SCIENCE ADVANCES | RESEARCH ARTICLE

OPTICS Copyright © 2022

The Authors, some
Anyonic-parity-time symmetry in rights reserved;
exclusive licensee
complex-coupled lasers American Association
for the Advancement
of Science. No claim to
Geva Arwas*†, Sagie Gadasi†, Igor Gershenzon, Asher Friesem, Nir Davidson, Oren Raz original U.S. Government
Works. Distributed
Non-Hermitian Hamiltonians, and particularly parity-time (PT) and anti-PT symmetric Hamiltonians, play an under a Creative
important role in many branches of physics, from quantum mechanics to optical systems and acoustics. Both the Commons Attribution
PT and anti-PT symmetries are specific instances of a broader class known as anyonic-PT symmetry, where the NonCommercial
Hamiltonian and the PT operator satisfy a generalized commutation relation. Here, we study theoretically these License 4.0 (CC BY-NC).
novel symmetries and demonstrate them experimentally in coupled lasers systems. We resort to complex coupling
of mixed dispersive and dissipative nature, which allows unprecedented control on the location in parameter space
where the symmetry and symmetry breaking occur. Moreover, tuning the coupling in the same physical system
allows us to realize the special cases of PT and anti-PT symmetries. In a more general perspective, we present and
experimentally validate a new relation between laser synchronization and the symmetry of the underlying
non-Hermitian Hamiltonian.

INTRODUCTION studied, where the symmetry breaking was used to generate a robust

Downloaded from https://www.science.org on April 24, 2024

Hamiltonian quantum theory provides an excellent description of single-mode operation (27, 28), reversing the pump dependence
isolated and closed systems. The Hermiticity of the Hamiltonian (29) or loss-induced lasing (30). A closely related concept that was
operator assures that the probability flows only between the various recently demonstrated in (31–33) is anti-PT symmetry. Optical
states of the system. To model open systems, which can exchange anti-PT systems offer additional methods to control light that can
probability and heat with their environment, non-Hermitian effec- be used, for example, to generate a refractionless propagation.
tive Hamiltonians are commonly used (1, 2). These non-Hermitian Here, we generalize the PT and anti-PT symmetries into an
Hamiltonians can display complex spectra and nonorthogonal ei- anyonic-PT. The general anyonic-PT and the two special cases of
genmodes, leading to several unusual properties. A notable exam- PT and anti-PT symmetries are all experimentally realized in a single
ple is the spectral degeneracy known as exceptional point (EP) physical system of two coupled lasers in a degenerate cavity. The
(3, 4), where two or more eigenmodes coalesce. Additional im- specific type of anyonic-PT symmetry is tuned by controlling the
portant examples are several novel symmetries that can appear phase of the complex coupling between the lasers (34). We both ex-
only in non-Hermitian Hamiltonians. In particular, parity-time (PT) perimentally and theoretically show how the symmetry is manifested,
symmetric systems (5) have gained much interest because of their and broken, along a line in the lasers’ relative loss and frequency
unique and counterintuitive properties. These Hamiltonians can parameter space. Furthermore, by taking into account the nonlinear-
have a real valued spectrum even when non-­Hermitian. This prop- ity of the laser system, we relate our results to the physics of syn-
erty, which is often referred to as “pseudo-­Hermiticity” (6), is not chronization and show how the various non-Hermitian symmetries
limited to PT symmetric systems and plays an important role in are manifested.
complex extensions of quantum theory (7, 8). Anyonic-PT symmetric Hamiltonians satisfy
As many concepts that originated from non-Hermitian physics,
the PT symmetry was first developed in the context of quantum me- PTℋ = ​e​​  −2i​  ℋPT​
​ (1)
chanics but turned out to be a powerful tool in several other branches
of physics, including electronic (9), acoustic (10, 11), and particularly where P and T are the parity and time reversal operators, and  is a
optics and photonics (12–15). In optical systems, the parity operator real constant. The specific cases  = 0 and  =  correspond to the
spatially reflects the system, and the “time reversal” operation inter- standard PT symmetry, where the Hamiltonian satisfy [PT, H] = 0.
changes the gain with the loss. The PT symmetry then emerges The cases  = ± /2 correspond to the anti-PT symmetry, where the
from a gain-loss balance (16–19), which can keep the Hamiltonian Hamiltonian anticommutes with the PT operation, {PT, H} = 0. The
invariant under the combined operation. This offers a novel control (anti-)PT commutation relations resemble the famous (fermionic)
over the light’s spatial profile and the device’s transmission proper- bosonic commutation relations. In this spirit, the general case was
ties (20–22). A PT symmetric system can be in two different phases, recently named anyonic-PT symmetry (35, 36). It should be clear
with a sharp symmetry breaking (17, 23) transition between the two. that this terminology does not suggest the existence of anyonic qua-
In the vicinity of the symmetry breaking, the system shows extreme siparticles in the system but rather a formal analogy with the com-
sensitivity to small perturbations, a highly promising feature for mutation relation Eq. 1.
sensing applications (24–26). PT symmetric lasers have been extensively The anyonic-PT symmetry is a novel way to manipulate light. For
example, we demonstrate how it can be used to control the phases
and the intensities of two coupled lasers, which can have arbitrary
Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot losses or frequencies. Furthermore, our methods enable to manipu-
7610001, Israel.
*Corresponding author. Email: arwasgeva@gmail.com late the location of the EPs in parameter space. In this way, the sys-
†These authors contributed equally to this work. tem can be tuned to the vicinity of an EP, without having to modify

