Manipulation of Quantum Strings by Edge Defect in Kagome Rydberg Atom Array

Wei Xu1 and Xue-Feng Zhang1, 2, ∗

1
Department of Physics, and Chongqing Key Laboratory for Strongly Coupled Physics, Chongqing University, Chongqing, 401331, China
2
Center of Quantum Materials and Devices, Chongqing University, Chongqing 401331, China
Optical tweezers can grab particles with their laser beam fingers, so these tiny point objects can be moved by
using the laser light. However, manipulation of topological defects, especially higher-dimension ones becomes
extremely difficult because they usually require non-local operation involving a great deal of operators, e.g.
string operator. Here, using a large-scale quantum Monte Carlo method, we demonstrate the quantum string
excitation can be grabbed in the Kagome Rydberg atom array with well-designed edges. After inserting an edge
arXiv:2410.21135v1 [cond-mat.quant-gas] 28 Oct 2024

defect, the fractional charge at one end of the string is found to be pinned, allowing the quantum string to be
manipulated. Since the quantum string is directed, its quantum fluctuations can be clearly observed due to the
movement of the dangling fractional charge at the other end. Then, we can drag the two ends of the string in
opposite directions, which is analogous to the separation of a quark-anti-quark pair connected by a gluon string.
As the separation distance increases, the quantum fluctuations of the string are strongly suppressed. Eventually,
a string-breaking phenomenon occurs. However, if both edge defects are located on the same side, the strings
are always broken because both dangling fractional charges take the same gauge charge and can not annihilate
each other into the vacuum. This work not only proposes an ideal experimental platform for analyzing quantum
string excitations in the ground state but also provides a much simpler method for manipulating quantum string
excitations in frustrated systems, such as spin ice.

Introduction— Field lines can depict the fundamental inter- (a) 𝑸=𝟎: 𝑸=-𝟏: 𝑸=𝟏:
(c)
actions between particles, e.g., electric field lines between a
positron and an electron and the gluon string between quarks.
On the other hand, due to the interplay between geometry
frustration and the strong interaction between spins, a line-
type structure, also known as a ’string,’ can emerge in quan-
tum frustrated magnetism[1–11]. Interactions between the
strings can result in various exotic topological phases of mat- (b)
C
ter, such as the incommensurate phase [4, 5], “Luttinger string 𝟎 𝝈− −𝟏 𝟎 𝝈+ +𝟏
A B
liquid” [3], and U(1) lattice gauge theory (LGT) [12, 13]. Un- 𝟎 +𝟏 +𝟏 𝟎
like the classical Dirac string in spin ice [8, 9], here the strings
are quantum, meaning the ground state of the string is a super-
position of all possible fluctuating string structures, which can FIG. 1. The schematic diagram of (a) different gauge charges and
be considered as the worldsheet in 2+1 dimensions. Although fields (thick red arrow: 32 ; thin red arrows: 13 ), (b) the effect of σ x ,
the quantum string can be indirectly observed by analyzing and (c) the quantum string (connected red arrows) pinned by the edge
the dynamical spectrum [3, 6], direct detection is still chal- defect (hand). The red arrows in (c) mark the “electric field” with
lenging. value one. The red triangle marks the gauge charge, and the blue
circle labels the ring exchange process.
As a typical frustrated system, the Kagome lattice is com-
posed of a corner-shared structure [6, 13–20]. The antiferro-
magnetic Ising interaction can impose a geometric constraint array is still under debate [32, 33], but the flexibility of the
analogous to the ice rule on the system [21], so that the dis- geometry and in situ manipulation make it possible to prepare
ordered ground state preserves macroscopic degeneracy. If classical string structures in the experiment [34]. However,
quantum fluctuations are introduced, higher-order perturba- searching for the manipulation method of quantum strings in
tive interactions can lift the degeneracy, so the system en- the ground state with more flexibility is still a critical task.
ters into the valence bond solid phase [15, 16, 22, 23], which In this manuscript, we study the Kagome Rydberg atom ar-
breaks the translational symmetry. Due to the strong lo- ray with open boundary conditions mimicking the real experi-
cal constraint, the elementary excitations should be fractional ment. Due to the long-range repulsive forces, the ground state
charges with quantum strings attached [6, 20]. Both fractional becomes the stripe phase in which the fractional charges are
charges and quantum strings are found to play a critical role in confined. As shown in Fig. 1, after kicking out one site at the
deconfined quantum criticality [6, 24, 25], and its low-energy boundary, a “directed” quantum string spontaneously emerges
physics can be understood by LGT [6, 13, 18, 19], which in- with one side attached to the edge defect. Then, by further
volves the U(1) lattice gauge field strongly coupled to mat- kicking out another site on the opposite side, both ends of
ter fields. Thanks to the rapid development of the Rydberg the quantum string are tweezed, so that we can easily stretch
atom array [26–31], the Kagome geometry can be easily im- the string until it breaks. The fractional charges at both bro-
plemented. The ground state of the Kagome Rydberg atom ken ends take different charges, just like the positive-negative
 2

