Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

PHYSICAL REVIEW A 97, 062309 (2018)

Experimental witness of genuine high-dimensional entanglement


Yu Guo, Xiao-Min Hu, Bi-Heng Liu,* Yun-Feng Huang, Chuan-Feng Li,† and Guang-Can Guo
CAS Key Laboratory of Quantum Information, University of Science and Technology of China, Hefei 230026, People’s Republic of China
and Synergetic Innovation Center of Quantum Information and Quantum Physics, University of Science and Technology of China, Hefei,
Anhui 230026, People’s Republic of China

(Received 16 January 2018; published 7 June 2018)

Growing interest has been invested in exploring high-dimensional quantum systems, for their promising
perspectives in certain quantum tasks. How to characterize a high-dimensional entanglement structure is one
of the basic questions to take full advantage of it. However, it is not easy for us to catch the key feature of
high-dimensional entanglement, for the correlations derived from high-dimensional entangled states can be
possibly simulated with copies of lower-dimensional systems. Here, we follow the work of Kraft et al. [Phys. Rev.
Lett. 120, 060502 (2018)], and present the experimental realizing of creation and detection, by the normalized
witness operation, of the notion of genuine high-dimensional entanglement, which cannot be decomposed into
lower-dimensional Hilbert space and thus form the entanglement structures existing in high-dimensional systems
only. Our experiment leads to further exploration of high-dimensional quantum systems.

DOI: 10.1103/PhysRevA.97.062309

To get full use of quantum potential, it is intuitive to ex- of high-dimensional systems may be mimicked by copies of
pand quantum systems to multiparticles and high dimensions. Bell pairs [20]. Essentially, the structure of high-dimensional
Within photonic systems, ten-photon entanglement [1] and entanglement is much more complicated. Recently, Kraft
entangled photon pairs in more than 100 dimensions [2] have et al. proposed the notion of genuine multilevel entanglement
been realized experimentally, thus extending available Hilbert (GMLE) [21], referring to high-dimensional quantum states
space by at least two orders of magnitude compared with that can give rise to correlations that cannot be simulated
common qubit pairs. Several eminent works utilizing high- by copies of lower-dimensional systems. Also, they provided
dimensional systems have showing tremendous advantages methods based on entanglement witness [22] to characterize
over qubit systems in fundamental investigations of quantum GMLE for both the bipartite and multipartite cases, which
mechanics [3–5], quantum simulations of a richer variety [6], cannot be distinguished with decomposable ones by the afore-
quantum communication protocols with improved robustness mentioned characterization tools [15–17]. In this article, we
and key rate [7], and more efficient and error-tolerant quantum experimentally realize the detection of bipartite GMLE with
computation [8]. high-quality entangled qutrits and ququarts.
On the other hand, characterization of high-dimensional In the scenario given by Ref. [21], spatially separated
entanglement remains a real challenge and has captured Alice and Bob are distributed with a two-ququart entangled
growing attention. Despite full characterization of quantum state, for example. After recoding their ququart with two
state which is extremely cumbersome [9] or requires extra qubits respectively, Alice and Bob now possess a four-qubit
assumptions [10,11] in practice, there are two main ways to state. Such four-qubit state is said to be decomposable if
characterize high-dimensional entangled states. The first one it can be decomposed into a product state of two copies
is by quantifying the ebit number [12–14] required to reproduce of two qubits. Otherwise, it is genuinely four-dimensional
such state. The second is by measuring the entanglement entangled. The difference of the indecomposable and decom-
dimensionality of it [15–17]. Specially, violation of a Bell posable state is apparent and profound, for the former can be
inequality [18] has been used to certify entanglement of performed with arbitrary local transformations on the local
dimension d = 11 [17]. While the former [12–14] provides an four-dimensional space, while for the later, local operations
operational measurement of how much entanglement there is, are restricted to two two-dimensional spaces and crossover
verifying entanglement dimensionality [15–17] focuses more between them is forbidden. More abundant operations mean
on the structure of high-dimensional states, which is what we that genuine multilevel state can produce some correlations that
are really concerned with in this article. can never be obtained by decomposable systems of the same
It seems that we have had powerful means [15–17] to dimension.
explore the high-dimensional entanglement structure. How- From the definition of GMLE, the set of decomposable
ever, the recent works [19,20] cautioned that Bell inequalities states S is convex, allowing us to construct an entanglement
may neglect key features of quantum systems, for behavior witness for GMLE. Reference [23] suggested the projector
based witnesses which are of the form W = αI − |ξ ξ |,
where α is the maximal squared overlap between |ξ  and the
*
bhliu@ustc.edu.cn decomposable states S. Based on such W, Ref. [21] provided