Arwas et al., Sci. Adv. 8, eabm7454 (2022) 1 June 2022 1 of 8

SCIENCE ADVANCES | RESEARCH ARTICLE

the frequencies or the losses, as required in the case of PT symmet-  ​​

​tan  = ​ ─ (4)
ric systems. 
Non-Hermitian physics also suggests a new perspective on syn-
chronization. A common wisdom (37) states that a stable unique For a fixed , this corresponds to a tilted straight line in the (,
synchronized state for two-phase oscillators with different frequen- ) parameter space, as schematically shown in Fig. 1A. We refer
cies can be obtained only through a dissipative coupling mechanism to this line as the anyonic-PT symmetry line. To show how the sym-
(38, 39) and that their frequency detuning must not exceed the metry is manifested in the spectrum of the system, the eigenvalues
strength of this coupling. Contrary to the expectation, we show how, and (non-normalized) eigenmodes of H are written as
by controlling the lasers’ loss, synchronization is possible for larger
[ ​   ​e​​  i ​,
​ ]
___________ T
z + ​​  ±​​
detuning and even when the coupling is purely dispersive. The allowed ​​​  ±​​  = ± i ​√− ​​​  2​ ​e​​  2i​ − ​z​​  2​ ​, ​V±​  ​​  = ​​ ​​ ─ 1​ ​​​​  ​​ (5)
frequency range follows a universal relation that we validate experi-
mentally. We show that, for all coupling types, the desynchroniza- The real and imaginary parts of ± correspond (up to a constant) to
tion generally results from a pseudo-Hermiticity symmetry. the frequency and the loss of the eigenmodes. + and V+ represent
the less lossy eigenmode. In Fig. 1C, we plot theoretical regime dia-
grams for the relative phase  = 1 − 2 and intensity I​ ​​r​​  = (​A21​ ​ − ​ ​  A22​ ​ ​  ) /
2 2 ii
RESULTS (​A1​ ​ + ​
​  A2​ ​)​  ​of the two laser fields Ei = Aie , when the system is in the
Although our analysis can be straightforwardly generalized to many V+ mode. In each image,  is fixed, and a (, ) regime diagram
other systems, here, we focus on the specific experimental system (both in units of the coupling magnitude ) is plotted. The different
used in what follows. It is composed of two coupled lasers with values of  correspond to a dispersive, complex, or dissipative coupling.
complex electric fields E1 and E2, relative loss  = (1 − 2)/2 and First, we consider the symmetry line that is visible for all values
frequency detuning  = (1 − 2)/2 (see Fig. 1). In each cavity of  (see Fig. 1A). In the dispersive case ( = ; Fig. 1C, left) the well-

Downloaded from https://www.science.org on April 24, 2024

round trip, part of the light from each laser is coupled, in a symmetric known PT symmetry line (13) is located at  = 0. Along this line,
manner, into the other laser (see Fig. 2 and Materials and Methods, in the so-called “unbroken” or “exact” regime ∣∣ < , the two
“Experimental setup” section). The evolution of the coupled laser lasers maintain equal intensity, while only their relative phase is changed.
fields is given by the laser rate equations (see Materials and In the “broken” regime ∣∣ > , one laser becomes stronger, while
Methods, “Laser rate equations” section) (40), which we write in a their relative phase is fixed at ±/2. Here, the Hamiltonian and the
vectorial form PT operator do not share the same eigenmodes, although the oper-
ators commute. At the symmetry breaking points, ∣∣ = ± , the
​ dE ​ = [iG(E, t ) + ​​  ​​  − i ​​  ​​  + ℋ ] E​
i ​ ─ (2) eigenvalues ±, as well as the eigenmodes V±, coalesce. This type of
0 0
dt
non-Hermitian degeneracy is known as the EP.
where E = [E1, E2]T, G(E, t) is a diagonal matrix that represents the In the dissipative case ( = − /2; Fig. 1C, right), the anti-PT
nonlinear gain, and 0 = (1 + 2)/2 and 0 = (1 + 2)/2 are scalar con- symmetry line is located at  = 0, where ∣ ∣ <  and ∣∣ > 
stants representing the lasers’ average loss and frequency, respectively. correspond to the unbroken and broken phases. Here, the EPs are
H is an effective traceless Hamiltonian matrix, parametrized by located at  = ± .
The new anyonic-PT symmetry is presented in the middle panel of