electron pair creation from the vacuum. Before breaking, the (a) (b)
quantum string exhibits the asymptotic freedom-like behavior.
Model— The system we consider is the Kagome Rydberg
atom array, in which two-level atoms are individually trapped
by optical tweezer lights. In each site, the ground state |g⟩
can be coupled to the Rydberg state |r⟩ via a two-photon pro-
cess with site-independent tunable Rabi frequency Ωi and de-
tuning ∆i . Atoms in the Rydberg state feel the long-range
repulsive interaction with a Van der Waals (VdW) type form
FIG. 2. (a) The density distribution of the atom in the Rydberg state
Vi j = V/R6i j , and the strength of V can be adjusted by changing (black dot) and the resonant configuration (red hexagon), and (b) the
the lattice spacing. Then, the Hamiltonian can be expressed as snapshot obtained from the QMC simulation and then subtracted the
follows: reference state. The blue dashed lines mark the possible region of the
X X X Ωi string’s fluctuation. The number of lattice sites is 561 with LY = 6.
H= Vi j ni n j − ∆i ni − σix , (1)
i< j i i
2

where σix = |g⟩i ⟨r|i + |r⟩i ⟨g|i denotes the exciting process be- In the spin-1/2 XXZ model, the spin exchange or XY inter-
tween the ground state and Rydberg state and ni = |r⟩i ⟨r|i action provides the quantum fluctuation. In the LGT, it repre-
is the density operator of atom in Rydberg state. The model sents the production of the gauge charge pair or the hopping
above can be mapped into the extended quantum transverse of the gauge charge at the next nearest neighbor dual-sites. In
Ising model by implementing the following transformation comparison, as demonstrated in Fig. 1(b), the Rabi term σix
ni ↔ (σzi + 1)/2. Then, the VdW potential term plays the role has the same effects but at the nearest neighbor site in the dual
of Ising interaction, so the Rydberg blockade effect imposes lattice. Both interactions keep the charge conservation. On
the local constraint. Meanwhile, the Rabi frequency and de- the other hand, a sixth-order perturbation can contribute to the
tuning correspond to the transverse and longitudinal magnetic ring exchange interaction ( ↔ ) in each hexagon without
field. When taking ∆ = V with lattice space a0 set to one, the breaking the triangle rule [22, 23]. Therefore, the string can
leading contribution of the diagonal term can be rewritten as gain kinetic energy from the movement of the gauge charge at
V P P 2 the end and also from the fluctuation of the main body.
2 k ( i∈△k ni − 1) which indicates only one atom in each tri-
angle can be excited to the Rydberg state. Such a strong local As presented in Fig. 1(c), if we kick out one site at one side,
constraint is called the triangle rule which causes the macro- it is equivalent to inserting an empty down-triangle or negative
scopic degeneracy. However, if further long-range repulsive gauge charge . Then, it can move into the bulk with a string
interactions are considered, the degeneracy can also be lifted attached, but the other end of the string is always pinned at
so that the ground state becomes the stripe phase in which the the edge defect, just like the dog leash. As the string length in-
atoms in one of three sublattices are fully excited (see Fig.1). creases, the stripe order is disrupted more severely. Therefore,
Quantum String—To interpret the excitation of fractional the string has a tension energy due to the long-range interac-
charges and quantum strings, it is better to first introduce the tions. If the Rabi frequency is small, the fluctuation energy of
lattice gauge mapping in brief. Same as the spin-1/2 XXZ the string cannot overcome the tension energy, so the gauge
model [6, 13, 17–20], as shown in Fig. 1(a), the “electric charge is confined. In contrast, with a larger Rabi frequency,
field” can be defined as Ell′ = ni − 1/3 at the bisector of the deconfinement is expected. In the following numerical simu-
corner-shared triangle, where l labels the triangles that com- lation, we set Ω = 0.36 to ensure that the gauge charges are
pose the honeycomb dual-lattice. The positive direction is set confined in the bulk, and the string attaching to the edge defect
as pointing from the down triangle to the up triangle. Then, can emerge with a finite length at the same time.
we can immediately find that the triangle rule is indeed the Results—The QMC method we utilized is the newly de-
“Gauss’s law”, and the violated configurations have different veloped high-performance large-scale loop algorithm [35],
charges calculated by summing over all field lines around one which can help achieve a large system size. To mimic the
triangle (as listed in Fig. 1(a)). Therefore, there are four dif- real experiment, we take open boundary conditions in both
ferent fractional excitations (gauge charge) classified by posi- directions. The shape of the edge is designed as shown in
tive or negative charge and up- or down-triangle. Meanwhile, Fig. 1(c) so that the outermost sites are fully occupied due to
their gap energies are approximately equal at ∆ = V and can the edge effect. The nearest-neighbor repulsive interaction V
be tuned by changing the detuning strength [6]. Furthermore, and the detuning ∆ are set to one, and the VdW long-range in-
after mapping, we can find that the stripe phase is very special teractions are truncated to the third nearest-neighbor site. The
because it represents a uniform “electric field” in the language maximum system size achieved reaches N = 561 atom array
of LGT. To explicitly observe the quantum string, we should sites. The temperature is set to β = 1/T = 100, so the ground
take it as the reference vacuum state. As shown in Fig. 1(c), state physics can be observed. All the results are calculated
after subtracted by the reference state, the non-zero electric by taking the average of 128 independent Markov chains with
field forms a single string with zero electric field in the back- 106 samples in each.
ground. Without the edge defect, the ground state is a perfect stripe
 3