cfli@ustc.edu.cn the following two states that locally maximize the distance to

2469-9926/2018/97(6)/062309(5) 062309-1 ©2018 American Physical Society


GUO, HU, LIU, HUANG, LI, AND GUO PHYSICAL REVIEW A 97, 062309 (2018)

the decomposable states. For the Schmidt rank-3 state, it is + 3(O(0,1) + O(0,2) + O(0,3))

|ξ1  = (|00 + |11 + |22)/ 3 (1) + (O(1,2) + O(1,3) + O(2,3))]/12 (4)

with α1 = [(3 + 5)/6]  0.934. For the Schmidt rank-4
1/2 with
state, it is
√ √ O(i,j ) = |iijj | + |jj ii|
|ξ2  = |00 3/2 + (|11 + |22 + |33)/(2 3) (2) + + + + − − − −
√ = |Xi,j Xi,j Xi,j Xi,j | + |Xi,j Xi,j Xi,j Xi,j |
with α2 = [(3 + 2 2)/6]1/2  0.986. + + + + − − − −
A negative expectation value of the observable W1 (W2 ) −|Yi,j Yi,j Yi,j Yi,j | − |Yi,j Yi,j Yi,j Yi,j |, (5)
clearly signifies that the state σ is genuine three-dimensional ±
√ ±

where |Xi,j  = 1/ 2(|i ± |j ), |Yi,j  = 1/ 2(|i ± i|j ).
entangled (genuine four-dimensional entangled).
Thus we can implement the witness of GMLE through local
To detect GMLE experimentally with W1 (W2 ), we need
operation and classical communication. (2) To have the ability
to take mainly two obstacles in our stride, as follows: (1) To
to prepare a high quality two-particle high-dimensional en-
decompose W1 (W2 ) into a set of local projective measurement
tangled state. Taking noise into consideration, quantum states
[24]. The witness operator W1 can be decomposed as
prepared in labs take the form of ρ = (1 − p)|ξ ξ | + pI/d
W1 = α1 I − |ξ1 ξ1 | (considering white noise only for simplification), where |ξ ξ |
⎛ ⎞ is the ideal state that we really want and d is the dimension
 2 
2 
of the system. For the Schmidt rank-3 (-4) state, it is obvious
= α1 I − ⎝ |iiii| + O(i,j )⎠ 3. (3)
that tr(Wσ ) would be always positive and therefore we cannot
i=0 i,j =0,i<j
detect GMLE experimentally, as long as p > 0.066 (p >
Similarly, W2 can be decomposed as 0.014). This makes the state preparation progress stringent and
can be hardly achieved by a widely used high-dimensional
W2 = α2 I − |ξ2 ξ2 | entanglement source based on orbital angular momentum
= α2 I − [(9|0000| + |1111| + |2222| + |3333|) (OAM) with the present technology [25,26]. However, a

D2

Bob

LCVR3

HWP45
TP

LCVR2
ppK

Source LCVR1

HWP2

HWP4
HWP3
QWP1
HWP5
HWP1
Alice
HWP6
D1

PBS PBS HWP HWP


HWP QWP IF
BD @404nm @808nm @808nm @808nm
LCVR @404nm @404nm @808nm
&808nm &808nm