( ​e​​  ​ − z )
i
Fig. 1C. In this case, the coupling is complex-valued with  = −0.88,
​​ℋ = ​ ​​​  z​  i​   ​e​​  ​​​​ ​​​​ (3)
and the anyonic-PT symmetry line is -rotated. As in the PT
symmetric case, the two lasers have equal amplitudes along the
where z =  − i. The frequency and the loss of each laser (rela- anyonic-PT symmetry line in the unbroken phase, and the symmetry
tive to their average values) are the real and imaginary parts of the breaking occurs at the EPs, located at (, ) = ± (cos , sin ).
Hamiltonian’s diagonal, while the off-diagonal terms describe the Beyond the symmetry lines, all phase diagrams reveal a rich struc-
complex coupling between the lasers, obtained by calculating a spa- ture. In particular, they show a discontinuity jump in the phase and/
tial overlap between the two laser fields (see Materials and Methods). or intensity of the lasers across the dashed lines, which we address
Because the coupling in our system is symmetric, the two coeffi- later on. Although the anyonic-PT symmetry line is tilted by an angle 
cients are identical. The magnitude of the coupling is given by  from the PT symmetric case, the full regime diagram is not rotated
and its phase by . This type of coupling, unless purely real, breaks in a trivial way, manifesting the different physical roles of  and .
Hermiticity and generically leads to a nonconservative dynamic, When the coupling is complex, there is no symmetry associated with
even in the absence of gain or loss in the system. We therefore refer reflections around the  = 0 axis. Notably,  > 0 does not always
to real coupling ( = 0, ) as dispersive and to imaginary coupling result in Ir > 0, i.e., the more lossy laser might have a stronger intensity.
( = ± /2) as dissipative. For complex coupling, the ratio between The experimental measurements are performed for two coupled
the dispersive and dissipative parts is set by . lasers in a degenerate cavity. To control , we adjust the distance
For simplicity, we begin the analysis with an approximated linear between the center of the two laser spots (see Fig. 1B and Materials
description of the dynamics, where we replace G(E, t) by a constant and Methods). Note that, by doing so, we also change , as the mag-
value G and therefore neglect the spatial and temporal dependence nitude and phase of the coupling are not independent. Once  is
of the gain medium. Under this approximation, the steady-state fixed, we use an intracavity digital mirror to apply a phase and am-
(dE/dt = 0) solutions of Eq. 2 are given by the eigenmodes of H. We plitude mask, which allows us to control the loss and the frequency
address the full nonlinear dynamics in a later section. of each laser individually and to scan  and  values. Last, we
The Hamiltonian in Eq. 3 is anyonic-PT symmetric, i.e., it satis- measure the intensity of the two lasers and their relative phase by
fies Eq. 1, provided that (see Materials and Methods) interfering the two beams (see Materials and Methods).

Arwas et al., Sci. Adv. 8, eabm7454 (2022) 1 June 2022 2 of 8

SCIENCE ADVANCES | RESEARCH ARTICLE

Downloaded from https://www.science.org on April 24, 2024

Fig. 1. Non-Hermitian symmetries and complex coupling. (A) Schematic diagram of the anyonic-PT symmetry line. The coupling type—from dispersive to dissipative—
given by  rotates the symmetry line in the (, ) plane. The dots represent the location of the EPs, on a circle of radius , where the symmetry breaks in each case.
(B) The experimentally controllable parameters. The background shows a single shot of the two coupled lasers. The interference fringes are used to extract the
relative phase in each measurement. (C) Theoretical plots of the relative intensity Ir (top row) and phase difference of the lasers  (bottom row), as given by the V+ eigen-
vector of the effective Hamiltonian. In each panel, the coupling is fixed and a (, ) regime diagram is plotted. The different columns are for  = , − 0.88, and − /2,
corresponding to purely dispersive, complex, and purely dissipative couplings, respectively. The dots mark the location of the EP. The dashed lines represent the condi-
tion for pseudo-Hermiticity of the Hamiltonian. (D) Experimental measurements of the relative intensity 2Ir (top row) and phase difference of the lasers  (bottom row).
The different columns correspond to  ≈ 10.9,6.7, and 2.2 MHz, respectively, with approximately the same  values of (C). The phase difference values are shaded in accor-
dance to their measured phase coherence C = ∣ < ei > ∣, where black regions correspond to poor coherence and lack of synchronization. Gray areas are experimentally
inaccessible.