(a) (b) (c) (a) (b)

(d) (e) (f)

(c) (d)
0.03

0.02 Asymptotic
𝛿𝐸 0.01 freedom
FIG. 3. (a-c) The density distribution of the atom in the Rydberg
state (black dot) and the resonant configuration (red hexagon) for String
0.00
different defect distances. The schematic pictures of quantum string breaking
excitations are presented in (d-f), respectively. 0 2 4 6
𝛿𝑥

FIG. 4. (a-b) The density distribution of the atom in the Rydberg

phase that can be detected by measuring the local density of state (black dot) and the resonant configuration (red hexagon) for dif-
the atom in the Rydberg state ⟨ni ⟩. In the stripe phase, all the ferent defect distances. (c) Schematic pictures of the straight quan-
atoms in the C sublattice are excited to the Rydberg state, ac- tum string corresponding to (a). (d) The energy difference of stretch-
companied by local quantum fluctuations. Then, we can turn ing by one more defect distance calculated by QMC.
off one edge site by changing the local detuning to a large neg-
ative value to ensure the atom always remains in state |g⟩. To
highlight the quantum string excitation, we subtract it by the the string are pinned, and the string vibration can be clearly
reference stripe state, which means the density in the C sub- observed in the bulk, which resembles the vibrating strings in
lattice is transformed as ni → 1 − ni . Furthermore, to illustrate a guitar or the ancient Chinese instrument–“Guzheng”.
the dynamics of the string, we also calculate the density of the If we change the position of both edge defects, the string
resonant hexagon [13, 20], which characterizes the strength of is stretched. As demonstrated in Fig. 3(b-c), the fluctuation
the ring exchange process. region continuously shrinks along with the increase of the dis-
As demonstrated in Fig. 2(a), the densities of both Ryd- tance between defects. Nevertheless, its shape always main-
berg excitation and resonant hexagon are higher around the tains a rectangular form. Such characteristics can be under-
edge defect. The interesting feature is that the sites in the blue stood from Fig. 3(d-f). When both ends of the string are fixed,
dashed lines labeled region have apparent high intensity, al- its length is also fixed because the string is directed, or no two
though the strength decays as the site is further away from segments of the string lie in the same row. Then, the configu-
the edge defect. To figure out the physical reason, we make a ration with the fewest resonant hexagons can be used to define
snapshot, which is similar to taking a picture of the Rydberg the boundary of the fluctuation region, e.g., the left boundary
atom distribution in the experiment. From Fig. 2(b), we can shown in Fig. 3(d-f). Consequently, it is easy to deduce how
immediately observe the string with finite length attached to the string accommodates a longer defect distance.
the edge. Meanwhile, we find that the string is directed, which When the defect distance δx increases to the width of the
means it cannot turn around to form a loop-like structure. The system Ly , the string is straightened, as shown in Fig. 4(a).
same phenomena are also observed in the XXZ model [20]. The resonant fluctuation densities appear higher along the
The main reason is that the string can only move up left or up string, but this may result from local quantum fluctuations be-
right. In addition, we can find some spots in other places, and cause the string cannot deform, as demonstrated in Fig. 4(c).
their appearance is due to local quantum fluctuations from the Furthermore, once the defect distance is further enlarged, the
vacuum which reflects the gauge charge confinement. string is overstretched. Then, we can observe that it breaks
At the other end of the quantum string, a gauge charge ran- into two strings attached to different sides, as explicitly shown
domly moves in the middle of the bulk. Then, we can imagine in Fig. 4(b).
that if an additional gauge charge appears, they will annihilate The string-breaking can be considered a sudden change in
each other if they have opposite charges. Therefore, we turn the system, so it should be reflected in the energies. Thus,
off one site on the other side to produce another string with we calculate the energies E(δx) with different defect distances
the gauge charge having the opposite charge. Remarkably, as δx. Here, to maintain the left-right symmetry of the system,
shown in Fig. 3(a), we can only observe a single string with the defect distance can only take even numbers. Then, we ob-
two ends attached to different edge defects. This strongly sup- tain the energy difference δE = E(δx +2)− E(δx) and plot it in
ports our expectation that the introduced gauge charge anni- Fig. 4(d). Before the string-breaking occurs, the energy cost
hilates the existing gauge charge in the bulk. Consequently, is continuously increasing while stretching the string, which
the two strings merge into a single string. Then, both ends of implies that the quantum string is more free when edge de-
 4

