, 26]. To address this issue, A promising application of the emerging quantum tech- some work has been dedicated to schemes which require nology is quantum-enhanced sensing, sometimes referred only a global control of the spin ensemble, such as using to as quantum metrology [1,2]. Using entangled states, steady state evolution [27], repeated energy transfer [7], continuous weak measurements [28], and the use of ge- one can, in principle, improve the measurementp sensi- tivity from the standard quantum limit (1= N) to the ometric phase gates [29]. Unfortunately, these methods Heisenberg limit (1=N)[1,3,4], where N is the number are still demanding currently when N is large, as they of- of probes or repetitions. However, preserving this quan- ten need complicated measurement-based feedback, high tum advantage is difficult in the presence of decoherence fidelity control and long preparation times. For example, [5]. For instance, a single-qubit Pauli Z-error can to- the optimized scheme in Ref. [7] uses O(N) rounds of ini- tally dephase a N-qubit GHZ-state, which would obtain tialization and evolution of an ancilla qubit. Our goal is Heisenberg-limited sensitivity in the noiseless case [6]. to improve the scaling with N so that one could possibly N-qubit Dicke states form a class of entangled states handle a larger error rate on the ancilla qubit. which are interesting for metrology [6{10]. Compared to In this paper, we present a Dicke state preparation other states used in quantum sensing, Dicke states have scheme that uses standard phase estimation [30], which been argued to be more robust to various noise sources prepares an eigenstate of a unitary operator by estimat- such as spin dephasing, spin damping, and spin number ing its eigenvalue. This algorithm is based on executing fluctuations [9]. Recent work has demonstrated a scheme projective measurements on the spin ensemble using an to use Dicke states for detecting the magnetic field in- ancilla qubit, and it will prepare a random Dicke state. duced by a single spin [7]. Another distinctive feature The scheme requires a ZZ-coupling between each spin of the use of Dicke states is that the optimal sensitivity in the ensemble to a single ancilla qubit. In Section III can be obtained through only global control on the set we detail how this coupling could be realized between an of spins [9, 10]. This is relevant for realizing practical ensemble of NV electronic spins and a superconducting quantum sensing using entangled states, as precise indi- flux qubit as ancilla. vidual spin qubit control can be difficult. Furthermore, Our scheme is efficient with respect to the number of arXiv:2104.14310v4 [quant-ph] 10 Sep 2021 superpositions of Dicke states can be used for quantum operations. It uses only dlog2 Ne + 1 rounds of phase error correction [11, 12]. estimation for preparing a random N-spin Dicke state. Dicke state preparation has been experimentally re- Each round of phase estimation measures a global opera- alized using photons [13, 14] and trapped-ion qubits tor of the spins, i.e., it applies an ancilla-qubit controlled 1 PN [15, 16], and there also exist many theoretical prepara- global Jz = 2 i=1 Zi rotation followed by ancilla qubit tion proposals suitable for a few qubits, see Ref. [17{ readout. The total time for performing the controlled 20] for example. However, it remains a challenge for rotations is upper-bounded by a constant and the prepa- large spin ensembles like N > O(100) diamond Nitrogen- ration timep thus scales as O(log2 N). With a probabil- Vacancy centers (negatively-charged NV [21]), each host- ity ∼ O(1= N), the prepared Dicke state would obtain ing an electronic spin S = 1. Since these NV center spins Heisenberg-limited sensitivity using only global control. are rather isolated from each other, it is costly to per- Besides the efficiency, our scheme also has some noise- form entangling gates between