Fig. 2. Degenerate cavity laser arrangement. It is composed of two lenses (L1 and L2) that form a 4f telescope configuration, reflective spatial light modulator (SLM) for
forming two lasers, an output coupler, and an Nd–yttrium-aluminum-garnet gain medium that is pumped by a xenon flash lamp. Coupling between the two lasers is
achieved by displacing the output coupler a distance z. The SLM is used to apply frequency detuning and relative loss between the lasers.

In Fig. 1D, we show the experimental results of the relative predictions of the V+ mode in Fig. 1C. As opposed to a coherently
phase and intensity of the two coupled lasers. The symmetry line driven system, where different eigenmodes can be resonantly excit-
can be easily identified in each case by its unbroken phase of equal ed, for our coupled lasers, the lasing state is the winner of a mode
intensities, as well as the sharp symmetry breaking—a signature competition (41), well approximated by the lower-loss eigenmode
of an EP. We find a remarkable agreement with the theoretical (42, 43).

Arwas et al., Sci. Adv. 8, eabm7454 (2022) 1 June 2022 3 of 8

SCIENCE ADVANCES | RESEARCH ARTICLE

However, the laser system is inherently nonlinear. The diagrams

in Fig. 1D contain regions of low coherence, where the black color
indicates that the relative phase between the lasers is not well de-
fined. In these regions, the lasers do not synchronize because of the
coexistence of degenerate modes, as we confirm by nonlinear simu-
lations (see the Supplementary Materials). The lack of phase coher-
ence cannot be captured by a linear analysis but, as we show below,
results from a pseudo-Hermiticity symmetry of ℋ (along the dashed
lines in Fig. 1C). We note that, at the vicinity of the EPs, linewidth
broadening (44, 45) can also affect the phase coherence of the la-
sers (46).
We now explore the behavior along the anyonic-PT symmetry
line in more detail. Figure 3A shows the amplitude ratio along the
line for various values of , as a function of the frequency detuning.
The unbroken phase is bounded by  = ± sin , which also marks
the location of the two EPs, with a sharp symmetry breaking. A clear
plateau of equal amplitudes in the unbroken phase is shown, in per-
fect agreement with the theory. In fact, as long as the anyonic-PT
symmetry is not broken, the linear modes are also exact solutions
of the full nonlinear laser rate equations, as explained below. The
different place of symmetry breaking for each  demonstrates our

Downloaded from https://www.science.org on April 24, 2024

ability to control the location of the EPs in the (, ) parameter
space. Operating at the vicinity of the EPs enhances the sensitivity
of the system to perturbations, which may be advantageous for
sensing applications (24–26).
Along the symmetry line, from Eq. 5, it follows that

​
 = − sin sin ​ (6)

such that  varies in the unbroken phase, between the two EPs, from
/2 to −/2 and then remains constant in the broken phase, beyond
the EPs. In Fig. 3B, we plot the measured phases as a function of ,
scaled by a factor of sin . The measured , plotted here for differ-
ent , shows a similar trend upon this scaling, in agreement with
Eq. 6. The large error bars at the vicinity of the symmetry breaking
Fig. 3. The anyonic-PT symmetry line. (A) Experimental measurements of the
indicates the marked enhancement of the noise at the EPs. amplitude ratio of the lasers A1/A2 along the anyonic-PT symmetry line, as a function
We now turn to address the general structure of the regime dia- of the detuning . Each panel corresponds to a different value of , as indicated
grams. The low coherence in the relative phase shown in Fig. 1D in the figure. The anyonic-PT unbroken regime is identified by the equal-amplitude
indicates that the lasers fail to synchronize. We next show that this plateau, while in the anyonic-PT broken regime (shaded), one laser has a larger
is a manifestation of pseudo-Hermiticity. This notion refers to amplitude. The EPs, located at  = ± sin , mark the place of symmetry breaking
operators that are non-Hermitian but, nevertheless, have a pure real for each . (B) Experimental measurements of the relative phase of the two lasers
spectrum (6). Here, the condition for H to be pseudo-Hermitian, i.e., along the symmetry line. Here, the detuning is given in units of sin . Upon this
Im[−] = Im [+] = 0, is shown by dashed lines in Fig. 1C, given by scaling, the location of the EPs and the onset of the symmetry breaking are the
same for all . The different colors correspond to the same  values of (A).