fects are close to each other. Such phenomena are reminiscent Note on related work—During the preparation of this
of the counterintuitive property of the strong force, which ex- manuscript, we noticed that a related experimental work ob-
hibits asymptotic freedom in quark confinement. served the string-breaking phenomena in the Kagome Ryd-
As a one-dimensional quantum object, the quantum string berg atom array using a different experimental scheme [34].
can usually be well described by a one-dimensional quantum In comparison, our work focuses on ground-state physics with
model, such as in triangular lattices [4, 5], Kagome lattices ten times larger system size, and study the interplay between
[6, 20], and even the high-Tc problem [36]. Here, the config- gauge charge and quantum string by the manipulation of it.
uration of the string can be mapped into an effective spin-1/2
chain by mapping the gauge field ↖ (↗) to effective spin up Acknowledgement– X.-F. Z. acknowledges funding from
(down) [20]. Then, the ring exchange interaction plays the the National Science Foundation of China under Grants No.
role of the effective spin-exchange interaction. Although the 12274046, No. 11874094, No.12147102 and No.12347101,
second and third-order long-range repulsive interactions con- Chongqing Natural Science Foundation under Grants No.
tribute to the complicated Ising interaction, the ring exchange CSTB2022NSCQ-JQX0018, Fundamental Research Funds
is dominant. Consequently, the effective model describing the for the Central Universities Grant No. 2021CDJZYJH-003,
string is approximately the spin-1/2 XY model, which can be and Xiaomi Foundation / Xiaomi Young Talents Program.
exactly solved by applying the Jordan-Wigner transformation.
Counting the spins in Fig. 3(d-f), we can find that the defect
distance can change the total magnetization as M = δx/2.
Therefore, zero magnetization corresponds to δx = 0 and ∗
corresponding author: [email protected]
also to half-filled Jordan-Wigner fermions, which preserve the [1] P. Fulde and F. Pollmann, Ann. Phys. (Berlin) 17, 441 (2008).
lowest energy. Then, stretching the string is equivalent to in- [2] Y. Jiang and T. Emig, Phys. Rev. Lett. 94, 110604 (2005).
serting additional Jordan-Wigner free fermions into a higher [3] Z. Zhou, C. Liu, Z. Yan, Y. Chen, and X.-F. Zhang, npj Quan-
Fermi energy band, so the energy cost becomes larger and tum Materials 7, 60 (2022), arXiv:2010.01750 [cond-mat.str-
larger. Thus, it exhibits asymptotic freedom-like behavior. el].
[4] Z. Zhou, C. Liu , D.-X. Liu, Z. Yan, Y. Chen , and X.-
Furthermore, we also attempt to place both edge defects on
F. Zhang, SciPost Physics 14, 037 (2023), arXiv:2005.11133
the same side in the numerical simulation. Then, we observed [cond-mat.str-el].
each defect is attached to a string. If the defects are close to [5] X.-F. Zhang, S. Hu, A. Pelster, and S. Eggert, Phys. Rev. Lett.
each other, the fluctuation of the strings becomes asymmet- 117, 193201 (2016).
ric, which implies there are interactions between the strings. [6] D.-X. Liu, Z. Xiong, Y. Xu, and X.-F. Zhang, Phys. Rev. B 109,