FIG. 1. Experimental setup. A cw violet laser (power is 3 mW, wavelength is 404 mm) serves as the pump source and is separated by a
beam displacer. Adjust the angles of HWP1-3 in the figure and the pump state can be written as a|0 + beiψ1 |1 + ceiψ2 |2 + deiψ3 |3 after we
encode H (V ) photon at upper path as |0 (|1) and H (V ) photon at lower path as |2 (|3). The pump photon is then incident to a Sagnac
interferometer to pump type-II cut periodically poled potassium titanyl phosphate (ppKTP) crystal (1 mm × 7 mm × 10 mm) and generates
the state of the form as a|00 + beiψ1 |11 + ceiψ2 |22 + deiψ3 |33. To ensure the phase stability between the two paths, the temperature of
the ppKTP is controlled with a homemade temperature controller and the temperature stability is 0.001 K. The LCVRs (LCVR1 and LCVR2
are set at 0◦ and LCVR3 is set at 90◦ ) can be set to introduce 0 phase by properly applying voltages, thus arbitrary states of the form of
a|00 + b|11 + c|22 + d|33 can be prepared. This state is then distributed to Alice and Bob who finish the witness process by measuring
their photon locally. BD: beam displacer; PBS: polarizing beam splitter; HWP: half wave plate; QWP: quarter wave plate; LCVR: liquid crystal
variable retarder.

062309-2
EXPERIMENTAL WITNESS OF GENUINE HIGH- … PHYSICAL REVIEW A 97, 062309 (2018)

high-quality high-dimensional entangled state can be achieved TABLE I. The setting angles of HWPs for the preparation of
with the phase stable high-dimensional path-polarization hy- the chosen states in our experiment when the pump photon is
brid entanglement source based on the beam displacer polar- horizontally polarized. In three-dimensional
√ case, the angle of HWP3
ization Mach-Zehnder interferometer [3,4]. is always 0◦ ; d = 0 and c = 1 − a 2 − b2 are always satisfied. In