​​​  2​  sin 2

​
 = ​ ─  ​​ (7) regions around the pseudo-Hermiticity lines. Pseudo-­Hermiticity
2
implies that all modes have the same loss resulting in poor phase
with 2 cos 2 − 2 + 2 > 0. For the purely dispersive and pure- coherence due to the coexistence of different lasing modes. For
ly dissipative cases, this becomes trivial straight lines, overlapping purely dispersive coupling Fig. 1D (left), this results in a unique
with the unbroken PT and the broken anti-PT phases, where the “+” shape. Here, in addition to the pseudo-Hermiticity in the hori-
eigenvalues are real. The complex coupling is different in this re- zontal unbroken-PT symmetry line, along the vertical  = 0 line,
gard: Here, Eq. 7 does not overlap with the anyonic-PT symmetry the Hamiltonian is truly Hermitian, trivially having real eigen-
line, where the spectrum is always complex. values. This + shape is extremely fragile—a slight change of 
For all coupling types, we see a sharp discontinuity jump in the from  changes the regime diagram markedly (see the Supplemen-
phase and/or intensity in the theoretical diagrams of Fig. 1C across tary Materials).
the pseudo-Hermiticity lines. Mathematically, the discontinuity can So far, we showed how the breakdown of synchronization is
be attributed to a branch cut of the complex ± and V± functions. In linked to the symmetry of the underlying effective linear Hamiltonian.
the experimental results in Fig. 1D, the nonlinearity and the noise in We now extend the analysis to include the nonlinear effects in the
the system broaden this discontinuity and generate low-­coherence system. With these, we can quantify the conditions for synchronization

Arwas et al., Sci. Adv. 8, eabm7454 (2022) 1 June 2022 4 of 8

SCIENCE ADVANCES | RESEARCH ARTICLE

and understand how the nonlinearity affects the non-Hermitian

symmetries of the Hamiltonian.
The nonlinearity in the coupled laser system results from the
term G(E, t) in Eq. 2, which describes the dynamics of the gain me-
dium. In a steady state, if one exists, then the gain of each laser takes
a constant value of Gi = P/(1 + ∣Ei∣2), where P is the pump strength
and ∣Ei∣2 is the laser intensity, normalized to its saturation value
(see Materials and Methods). In the anyonic-PT unbroken phase,
both lasers have the same amplitude, yielding G1 = G2. Hence, the
solutions of Eq. 2 are exact eigenmodes of H (47, 48), and the
anyonic-PT unbroken phase is not affected by the nonlinearity.
The pseudo-Hermiticity symmetry line, however, is not robust in this
sense and is expected to vary with different strengths of nonlinearity
in the system.
In the broken symmetry phase, or away from the symmetry line,
the lasers do not necessary synchronize or even lase at all. The con-
dition for a synchronized steady state is that the system converges to
a stable fixed point of Eq. 2, namely, that the two laser fields have
fixed intensities and oscillate in synchrony, with a “locked” relative
phase. The dynamic of the phase difference between the lasers is
given by (see Materials and Methods)

Downloaded from https://www.science.org on April 24, 2024

[ ​A2​  ​​ ]
d ​A​  ​​ ​A​  ​​
​​​ ─ ​ = − 2 + ​ ​​ ​ ─1 ​  cos ( +  ) − ​ ─2 ​  cos ( −  ) ​ ​​​​ (8)
dt ​A1​  ​​
Note that this relation only depends on the amplitude ratio. The
loss  and the gain G(E, t) do not appear here. Therefore, it also
applies to many different coupled oscillatory systems and other types
of nonlinearities. In Fig. 4, we show the (A1/A2, ) regime dia-
gram of the measured synchronization for the same values of  as
in Fig. 1.
Let us first discuss synchronization with equal amplitudes A1 =
A2 (red line in Fig. 4). Here, the condition d/dt = 0 in Eq. 8 reduces Fig. 4. Synchronization regions. The different panels correspond to  = , − 0.88,
to Eq. 6 and coincides with the phase along the anyonic-PT symme- and − /2, as in Fig. 1D. The black lines are the theoretical synchronization criteria
try line. Hence, equal amplitude synchronization is possible only if of Eq. 9, while the scattered points show the experimental data. The color rep-
resents the measured phase coherence, where the yellow color indicates perfect
∣∣ is smaller than sin , the dissipative part of the coupling.
synchronization. Each data point is the average of 10 measurements. The white
Therefore, the anyonic-PT symmetry must be unbroken. For purely points show the  value of the EPs, which border the frequency range for uniform
dissipative coupling, this reduces to the well-known condition for amplitude synchronization.
synchronization ∣ ∣ <  (37), and the border coincides with the
anti-PT symmetric EPs (38). Conversely, when the dissipative part
vanishes, as in the PT symmetric case, uniform amplitude synchro-
nization is impossible. DISCUSSION
In the case of unequal amplitudes, Eq. 8 implies that syn- To conclude, we demonstrated and investigated anyonic-PT
chronization is possible, in principle, for every  and . The symmetries using a degenerate cavity laser. We provided detailed
synchronization condition in the region is (see Materials and regime diagrams for the case of two coupled lasers, showing how
Methods) the novel anyonic-PT symmetry, as well as the special cases of PT
and anti-PT symmetries, is manifested. At the heart of our experi-