These two strings cannot merge into one string because the L140404 (2024), arXiv:2301.12864 [cond-mat.str-el].
fractional charges at the free ends of both strings take the same [7] Z. Xiong, Phys. Rev. B 108, 195107 (2023).
[8] D. J. P. Morris, D. A. Tennant, S. A. Grigera, B. Klemke,
charge.
C. Castelnovo, R. Moessner, C. Czternasty, M. Meissner, K. C.
Conclusion and Outlook—We successfully pin the quan- Rule, J.-U. Hoffmann, K. Kiefer, S. Gerischer, D. Slobinsky,
tum string at the boundary by introducing an edge defect in the and R. S. Perry, Science 326, 411 (2009).
Kagome Rydberg atom array, allowing us to manipulate it ef- [9] C. Castelnovo, R. Moessner, and S. Sondhi, Nature 451, 42
fectively. With the help of an additional edge defect, both ends (2008).
of the string can be grabbed tightly, enabling us to stretch this [10] E. Mengotti, L. J. Heyderman, A. F. Rodrı́guez, F. Nolting, R. V.
one-dimensional quantum curve smoothly. Although the en- Hügli, and H.-B. Braun, Nature Physics 7, 68 (2011).
[11] Y. Wan and O. Tchernyshyov, Phys. Rev. Lett. 108, 247210
tire process is not dynamic but simulated in the ground state, it
(2012).
can still be considered as following a quasi-static path. After [12] Z. Zhou, Z. Yan, C. Liu, Y. Chen, and X.-F. Zhang, arXiv
measuring the density distribution of the Rydberg excitation e-prints , arXiv:2212.10863 (2022), arXiv:2212.10863 [quant-
and resonant hexagons, we find that the string is directed, and ph].
its fluctuation is confined to a rectangular region. Finally, by [13] X.-F. Zhang, Y.-C. He, S. Eggert, R. Moessner, and F. Poll-
examining the variation of the energy, we observe asymptotic mann, Phys. Rev. Lett. 120, 115702 (2018), arXiv:1706.05414
freedom phenomena before string-breaking occurs, which can [cond-mat.str-el].
[14] T.-H. Han, J. S. Helton, S. Chu, D. G. Nocera, J. A. Rodriguez-
be well understood by the free Jordan-Wigner fermionic the- Rivera, C. Broholm, and Y. S. Lee, Nature (London) 492, 406
ory. (2012), arXiv:1307.5047 [cond-mat.str-el].
As mentioned before, in the real experiment, it is easy [15] K. Damle and T. Senthil, Phys. Rev. Lett. 97, 067202 (2006).
to kick out a site by changing the detuning. Definitely, the [16] S. V. Isakov, S. Wessel, R. G. Melko, K. Sengupta, and Y. B.
tweezer light at that site can also be switched off. However, Kim, Phys. Rev. Lett. 97, 147202 (2006).
considering the possibility of dynamically moving the string [17] Y.-C. He, D. N. Sheng, and Y. Chen, Phys. Rev. Lett. 112,
137202 (2014).
step by step in future experiments, the previous scheme would
[18] Y.-C. He and Y. Chen, Phys. Rev. Lett. 114, 037201 (2015).
be better. In addition, annealing to the ground state would not [19] Y.-C. He, S. Bhattacharjee, F. Pollmann, and R. Moessner,
be easy due to the topological features of the string, so search- Phys. Rev. Lett. 115, 267209 (2015).
ing for a better annealing method would also be an interesting [20] X.-F. Zhang and S. Eggert, Phys. Rev. Lett. 111, 147201 (2013),
direction [37, 38]. arXiv:1305.0003 [cond-mat.quant-gas].
 5