The experimental setup is shown in Fig. 1. A cw violet four-dimensional case,the angle of HWP3 is always 22.5 and we set
laser is separated by 4.21 mm with beam displacer (BD) b = c = d, thus b = (1 − a 2 )/3.
working at 404 mm, then these two beams are incident to a
Sagnac interferometer to pump type-II cut periodically poled Three-dimensional case
potassium titanyl phosphate (ppKTP) crystal and generate √ line 1 : a = 0.65 line 2 : b = 0.65
two photon polarization entanglement (|H H  + |V V )/ 2 in
each path [27]. We encode the horizontal polarized (H ) photon b HWP1 HWP2 a HWP1 HWP2
0.25 −19.8◦ −40.8◦ 0.25 −19.8◦ −4.2◦
and the vertical polarized (V ) photon at upper path as |0 and
0.35 −22.0◦ −36.9◦ 0.35 −22.0◦ −8.1◦
|1 respectively, and we encode the H and V photons at lower ◦
0.45 −25.7 −32.2◦ 0.45 −25.7◦ −12.8◦
path as |2 and |3. Through the adjustment of half wave plates
0.55 −31.0◦ −27.2◦ 0.55 −31.0◦ −17.8◦
(HWPs) working at 404 nm and the computer-controlled liquid 0.65 −37.7 ◦
−22.5◦ 0.65 −37.7◦ −22.5◦
crystals variable retarders (LCVRs) in Fig. 1, arbitrary pure 0.75 −44.4 ◦
−18.4◦ 0.75 −44.4◦ −26.5◦
states of the form a|00 + beiψ1 |11 + ceiψ2 |22 + deiψ3 |33 line 3 : a + b = 1 line 4 : a = b
(a, b, c, and d are real and satisfy a 2 + b2 + c2 + d 2 = 1)
can be prepared. The two photons are then distributed to a HWP1 HWP2 a(b) HWP1 HWP2
Alice and Bob and measured by them locally. When the state 0.25 −28.2◦ −3.2◦ 0.45 −12.8◦ −22.5◦
is prepared on (|00 + |11 + |22 + |33)/2, the interference 0.35 −22.0◦ −8.1◦ 0.55 −23.6◦ −22.5◦
visibility between |00 and |11 (|00 and |33) is over 0.996 0.45 −18.2◦ −16.9◦ 0.577 −27.4◦ −22.5◦
(0.990). Such visibilities are sufficient for the detection of 0.55 −18.2◦ −28.1◦ 0.65 −37.7◦ −22.5◦
0.65 −22.0◦ −36.9◦ 0.70 −44.2◦ −22.5◦
GMLE. When the pump power is 3 mW, we get a photon
0.75 −28.2◦ −41.8◦
pair rate at 2500 counts/s and the overall detection efficiency
Four-dimensional case
is 0.222.
The measurement setup, composed of QWPs, HWPs, BDs, a HWP1 HWP2 a HWP1 HWP2
and PBSs, allows us to project Alice’s (Bob’s) photon to 0.756 −35.5◦ −38.0◦ 0.866 −40.6◦ −41.8◦
± ± −42.5◦ −43.2◦ −43.4◦ −43.8◦
arbitrary computational basis {|l} and |Xi,j  and |Yi,j . Two 0.918 0.945
fiber-coupled single-photon detectors D1 and D2 are used to
detect the photons. Interference filters with a bandwidth of 3
nm before the detectors are used to remove the background value of tr(W1 σ ) on the lines (light blue lines) and the experi-
+
photon noise. We take Alice’s photon to be projected to |Y0,2  mental results of the chosen states along these lines (red dots)
as an example and others can be achieved analogously. Alice respectively. The experimental and theoretical results match
sets HWP4, HWP5, HWP6, and QWP1 in Fig. 1 at 0◦ , 45◦ , very well and demonstrate our ability to witness genuine three-
+
22.5◦ , and 0◦ , respectively, thus can project her photon to |Y0,2 . dimensional entanglement for almost all theoretical allowed
After getting all projection probabilities defined by Eq. (3) or states. The integration time for each data is 20 s to reduce the
Eq. (4), the expectation value of W on a certain state σ , i.e., statistic error. Specially, the minimum of tr(W1 σ ) is −0.066√
tr(Wσ ) can be calculated. We can then confirm that σ is GMLE when the coefficients in Eq. (1) satisfy a = b = c = 1/ 3
with a negative value of tr(Wσ ). [represented by the third dot in Fig. 2(e)]. The experimental
We prepared 23 three-dimensional entangled states and four result is
four-dimensional states to perform the witness task. These
tr(W1 |ξ1 ξ1 |) = −0.050 ± 0.004; (6)
states are all of the form a|00 + b|11 + c|22 + d|33, where
the coefficients a, b, c, and d are real and satisfy a 2 + b2 + the error bar is due to the statistics error and is estimated by
c2 + d 2 = 1, and can be prepared by properly setting the angles Monte Carlo simulation, and the fidelity [28] of state |ξ1  is
of HWP1-3 in Fig. 1; detailed information is shown in Table I. 0.986 ± 0.004. To observe genuine four-dimensional entan-
The experimental results are shown in Fig. 2. For the detection glement, one needs to improve the fidelity of the state. Thus
of three-dimensional entanglement, we set d = 0, thus the state we make a more careful adjustment on the coefficients a,b,c,d
to be detected can be uniquely affirmed by a and b. Figure 2(a) in the four-dimensional case and increase the data collection
presents the distribution of the experimental states with differ- time to 30 s to suppress the statistic error. The experimental
ent a and b. Within the light blue region, the expectation values witness results of four-dimensional entanglement are shown in
of W1 on these states are negative, which means these states can Fig. 2(f), where we restrict that b = c = d and vary the ratio
be witnessed as genuine three-dimensional entangled. Outside of a to b. When measuring |ξ2  [Eq. (2) and represented by the
of this region, positive expectation values imply the failure to second dot in Fig. 2(f)], the minimum of tr(W2 σ ) is −0.014.
draw a conclusion. However, if ignoring the notion of GMLE, Our experimental result reads
these states are all entangled and have no distinctions. We
tr(W2 |ξ2 ξ2 |) = −0.0091 ± 0.0023, (7)
choose the states represented by the dots along four straight
lines, satisfying a = 0.65 (pink line), b = 0.65 (green line), where the fidelity [28] of state |ξ2  is 0.991 ± 0.003. This
a + b = 1 (red line), and a = b (blue line), in the ab plane clearly proves with high statistical significance that the
to be witnessed. Figures 2(b)–2(e) present both the theoretical observed state |ξ2  is genuine four-dimensional entangled.