2 [ ( ​A2​  ​​ ) ( ​A1​  ​​ ) ]
1/2
 ​ ​​ ​​ ​​ ​​─ ​A​  ​​ 2 ​A​  ​​ 2 ments is the ability to control the nature of the coupling—from
​∣∣≤ ​ ─ ​  1 ​​ ​​​​  ​  + ​​ ​​─
​  2 ​​ ​​​​  ​ − 2cos 2​ ​​​​  ​​ (9)
purely dispersive to purely dissipative. The mixed case, where the
coupling is complex, presents a rich structure that can be advanta-
indicated by the black lines in Fig. 4. For all values of , we see a geous for future applications. The presence of a robust symmetry
good agreement between the regions of high coherence and the line in parameter space, which depends on both the frequency and
synchronization criteria of Eq. 9. In particular, larger frequency the relative loss between the lasers, can be used for calibration, for
detunings require larger amplitude ratio for the lasers to syn- sensing applications, or to detect a frequency loss–correlated noise.
chronize. For purely dispersive coupling, we get a unique linear Furthermore, by controlling the coupling, one can control the loca-
behavior at the origin. Paradoxically, here, it is more difficult to tion of the EPs in parameter space.
synchronize the lasers when they have the same frequency. This An interesting direction for future research is the possibility to
is because for dispersive coupling with  = 0, the PT symmetry, dynamically alter the coupling type. For instance, by dynamically
until broken, forces equal amplitudes, preventing the lasers from changing the coupling from purely dispersive to purely dissipa-
synchronizing. tive, a PT symmetric system can be transformed into an anti-PT

Arwas et al., Sci. Adv. 8, eabm7454 (2022) 1 June 2022 5 of 8

SCIENCE ADVANCES | RESEARCH ARTICLE

symmetric system continuously. In our degenerate laser cavity, this Detection arrangement
could be achieved by dynamically changing the distance between One mirror serves as an output coupler, where light emerges and
the lasers. Furthermore, by dynamically changing the coupling type, propagates to the detection arrangement. The measurement of the
the EP itself can move along a circle around a fixed point in pa- lasers’ amplitudes and relative phase is carried out by using an
rameter space, rather than encircling the EP, which is of recent interferometer, schematically depicted in fig. S1. In one arm of the
interest (49). interferometer, one laser is selected and expanded using a pinhole
While synchronization is inherently a nonlinear phenomenon, and a lens to serve as a reference field. In the second arm, the laser
we demonstrated how the non-Hermitian (linear) framework can field on the SLM is imaged by a 4f telescope. The light from both
provide valuable insight. The lasers fail to synchronize because of arms is then recombined on a complementary metal-oxide semi-
the reality of the spectrum’s pseudo-Hermiticity symmetry, as all conductor detector with a small angle, resulting in interference fringes
modes have equal loss. In the more familiar case of dispersive cou- on top of each laser field. Figure 1B shows a typical interference
pling, the pseudo-Hermiticity and the PT symmetries overlap. How- image. Each data point is averaged over 10 measurements.
ever, when the coupling is complex, we found an intriguing structure,
where the reality of the spectrum is along hyperbolic lines in the regime Laser rate equations
diagrams. The relation between synchronization and pseudo-­ Consider an array of many coupled lasers. The dynamics of the laser
Hermiticity is not limited to just two lasers. This opens an arena for field and the gain medium is given by (40)

​​  p ​​​​ [​​(​Gm ​  ​​  − ​∑​ ​​ ​​  mn​​ ​En​  ​​​]​​​​

future study of complex band structures in laser lattices (50). In this ​dE​  m​​ 1
case, the pseudo-Hermiticity symmetry can be seen as an imaginary ​​​ ─  ​  = ​ ─ ​  ​​  − ​​  m​​  − i ​​  m​​  ) ​Em (10)
dt n
flat-band analog (51).
c[ ( ​Is​  at​​ )]
​dG​  m​​ ​∣​Em
1  ​​ ​​ ​P​  ​​  − ​G​  ​​​ ​​1 + ​ ─ ​  ​​∣​​  2​
​​​ ─  ​  = ​ ─
​
 ​  ​​ m m  ​​ ​​​ ​​​​ (11)
dt