[21] L. Pauling, Journal of the American Chemical Society 57, 2680 [31] P. Scholl, M. Schuler, H. J. Williams, A. A. Eberharter,
(1935). D. Barredo, K.-N. Schymik, V. Lienhard, L.-P. Henry, T. C.
[22] R. Moessner, S. L. Sondhi, and P. Chandra, Phys. Rev. Lett. 84, Lang, T. Lahaye, A. M. Läuchli, and A. Browaeys, Nature
4457 (2000). (London) 595, 233 (2021), arXiv:2012.12268 [quant-ph].
[23] S. V. Isakov and R. Moessner, Phys. Rev. B 68, 104409 (2003). [32] R. Samajdar, W. W. Ho, H. Pichler, M. D. Lukin, and
[24] T. Senthil, A. Vishwanath, L. Balents, S. Sachdev, and M. P. A. S. Sachdev, Proceedings of the National Academy of Science
Fisher, Science 303, 1490 (2004). 118, e2015785118 (2021), arXiv:2011.12295 [cond-mat.quant-
[25] T. Senthil, L. Balents, S. Sachdev, A. Vishwanath, and M. P. A. gas].
Fisher, Phys. Rev. B 70, 144407 (2004). [33] Z. Yan, Y.-C. Wang, R. Samajdar, S. Sachdev, and Z. Y. Meng,
[26] A. Browaeys and T. Lahaye, NATURE PHYSICS 16, 132 Phys. Rev. Lett. 130, 206501 (2023), arXiv:2301.07127 [cond-
(2020). mat.quant-gas].
[27] F. Nogrette, H. Labuhn, S. Ravets, D. Barredo, L. Béguin, [34] D. Gonzalez-Cuadra, M. Hamdan, T. V. Zache, B. Braver-
A. Vernier, T. Lahaye, and A. Browaeys, Phys. Rev. X 4, man, M. Kornjaca, A. Lukin, S. H. Cantu, F. Liu, S.-T. Wang,
021034 (2014). A. Keesling, M. D. Lukin, P. Zoller, and A. Bylinskii, arXiv
[28] H. Bernien, S. Schwartz, A. Keesling, H. Levine, A. Omran, e-prints , arXiv:2410.16558 (2024), arXiv:2410.16558 [quant-
H. Pichler, S. Choi, A. S. Zibrov, M. Endres, M. Greiner, ph].
V. Vuletić, and M. D. Lukin, Nature (London) 551, 579 (2017), [35] W. Xu and X.-F. Zhang, arXiv e-prints , arXiv:2409.17835
arXiv:1707.04344 [quant-ph]. (2024), arXiv:2409.17835 [cond-mat.str-el].
[29] G. Semeghini, H. Levine, A. Keesling, S. Ebadi, T. T. Wang, [36] J.-L. Wang, S.-J. Hu, and X.-F. Zhang, arXiv e-prints ,
D. Bluvstein, R. Verresen, H. Pichler, M. Kalinowski, R. Sama- arXiv:2406.01980 (2024), arXiv:2406.01980 [cond-mat.str-el].
jdar, A. Omran, S. Sachdev, A. Vishwanath, M. Greiner, [37] Z. Yan, Z. Zhou, Y.-H. Zhou, Y.-C. Wang, X. Qiu, Z. Y.
V. Vuletić, and M. D. Lukin, Science 374, 1242 (2021), Meng, and X.-F. Zhang, npj Quantum Information 9, 89 (2023),
arXiv:2104.04119 [quant-ph]. arXiv:2105.07134 [quant-ph].
[30] S. Ebadi, T. T. Wang, H. Levine, A. Keesling, G. Semeghini, [38] Y.-M. Ding, Y.-C. Wang, S.-X. Zhang, and Z. Yan, Physical
A. Omran, D. Bluvstein, R. Samajdar, H. Pichler, W. W. Ho, Review Applied 22, 034031 (2024), arXiv:2310.04291 [quant-
S. Choi, S. Sachdev, M. Greiner, V. Vuletić, and M. D. Lukin, ph].
Nature (London) 595, 227 (2021), arXiv:2012.12281 [quant-
ph].