Arxiv:2104.11258V1 [Quant-Ph] 22 Apr 2021 POVM) [17]

Quantum tomography of three-qubit generalized Werner states

Artur Czerwinski∗ Institute of Physics, Faculty of Physics, Astronomy and Informatics Nicolaus Copernicus University, Grudziadzka 5, 87–100 Torun, Poland

In this article, we introduce a framework for state tomography of three-qubit generalized Werner states. The framework utilizes the single-qubit SIC-POVM and involves the Poisson noise. The accuracy of state estimation is quantiﬁed and presented on graphs. As a special case, we compare the eﬃciency of the framework for the pure three-qubit generalized Werner state and the W state. The reconstructed states are presented graphically and discussed. Keywords: generalized Werner states, three-qubit entanglement, quantum state tomography, SIC-POVM

I. INTRODUCTION II. FRAMEWORK FOR STATE TOMOGRAPHY

Quantum state tomography (QST) plays a crucial role A. Three-qubit generalized Werner states in many quantum information protocols since it pro- vides a variety of techniques to determine a mathematical In 1989, R. Werner introduced one-parameter class of representation of a quantum system [1]. In particular, mixed quantum states which feature non-classical corre- photonic tomography has been developed to character- lations [18]. In particular, two-qubit Werner states have ize quantum state of photons, including polarization [2]. been implemented for various important advancements Since noise and errors are inherent to any measurement, in quantum information [19]. In this work, we analyze one needs to implement statistical methods which pro- three-qubit generalized Werner states [20, 21]: duce reliable estimates of the actual states [3, 4]. Estima- tion methods can be compared in terms of their efficiency 3q 1 − η in state reconstruction [5]. ρW (η) = η |GHZihGHZ| + I8, (1) Entangled states which feature non-classical correla- 8 tions are an important resource for quantum communica- √ tion and computing [6, 7]. In particular, entangled pho- where |GHZi = (|000i + |111i)/ 2 and 0 ≤ η ≤ 1. 3q ton pairs can be used for multiple applications, such as The definition means that ρW (η) is a mixture of a quantum key distribution (QKD) [8]. Other well-known Greenberger-Horne-Zeilinger (GHZ) state [22, 23] and applications of entangled states relate to: superdense the completely unpolarized state (i.e., maximally mixed coding [9, 10], quantum teleportation [11, 12], quantum states). The GHZ state can be obtained from three computing [13], quantum interferometric optical lithog- polarization-entangled spatially separated photons [24], raphy [14], etc. Entanglement in photonic states is not which makes it suitable for our framework. This kind limited only to photon pairs, but can be extended to of entangled state is famous for its importance in quan- three-qubit polarization entanglement [15]. Therefore, tum information, including quantum teleportation [25], the ability to characterize multipartite entangled states quantum secret sharing [26], or quantum cryptography based on measurements remains a relevant issue. [27]. Nonetheless, GHZ states remain a topic of intensive In this work, we follow the generalized approach to research, see, e.g., Ref. [28]. quantum measurement, which relies on positive operator- Since for η = 1 the density matrix Eq. 1 represents the valued measures (POVMs) [16]. More specifically, in our GHZ state, we pay special attention to this case. In our QST framework, we apply a symmetric, information- QST framework, we compare the accuracy of GHZ state ally complete, positive operator-valued measure (SIC- reconstruction with another three-qubit entangled state defined as: arXiv:2104.11258v1 [quant-ph] 22 Apr 2021 POVM) [17]. Additionally, we impose the Poisson noise to make the measurement scheme realistic. For different 1 scenarios, we test the framework numerically on three- |W i = √ (|001i + |010i + |100i) , (2) qubit generalized Werner states. Separately, we compare 3 the estimation of GHZ and W states. The accuracy of the framework is quantified and presented on graphs. which is commonly referred to as the W state. |GHZi In Sec. II, we introduce in detail the framework for and |W i are representatives of two very different classes QST of three-qubit states. Then, in Sec. III, the results of tripartite entangled states [29]. The W state was are presented and discussed. The main conclusions are proposed as a resource for several applications, such provided in the summary. as secure quantum communication [30]. Polarization- entangled W states can be obtained by parametric down- conversion from a single-photon source [31], which implies that it is applicable in the framework considered in ∗ [email protected] the present article. 2