Downloaded from https://www.science.org on April 24, 2024

MATERIALS AND METHODS where m, m, and Pm are the loss, frequency, and pump of each
Experimental setup laser; nm is the coupling matrix; and p, c, and Isat are the cavity
A schematic diagram of the experimental arrangement is presented round trip time, the gain medium fluorescence lifetime, and the sat-
in Fig. 2 (see the Supplementary Materials for a detailed version). It uration intensity. We write the equations in a dimensionless form
consists of two lenses that form a 4f telescope configuration, a gain by rescaling the units of the electric field and time by setting Isat = 1
medium, reflective spatial light modulator (SLM), and an output and p = 1. For our symmetric coupling, we can write 12 = 21 ≡
coupler, with a total propagation length of 5 m per round trip. The iei. Using 1,2 = 0 ±  and 1,2 = 0 ± , Eq. 10 takes the form
arrangement is essentially a perfect degenerate cavity laser (34), of Eq. 2 for the two lasers, with H given by Eq. 3. The initiation of
where the field of each point at one end (the input) maps onto itself lasing typically begins with a relaxation oscillations period (41).
after a complete roundtrip. The gain medium is a doped Nd–yttrium-­ When (and if) the system reaches a synchronized steady state, the
aluminum-garnet rod of 10-mm diameter and 11-cm length that is gain (Eq. 11) takes the value Gm = Pm/(1 + ∣Ei∣2). For a pump
pumped by a xenon flash lamp, generating a quasi–continuous-­ strength that is slightly above the lasing threshold value, the inten-
wave laser pulse of 200-s duration. The reflective SLM (43) at the sity is weak, and we can approximate the steady state gain by Gm ≈
input plane is used as a digital mask in which two holes of diameter Pm. If we also assume uniform pumping Pm = P, then the gain terms
300 m are imprinted. The field of each hole matches that of a single in Eq. 2 are proportional to the unit matrix, and the fixed points can
spatial Gaussian mode with waist of w0 ≈ 150 m (see the Supple- be approximated by the eigenstates of H. When both lasers have the
mentary Materials), so the field of the holes can be viewed as that same intensity, the eigenstates of H are exact solutions, because
from two independent lasers. The SLM can thus be used to control G1 = G2, irrespective of the pump strength or the intensity.
the relative loss and frequency detuning of the two lasers. Coupling
between the lasers is obtained by shifting the output coupler from Anyonic PT symmetry condition
the exact imaging plane (more details below). Here, we show explicitly the condition for our effective Hamiltoni-
an to have anyonic-PT symmetry. To avoid confusion, here, we
~
Coupling mechanism replace  by ​​ ​​in the anyonic commutation in Eq. 1. We define the
The coupling between the lasers is generated by shifting the out- time reversal and parity operators in the standard way (1), such
put coupler by z = 10 cm from the exact imaging plane. As a that T performs complex conjugation and the parity operator here
result, a portion of the light from each laser leaks to the other is given by
laser (see Fig. 2). The complex-valued coupling coefficients nm
(by convention Hnm = − inm) are given by the normalized over- ​​P = ​(0​​​ ​  1)
​​​ ​​​​ (12)
1 0
lap integral of the expanded field of one laser with that of the
other one at the imaging plane (see the Supplementary Mate- In addition, we have P 2 = 1, T 2 = 1, and [P, T] = 0. We can re-
rials for explicit derivation). As the field propagation from one write Eq. 1 as
laser to the other is symmetric, so are the coupling coefficients. ˜ ˜
The effective Hamiltonian is therefore a complex-symmetric ℋ = ​PTe​​  −2i​ ​​  ℋPT = ​e​​  −2i​ ​​  Pℋ * P​
​ (13)
matrix, while generally not Hermitian, i.e., Hnm = Hmn but​​
*
ℋ​nm  ​ ≠ ​
​ ℋ ​  mn​​​. In our experiments, we vary the coupling by For the Hamiltonian of Eq. 3, we have
changing the distance between the lasers (reflective spots on the
( ​e​​  ​ − z ) ( ​e​​  ​ ​z​​  *​ )
i
SLM), so as to generate a dispersive, dissipative, or complex cou- ˜ ​​  *​​  ​   ​e​​  −i
​​​ ​​​  z​  i​   ​e​​  ​​​​ ​​  = ​e​​  2i​ ​​​ ​​​  − ​z−i  ​​​​ ​​​​ (14)
pling between the lasers.