062309-3
GUO, HU, LIU, HUANG, LI, AND GUO PHYSICAL REVIEW A 97, 062309 (2018)

FIG. 2. Distribution of experimental states and the value of tr(Wσ ). (a) The distribution of experimental states with different coefficients
a’s and b’s in the case of three dimensions. The light blue region represents the states that the expectation value of W on them is negative.
The colored dots are the experimental states which are distributed along four straight lines satisfying a = 0.65 (pink line), b = 0.65 (green
line), a + b = 1 (red line), and a = b (blue line). (b)–(e) Experimental results of tr(Wσ ) along the lines in (a) respectively. Light blue lines
are the theoretical value along these lines. The red dots represent the experimental results of the chosen states. (f) Experimental results of
four-dimensional case. Restricting b = c = d, the light blue line is theoretical value of tr(Wσ ) when scanning a. Experimental values are given
by the red dots as well. Error bars are due to the statistics error.

We would like to emphasize that the aforementioned charac- present experiment represents a step towards the practical
terization tools [15–17] fail to signify GMLE, indicating the realization of the protocols to test if a state is genuine high-
superiority of the witness method we used [21]. More impor- dimensional entangled.
tantly, it is hard to certify GMLE with a widely used high-
dimensional source based on OAM for their poor performance This work was supported by the National Key Re-
of resisting noise with present technology [25,26]. Our phase search and Development Program of China (Grants
stable high-dimensional path-polarization hybrid entangle- No. 2017YFA0304100, No. 2016YFA0301300, and No.
ment source shows a significant advantage in preparation of 2016YFA0301700), NNSFC (Grants No. 11774335, No.
high-quality arbitrary high-dimensional entangled states. 61327901, No. 11474268, and No. 11504253), the Key
In conclusion, we have reported experiment of detection Research Program of Frontier Sciences, CAS (Grant No.
of genuine high-dimensional entanglement, which cannot be QYZDY-SSW-SLH003), the Fundamental Research Funds for
simulated by lower-dimensional systems and thus can present the Central Universities, and Anhui Initiative in Quantum
essentially different resources. Utilizing a phase stable high- Information Technologies (Grants No. AHY020100 and No.
dimensional path-polarization hybrid entanglement source, the AHY060300).

[1] X.-L. Wang et al., Experimental Ten-Photon Entanglement, Entangled Photonic Qutrits, Phys. Rev. Lett. 120, 180402
Phys. Rev. Lett. 117, 210502 (2016). (2018).
[2] M. Krenn, M. Huber, R. Fickler, R. Lapkiewicz, S. Ramelow, [5] A. Cabello and M. T. Cunha, Proposal of a Two-Qutrit Con-
and A. Zeilinger, Generation and confirmation of a (100 × 100)- textuality Test Free of the Finite Precision and Compatibitity
dimensional entangled quantum system, Proc. Natl. Acad. Sci. Loopholes, Phys. Rev. Lett. 106, 190401 (2011).
USA 111, 6243 (2014). [6] M. Neeley et al., Emulation of a quantum spin with a supercon-
[3] X.-M. Hu, J.-S. Chen, B.-H. Liu, Y. Guo, Y.-F. Huang, Z.-Q. ducting phase qudit, Science 325, 722 (2009).
Zhou, Y.-J. Han, C.-F. Li, and G.-C. Guo, Experimental Test [7] I. Ali-Khan, C. J. Broadbent, and J. C. Howell, Large-Alphabet
of Compatibility-Loophole-Free Contextuality with Spatially Quantum Key Distribution Using Energy-Time Entangled Bi-
Sepatated Entangled Qutrits, Phys. Rev. Lett. 117, 170403 partite States, Phys. Rev. Lett. 98, 060503 (2007).
(2016). [8] B. P. Lanyon, M. Barbieri, M. P. Almeida, T. Jennewein, T. C.
[4] X.-M. Hu, B.-H. Liu, Y. Guo, G.-Y. Xiang, Y.-F. Huang, Ralph, K. J. Resch, G. J. Pryde, J. L. O’Brien, A. Gilchrist, and A.
C.-F. Li, C.-G. Guo, M. Kleinmann, T. Vértesi, and A. G. White, Simplifying quantum logic using higher-dimensional
Cabello, Observation of Stronger-than-Binary Correlations with Hilbert spaces, Nat. Phys. 5, 134 (2009).