3q B. Measurements receive coincidence counts, nκ , which can be modeled numerically as:

To extract information required for QST of three-qubit 3q 3q 3q generalized Werner states, we utilize a measurement nα = Neκ Tr Mκ ρW (η) , (6) scheme based on the SIC-POVM such that dim H = 2. For qubits, the SIC-POVM is defined by means of four where Neκ ∈ Pois(N ), i.e., the ensemble size for each vectors: act of measurement is selected randomly from the Pois- son distribution characterized by the mean value N , cf. 1 r2 Ref. [36]. On the basis of Eq. 6 we can numerically gen- |ξ1i = |0i |ξ2i = √ |0i + |1i 3q 3 3 erate noisy data for any three-qubit Werner state ρW (η). r We assume that the experimenter knows nothing about 1 2 i 2π |ξ3i = √ |0i + e 3 |1i (3) the state which has to be determined. For this reason, 3 3 the expected photon counts are given as: r 1 2 i 4π 3q 3q 3q |ξ4i = √ |0i + e 3 |1i , m = N Tr M σ , (7) 3 3 κ κ where σ3q represent a general 9 × 9 density matrix which where by {|0i , |1i} we denote the standard basis in H. can be factorized according to the Cholesky decomposi- According to the definition of the SIC-POMC, the mea- tion: surement operators are defined as rank−1 projectors: T †T σ3q = (8) 1 † P := |ξ ihξ | , (4) Tr(T T ) k 2 k k and T denotes a lower-triangular matrix which depends 4 P on 64 real parameters: t1, . . . , t64. The Cholesky decom- which satisfy k=1 Pk = I2. These four measurement operators are sufficient to perform single-qubit tomogra- position is commonly applied in QST frameworks because 3q phy [32]. The measurement scheme based on the SIC- it guarantees that the estimated state is physical, i.e. σ POVM can be realized on photons in order to character- is Hermitian, positive semi-defnite, of trace one [37, 38]. ize the polarization state of light. This scheme is mini- In this way, the problem of quantum state estima- mal, but at the same time it allows for efficient and reli- tion means that one needs to determine the parameters able single-qubit tomography. t1, . . . , t64 which fully characterize T . To figure out the In this framework, we assume that the three-qubit values that optimally fit to the noisy measurements, we 2 state is encoded in the polarization mode of photons. implement the χ −estimation, see, e.g., Ref. [39]. There- Each photon travels in a separate arm of the experimen- fore, we search for the minimum value of the following tal setup and undergoes measurements according to the function: 2 SIC-POVM scheme. Therefore, for state reconstruction 64 n3q − m3q of three photons, we propose three-qubit operators: 2 X κ κ χ (t1, . . . , t64) = 3q . (9) κ=1 mκ M 3q := P ⊗ P ⊗ P , (5) κ i j k This procedure allows one to simulate an experimental 3q where i, j, k = 1,..., 4 and, for simplicity, we used one scenario for any input state ρW (η) – first we generate symbol to denote the three-qubit operators, i.e. κ ≡ noisy photon counts Eq. 6 and then we can recover the state by finding the parameters t1, . . . , t64 for which the (i, j, k). Naturally, from the definition Eq. 5, we have 2 64 three-qubit measurement operators. This measure- χ function is minimized. We proceed analogously when ment scheme is minimal as far as three-qubit states are considering QST of the W state given in Eq. 2. concerned. Therefore, it appears intriguing to investigate the performance of such measurements in a noisy D. Performance analysis scenario.