the electronic spins [22{ resilience: phase estimation can be realized with inte- 24]. This limitation excludes quantum algorithms for grated dynamical decoupling, which provides robustness 2 to the dephasing of the ancilla qubit as well as the de- The variance of θ can be calculated by the error propa- phasing of the spins in the ensemble. Furthermore, by gation formula repeating the projective measurements and performing a 2 simple majority vote, the effects of ancilla qubit decay 2 (∆M(θ)) (∆θ) = 2 ; (2) and flipped measurements (due to ancilla qubit dephas- j@θ hM(θ)ij ing or imperfect measurement) can be mitigated. This paper is organized as follows. In Sec.IA, we where the expectation value is with respect to the ini- briefly review Dicke states and Heisenberg-limited sens- tial state jN; mzi and M(θ) is the Heisenberg-evolved ing. In Sec. II we present the idea of using phase esti- operator. If we were to measure M = αJx + βJz, then mation to prepare Dicke states. In Sec. III we discuss hM(θ)i = hJzi (β cos(θ) + α sin(θ)) the Hamiltonian and a possible experimental setup with (3) multiple NV centers coupled to a flux qubit. In Sec.IV ≈ mz(β + θα); we numerically consider the performance of the scheme given the dominant noise sources. Finally, we discuss the for small θ (note that hN; mzj Jx jN; mzi = 0). We mea- results in Sec.V. sure Jx by choosing β = 0; α = 1, its expectation value hJx(θ)i has an optimal dependence on θ when mz is large. 2 However, the variance (∆Jx(θ)) will be large in a rotated A. Dicke states Dicke state, precluding any Heisenberg gains. 2 The proposal is instead to measure M = Jz , so that For simplicity, we assume even spin number N the variance is given by (see details in Ref. [10]): throughout this paper (odd spin-number N can be 2 2 2 2 2 treated similarly). The N-spin (or qubit) Dicke (∆θ) = [(∆Jx ) f(θ) + 4hJx i − 3hJy i state jN; m i with m 2 {− N ;:::; N g is a uniform, 2 2 2 2 2 2 z z 2 2 − 2hJz i × (1 + hJx i) + 6hJzJx Jzi] = [4(hJx i − hJz i) ] permutation-symmetric, superposition of N-bit strings (4) jxi where all bit-strings have N=2 + mz spins in j0i, 2 2 i.e., their Hamming weight is N=2 − mz. For example, (∆Jz ) 2 1 with f(θ) = 2 2 2 + tan (θ). The minimal vari- jN = 4; m = 0i = p (j0011i+j0101i+j0110i+j1001i+ (∆Jx) tan (θ) z 6 ance is obtained when tan2(θ) = p(∆J 2)2=(∆J 2)2. For j1010i + j1100i). A Dicke state jN; mzi is an eigenstate z x of the collective spin operator Dicke state jN; mzi the minimal variance (obtained at θ ≈ 0) is N 1 X Jz = Zi; (1) 2 4 2 2 2 2mz + 2 64mz − 16mz i=1 (∆θmin) = 2 2 + 2 2 2 : N + 2N − 12mz (N + 2N − 12mz) with eigenvalue mz. Here Zi is the Pauli Z operator on (5) 1 PN the spin labeled i. In addition, we have Jx = 2 i=1 Xi 1 PN Note that the sensitivity can surpass the standard quan- and Jy = 2 i=1 Yi. p To use such states for metrology, one imagines that tum limit when mz ∼ O( N) and is Heisenberg-limited 2 the prepared quantum state is transformed by e−iθJy when mz ∼ O(1). In addition, when mz = 0, (∆θmin) = 2 and the goal is to estimate the rotation angle θ which N(N+2) saturates the quantum Cram´er-Raobound [6]. 2 is assumed to be small. A standard metrological method The expectation value hJz i can in principle be obtained (for NV centers, limited by T2 and optical measurement by measuring Jz, squaring its outcome and gathering suf- accuracy) is Ramsey spectroscopy [31] using a single ficient statistics by repeating the measurements. We are qubit state repeatedly (or, equivalently, using a prod- thus especially interested in Dicke states close to jN; 0i, uct state of multiple qubits). In this context, the Ram- i.e., jN; mzi with mz ∼ O(1). Other than this motiva- sey method corresponds to preparing a simple prod- tion, we do not focus on aspects of using a (noisy) Dicke i π J uct state e 2 y j00 ::: 0i and letting it thus evolve to state for metrology in this paper.

Load more

  14

Copy Link