062309-4
EXPERIMENTAL WITNESS OF GENUINE HIGH- … PHYSICAL REVIEW A 97, 062309 (2018)

[9] M. Agnew, J. Leach, M. Mclaren, F. Stef Roix, and R. W. Boyd, [18] D. Collins, N. Gisin, N. Linden, S. Massar, and S. Popescu, Bell
Tomography of the quantum state of photons entangled in high Inequalities for Arbitrarily High-Dimensional Systems, Phys.
dimensions, Phys. Rev. A 84, 062101 (2011). Rev. Lett. 88, 040404 (2002).
[10] D. Gross, Y.-K. Liu, S. T. Flammia, S. Becker, and J. Eisert, [19] Y. Cai, Quantum sizes: Complexity, Dimension and Many-
Quantum State Tomography via Compressed Sensing, Phys. box Locality, PhD. thesis, National University of Singapore,
Rev. Lett. 105, 150401 (2010). Singapore, 2015.
[11] G. A. Howland, S. H. Knarr, J. Schneeloch, D. J. Lum, and [20] W. Cong, Y. Cai, J. D. Bancal, and V. Scarani, Witnessing
J. C. Howell, Compressively Characterizing High-Dimensional Irreducible Dimension, Phys. Rev. Lett. 119, 080401 (2017).
Entangled States with Complementary, Random Filtering, Phys. [21] T. Kraft, C. Ritz, N. Brunner, M. Huber, and O. Gühne,
Rev. X 6, 021018 (2016). Characterizing Genuine Multi-Level Entanglement, Phys. Rev.
[12] A. Tiranov, S. Designolle, E. Z. Cruzeiro, J. Lavoie, N. Brunner, Lett. 120, 060502 (2018).
M. Afzelius, M. Huber, and N. Gisin, Quantification of multi- [22] B. M. Terhal, Bell inequalities and the separability criterion,
dimensional entanglement stored in a crystal, Phys. Rev. A 96, Phys. Lett. 271, 319 (2000).
040303(R) (2017). [23] M. Barbieri, F. De Martine, G. Di Nepi, and P. Mataloni, De-
[13] A. Martin, T. Guerreiro, A. Tiranov, S. Designolle, F. Fröwis, tection of Entanglement with Polarized Photons: Experimental
N. Brunner, M. Huber, and N. Gisin, Quantifying Photonic Realization of an Entanglement Witness, Phys. Rev. Lett. 91,
High-Dimensional Entanglement, Phys. Rev. Lett. 118, 110501 227901 (2003).
(2017). [24] O. Gühne, P. Hyllus, D. Bruß, A. Ekert, M. Levenstein, C.
[14] P. Erker, M. Krenn, and M. Huber, Quantifying high dimensional Macchiavello, and A. Sanpera, Detection of entanglement with
entanglement with two mutually unbiased bases, Quantum 1, 22 few local measurements, Phys. Rev. A 66, 062305 (2002).
(2017). [25] R. Fickler, R. Lapkiewicz, M. Huber, M. P. Lavery, M. J. Padgett,
[15] A. Sanpera, D. Bruß, and M. Lewenstein, Schmidt-number and A. Zeilinger, Interface between path and orbital angular mo-
witnesses and bound entanglement, Phys. Rev. A 63, 050301(R) mentum entanglement for high-dimensional photonic quantum
(2001). information, Nat. Commun. 5, 4502 (2014).
[16] G. Sentís, C. Eltschka, O. Gühne, M. Huber, and J. [26] M. Malik, M. Erhard, M. Huber, M. Krenn, R. Fickler, and
Siewert, Quantifying Entanglement of Maximal Dimension A. Zeilinger, Multi-photon entanglement in high dimensions,
in Bipartite Mixed States, Phys. Rev. Lett. 117, 190502 Nat. Photon. 10, 248 (2016).
(2016). [27] A. Fedrizzi, T. Herbst, A. Poppe, T. Jennewein, and A. Zeilinger,
[17] A. C. Dada, J. Leach, G. S. Buller, M. J. Padgett, and E. Ander- A wavelength-tunable fiber-coupled source of narrowband en-
sson, Experimental high-dimensional two-photon entanglement tangled photons, Opt. Express 15, 15377 (2007).
and violations of generalized Bell inequalities, Nat. Phys. 7, 677 [28] Here the fidelity is estimated by the presented data, and we have
(2011). not done a complete state tomography.

062309-5

You might also like