Two figures of merit are introduced to investigate the performance of the measurement scheme on generalized C. Methods of state reconstruction 3q Werner states. First, every input Werner state ρW (η) is compared with the result of χ2−estimation, σ3q, by We analyze the accuracy of the measurement scheme computing the quantum fidelity [40–42]: based on the SIC-POVM under conditions which include the Poisson noise, which is a typical source of error in q√ √ 2 3q 3q 3q 3q 3q single-photon counting [33–35]. Then, by N we denote F(σ , ρW (η)) := Tr σ ρW (η) σ . (10) the average ensemble size, i.e. the number of quantum systems produced by the source per measurement. In This figure is commonly applied to evaluate the accu- the three-qubit framework, we assume that the detectors racy of a QST framework, in particular, under imperfect 3 measurement settings, see, e.g., Ref. [43–45]. In our sce- QST remains at a decent level, i.e., F(η) > 0.87 for all nario, the quantum fidelity is treated as a function of η η. The plot seems irregular, which is caused by the ran- (and denoted by F(η)) to track the precision of the frame- dom nature of the noise introduced into the measurement work along the domain of generalized Werner states. scheme. Next, we analyze how much the estimated states are Finally, for N = 10, we witness a significant decline in mixed. Thus, we follow the standard formula for the the quality of state recovery. As expected, the Poisson purity [16], which implies that we compute γ ≡ Tr(ρ2), noise has a damaging impact on the quantum fidelity if where ρ denotes a density matrix. The purity of the we utilize only 10 systems per measurement. We notice states resulting from the QST framework is calculated that the accuracy of state estimation irregularly goes up and compared with the input states. and down, reflecting the randomness of noise. Neverthe- less, a conclusion can be formulated that F(η) improves as we increase η. This implies that the scenario with III. RESULTS AND ANALYSIS N = 10 performs better if the three-qubit generalized Werner state is close to pure. We consider three measurement scenarios that differ in 1 ● the number of quantum systems provided by the source Ensemble size ●★ ★●★■■ ★●■ per measurement (ensemble size). To be more specific, ● theory ★ 1000 ■ 100 ▼ 10 ★●■■ 0.8 ■★●■ we assume that N = 10, 100, 1 000. This allows us to ★● ★●■ ▼▼ ★● ▼ ▼▼ investigate the performance of the framework versus the ★■★●■▼▼ ▼ 0.6 ★●▼▼ ▼ amount of Poisson noise which is strictly connected with ▼ ▼●▼ ■ ●■★●■ ▼■★■★●★ the average number of systems involved in one measure- γ(η) ■ ●■▼■ ▼ ▼■▼★●★●▼ ment. 0.4 ■★●■★● ▼ ▼ ■★●■▼★●▼ ▼ ▼ ▼▼ ★■★●■★●▼ ▼▼▼ ▼ ▼▼▼ ▼■★■★●■●▼ 0.2 ▼ ▼ ★■▼★●■★●● ■■★●■★●■★●■● ★●■★●■★●■★●■■★●★●■★●★● A. Tomography of three-qubit generalized Werner states 0 0 0.2 0.4 0.6 0.8 1 η For the three values of N , we perform QST along the admissible interval of η. In Fig. 1, one can observe the FIG. 2: Plots present the purity, γ(η), in QST of plots of the quantum fidelity, F(η). First, for N = 1 000, three-qubit generalized Werner states. one can observe high quality of quantum state estimation. We notice that F(η) > 0.975 for all η. This confirms that the Poisson noise has only a limited impact on state esti- In Fig. 2, one finds the purity of the estimated states 3q mation if we utilize a relatively high number of quantum σ compared with the actual purity the three-qubit gen- 3q systems per measurement. Although the framework is eralized Werner states ρW (η). The plot confirms the based on the minimal set of measurement operators, it is previous observations. For N = 1 000, the plot of γ(η) robust against the noise in this scenario. strictly overlaps with the theoretical value. If N = 100, we detect minor discrepancies between the purity of 1 ★ ★★★ ★ ★ ★ ★ the estimates and the actual value. Furthermore, when ★ ★ ★★ ★★★ ★★★★★★★★★★★★ ★ ★★ ★★ ★★★ ★★ ■ ■■ ■ ★ ★★ ★ ■ ■■ ■ ■■■■ ■ ■ ■ N = 10, we notice that the plot departs significantly from ■ ■ ■ ■■ ■ ■ ■ ■■■■ ■ the theoretical curve. Interestingly, in the case of purity, 0.9 ■ ■ ■ ■ ■ ■ ■ ■■ ■■ ■ ▼ ■ ▼ ▼ ▼ the estimated figures coincide with the actual values for ▼ ▼▼ ▼ ▼ ▼ ▼ a middle range of η, i.e., 0.4 < η < 0.8. ▼ 0.8 ▼▼ ▼ ▼ ▼▼ ▼ ▼ ▼

F(η) ▼ ▼ ▼ ▼ ▼ ▼ 0.7 ▼ B. Tomography of GHZ state and W state ▼ ▼▼ ▼ ▼ ▼ ▼ ▼ Ensemble size ▼ ▼ ▼ 0.6 ▼ ★ 1000 ■ 100 ▼ 10 ▼ As already indicated, for η = 1, the quantum state Eq. 1 corresponds to the GHZ state. Since |GHZi plays a crucial role in quantum information theory, it appears 0 0.2 0.4 0.6 0.8 1 justifiable to consider this special case separately. In η Fig. 3, one finds, in the left-hand side column, the reconstructed state for three selected ensemble sizes. In the FIG. 1: Plots present the fidelity, F(η), in QST of case of N = 1 000 and N = 100, we present only the real three-qubit generalized Werner states. part of the estimated density matrix because the imaginary part is negligible. For N = 10, we visualize both Next, if N = 100, we observe a moderate impact of the real and imaginary parts to show inaccuracies of state Poisson noise on the state tomography. The accuracy of estimation in this scenario. Similarly, in the right-hand 4

state |GHZi |W i N

1000

100

FIG. 3: Density matrices obtained from the framework for two input states: |GHZi and |W i. For N = 10, the panels show real (a) and imaginary (b) parts of the reconstructed state. side column of Fig. 3, one has the results of applying the |W i can be accurately reconstructed by the framework. QST framework to the W state. The fidelity between |GHZi and its estimate is 0.9934. First, for N = 1 000, we notice that both |GHZi and For |W i the overlap equals 0.9997. This confirms that the 5 framework is efficient not only for three-qubit generalized is vulnerable to the Poisson noise, which results in poor Werner states but also for the W state which features a fidelity. However, the accuracy of state estimation im- different kind of entanglement. proves as we reduce the entropy of the input states. If N = 100, one can spot minor flaws in the results of As a special case, we investigated the performance of state estimation. It can be expressed numerically by the the framework on the GHZ state, which corresponds to quantum fidelity, which equals 0.9540 for the GHZ state the pure three-qubit generalized Werner state. The accu- estimation and 0.9682 in the case of the W state. racy of the tomographic technique was compared with the Finally, if N = 10, we notice a considerable amount of estimation of the W state, which is considered a distinct errors in the reconstructed states. In particular, the state kind of three-qubit entanglement. It was demonstrated tomography of |GHZi yields the fidelity equal 0.8216. that the framework is more accurate at reconstructing Errors are noticeable not only in the real part of the the W state, which was particularly evident for the small reconstructed density matrix but also in the imaginary ensemble size. Thus, we can conclude that the two repre- which should be flat. It appears that the Poisson noise is sentatives of three-qubit entanglement, i.e., |GHZi and less detrimental for W state estimation, which results in |W i, differ in the quality of state tomography under the the fidelity: 0.9424. To conclude, we can state that the Poisson noise. framework seems more robust against the Poisson noise In the future, the framework will be tested on other if is applied to the W state reconstruction rather than classes of multipartite entangled states. Additionally, al- the GHZ state. ternative kinds of experimental noise will be incorporated into the model to study the robustness of the method in different scenarios. IV. DISCUSSION AND SUMMARY

In the article, we introduced a framework for three- qubit quantum tomography, which is based on the SIC- ACKNOWLEDGMENTS POVM and involves the Poisson noise as a source of experimental uncertainty. By numerical simulations, we The author acknowledges ﬁnancial support from the tested the framework on three-qubit generalized Werner Foundation for Polish Science (FNP) (project First Team states. In particular, we discovered that if we exploit only co-ﬁnanced by the European Union under the European 10 copies of the system per measurement, the framework Regional Development Fund).

[1] M. G. A. Paris and J. Reháˇcek(eds.),ˇ Quantum State [12] M. Anderson, T. Muller, J. Skiba-Szymanska, Estimation (Lecture Notes in Physics), Springer, Berlin- A. B. Krysa, J. Huwer, R. M. Stevenson, J. Heffernan, Heidelberg (2004) doi:10.1007/b98673 D. A. Ritchie, and A. J. Shields, Phys. Rev. Appl. 13, [2] G. M. D’Ariano, M. G. A. Paris, and M. F. Sac- 054052 (2020) doi:10.1103/PhysRevApplied.13.054052 chi, Adv. Imaging Electron Phys. 128, 205 (2003) doi: [13] R. Jozsa and N. Linden, Proc. R. Soc. A 459, 2011 (2003) 10.1016/S1076-5670(03)80065-4 doi:10.1098/rspa.2002.1097 [3] Z. Hradil, Phys. Rev. A 55, 1561(R) (1997) doi: [14] A. N. Boto, P. Kok, D. S. Abrams, S. L. Braunstein, 10.1103/PhysRevA.55.R1561 C. P. Williams, and J. P. Dowling, Phys. Rev. Lett. 85, [4] T. Opatrny, D.-G. Welsch, and W. Vogel, Phys. Rev. A 2733 (2000) doi:10.1103/PhysRevLett.85.2733 56, 1788 (1997) doi:10.1103/PhysRevA.56.1788 [15] D. R. Hamel, L. K. Shalm, H. Hübel, A. J. Miller, F. Mar- [5] A. Acharya, T. Kypraios, and M. Gut¸˘a, J. Phys. sili, V. B. Verma, R. P. Mirin, S. W. Nam, K. J. Resch, A: Math. Theor. 52, 234001 (2019) doi:10.1088/1751- and T. Jennewein, Nature Photon. 8, 801–807 (2014) doi: 8121/ab1958 10.1038/nphoton.2014.218 [6] A. Einstein, B. Podolsky, and N. Rosen, Phys. Rev. 47, [16] M. A. Nielsen and I. L. Chuang, Quantum Computation 777 (1935) doi:10.1103/PhysRev.47.777 and Quantum Information, Cambridge University Press, [7] J. S. Bell, Physics Physique Fizika 1, 195 (1964) doi: Cambridge (2000) doi:10.1017/CBO9780511976667 10.1103/PhysicsPhysiqueFizika.1.195 [17] J. M. Renes, R. Blume-Kohout, A. J. Scott, [8] A. K. Ekert, Phys. Rev. Lett. 67, 661 (1991) doi: C. M. Caves, J. Math. Phys. 45, 2171-2180 (2004) doi: 10.1103/PhysRevLett.67.661 10.1063/1.1737053 [9] C. H. Bennett and S. J. Wiesner, Phys. Rev. Lett. 69, [18] R. F. Werner, Phys. Rev. A 40, 4277 (1989) doi: 2881 (1992) doi:10.1103/PhysRevLett.69.2881 10.1103/PhysRevA.40.4277 [10] K. Mattle, H. Weinfurter, P. G. Kwiat, and [19] M. Barbieri, F. De Martini, G. Di Nepi, and A. Zeilinger, Phys. Rev. Lett. 76, 4656 (1996) doi: P. Mataloni, Phys. Rev. Lett. 92, 177901 (2004) doi: 10.1103/PhysRevLett.76.4656 10.1103/PhysRevLett.92.177901 [11] C. H. Bennett, G. Brassard, C. Crepeau, R. Jozsa, [20] W. Dürand J. I. Cirac, Phys. Rev. A61, 042314 (2000) A. Peres, and W. K. Wootters, Phys. Rev. Lett. 70, 1895 doi:10.1103/PhysRevA.61.042314 (1993) doi:10.1103/PhysRevLett.70.1895 [21] J. Siewert and C. Eltschka, Phys. Rev. Lett. 108, 230502 (2012) doi:10.1103/PhysRevLett.108.230502 6

[22] D. M. Greenberger, M. A. Horne, A. Shimony, and [34] S. W. Hasinoff, Photon, poisson noise, in: K. Ikeuchi A. Zeilinger, Am. J. Phys. 58, 1131 (1990) doi: (eds.), Computer Vision, Springer, Boston, MA, pp. 608- 10.1119/1.16243 610 (2014) doi:10.1007/978-0-387-31439-6˙482 [23]M. Zukowski,˙ A. Zeilinger, M. A. Horne, and H. We- [35] D. Shin, A. Kirmani, V. K. Goyal and J. H. Shapiro, infurter, Acta Phys. Pol. A 93, 187 (1998) doi: IEEE Trans. Comput. Imaging. 1, 112 (2015) doi: 10.12693/APhysPolA.93.187 10.1109/TCI.2015.2453093 [24] D. Bouwmeester, J.-W. Pan, M. Daniell, H. Weinfurter, [36] R. T. Thew, K. Nemoto, A. G. White, and A. Zeilinger, Phys. Rev. Lett. 82, 1345 (1999) doi: W. J. Munro, Phys. Rev. A 66, 012303 (2002) doi: 10.1103/PhysRevLett.82.1345 10.1103/PhysRevA.66.012303 [25] A. Karlsson and M. Bourennane, Phys. Rev. A 58, 4394 [37] D. F. V. James, P. G. Kwiat, W. J. Munro, and (1998) doi:10.1103/PhysRevA.58.4394 A. G. White, Phys. Rev. A 64, 052312 (2001) doi: [26] M. Hillery, V. Buzek, and A. Berthiaume, Phys. Rev. A 10.1103/PhysRevA.64.052312 59, 1829 (1999) doi:10.1103/PhysRevA.59.1829 [38] J. Altepeter, E. Jerey, and P. Kwiat, Adv. At. Mol. Opt. [27] J. Kempe, Phys. Rev. A 60, 910 (1999) doi: Phys. 52, 105 (2005) doi:10.1016/S1049-250X(05)52003- 10.1103/PhysRevA.60.910 2 [28] Y. Zhao, R. Zhang, W. Chen, X.-B. Wang, and J. Hu, [39] B. Jack, J. Leach, H. Ritsch, S. M. Barnett, M. J. Pad- npj Quantum Inf. 7, 24 (2021) doi:10.1038/s41534-021- gett, and S. Franke-Arnold, New J. Phys. 11, 103024 00364-8 (2009) doi:10.1088/1367-2630/11/10/103024 [29] W. Dür,G. Vidal, and J. I. Cirac, Phys. Rev. A 62, [40] A. Uhlmann, Rep. Math. Phys. 24, 229 (1986) doi: 062314 (2000) doi:10.1103/PhysRevA.62.062314 10.1016/0034-4877(86)90055-8 [30] W. Jian, Z. Quan,and T. Chao-Jing, Commun. Theor. [41] R. Jozsa, J. Mod. Opt. 41, 2315 (1994) doi: Phys. 48, 637 (2007) doi:10.1088/0253-6102/48/4/013 10.1080/09500349414552171 [31] T. Yamamoto, K. Tamaki, M. Koashi, and [42] I. Bengtsson and K. Zyczkowski,˙ Geometry of Quan- N. Imoto, Phys. Rev. A 66, 064301 (2002) doi: tum States. An Introduction to Quantum Entanglement, 10.1103/PhysRevA.66.064301 Cambridge University Press, Cambridge (2006) doi: [32]J. Reháˇcek,ˇ B.-G. Englert, and D. Kasz- 10.1017/CBO9780511535048 likowski, Phys. Rev. A 70, 052321 (2004) doi: [43] J. G. Titchener, M. Gräfe,R. Heilmann, A. S. Solntsev, 10.1103/PhysRevA.70.052321 A. Szameit, and A. A. Sukhorukov, npj Quantum Inf. 4, [33] S. Hernandez-Marin, A. M. Wallace, and G. J. Gibson, 19 (2018) doi:10.1038/s41534-018-0063-5 IEEE Trans. Pattern Anal. Mach. Intell. 29, 2170 (2007) [44] H. Yuan, Z.-W. Zhou, and G.-C. Guo, New J. Phys. 18, doi:10.1109/TPAMI.2007.1122 043013 (2016) doi:10.1088/1367-2630/18/4/043013 [45] D. Rosset, R. Ferretti-Schöbitz,J.-D. Bancal, N. Gisin, and Y.-C. Liang, Phys. Rev. A 86, 062325 (2012) doi: 10.1103/PhysRevA.86.062325