Arwas et al., Sci. Adv. 8, eabm7454 (2022) 1 June 2022 6 of 8

SCIENCE ADVANCES | RESEARCH ARTICLE
~
The off-diagonals give e​​ ​​ i​ = ​e​​  i(2​ ​−)​​, which means that we must 9. J. Schindler, A. Li, M. C. Zheng, F. M. Ellis, T. Kottos, Experimental study of active LRC
~ ~ circuits with PT symmetries. Phys. Rev. A 84, 040101 (2011).
have ​ ​​  = ​ (optionally, ​​​​  and  can differ by , but this makes no
10. X. Zhu, H. Ramezani, C. Shi, J. Zhu, X. Zhang, PT-symmetric acoustics. Phys. Rev. X 4,
difference), as in our original~definition in Eq. 1. The diagonal terms
031042 (2014).
give z = − z*e2i or R ​ e [ z ​e​​  −i​ ​​ ] = 0​, so that, with z =  − i, we 11. R. Fleury, D. Sounas, A. Alu, An invisible acoustic sensor based on parity-time symmetry.
obtain the anyonic-PT symmetry line (Eq. 4). Nat. Commun. 6, 5905 (2015).
The anyonic-PT symmetry can also be understood as a general- 12. L. Feng, R. El-Ganainy, L. Ge, Non-hermitian photonics based on parity–time symmetry.
ization of the more familiar PT symmetric case. Every anyonic-PT Nat. Photonics 11, 752–762 (2017).
13. Ş. Özdemir, S. Rotter, F. Nori, L. Yang, Parity–time symmetry and exceptional points
symmetric Hamiltonian can be written as H = eiHPT, where HPT in photonics. Nat. Mater. 18, 783–798 (2019).
is PT symmetric. This is because Eq. 1 is formally equivalent to 14. M.-A. Miri, A. Alù, Exceptional points in optics and photonics. Science 363, eaar7709
[PT, e−iH] = 0. The celebrated PT symmetric two-site system [see, (2019).
e.g., (2)] is given in our notation by 15. S. Klaiman, U. Günther, N. Moiseyev, Visualization of branch points in PT-symmetric
waveguides. Phys. Rev. Lett. 101, 080402 (2008).
16. C. E. Rüter, K. G. Makris, R. El-Ganainy, D. N. Christodoulides, M. Segev, D. Kip, Observation
​​​ℋ​  PT​​  = ​(iΔα
κ − iΔα)
​​​ ​  ​  κ​ ​​ ​​​​ (15) of parity–time symmetry in optics. Nat. Phys. 6, 192–195 (2010).
17. A. Guo, G. Salamo, D. Duchesne, R. Morandotti, M. Volatier-Ravat, V. Aimez, G. Siviloglou,
D. Christodoulides, Observation of PT-symmetry breaking in complex optical potentials.
Phys. Rev. Lett. 103, 093902 (2009).
which is a special case of Eq. 3 with purely real coupling ( = 0) and
18. R. El-Ganainy, K. Makris, D. Christodoulides, Z. H. Musslimani, Theory of coupled optical
vanishing frequency detuning ( = 0). When multiplying by the PT-symmetric structures. Opt. Lett. 32, 2632–2634 (2007).
phase factor ei, the coupling becomes complex as required, while 19. Y. D. Chong, L. Ge, A. D. Stone, PT-symmetry breaking and laser-absorber modes
the line  = 0 is rotated in the (, ) plane. in optical scattering systems. Phys. Rev. Lett. 106, 093902 (2011).
20. Z. Lin, H. Ramezani, T. Eichelkraut, T. Kottos, H. Cao, D. N. Christodoulides, Unidirectional
invisibility induced by PT-symmetric periodic structures. Phys. Rev. Lett. 106, 213901 (2011).
Synchronization borders

Downloaded from https://www.science.org on April 24, 2024

21. K. G. Makris, R. El-Ganainy, D. Christodoulides, Z. H. Musslimani, Beam dynamics in PT
To derive the synchronization regimes of Fig. 4, we first rewrite symmetric optical lattices. Phys. Rev. Lett. 100, 103904 (2008).
Eq. 10 in terms of the amplitudes and phases using Ei = Aieii to 22. S. Longhi, PT-symmetric laser absorber. Phys. Rev. A 82, 031801 (2010).
obtain Eq. 8 for the dynamic of the phase difference . Frequency 23. B. Peng, Ş. K. Özdemir, F. Lei, F. Monifi, M. Gianfreda, G. L. Long, S. Fan, F. Nori,
synchronization essentially means that the lasers are phase-locked, C. M. Bender, L. Yang, Parity–time-symmetric whispering-gallery microcavities. Nat. Phys.
10, 394–398 (2014).
i.e., d/dt = 0. The steady-state  depends on the amplitude ratio—, 24. J. Wiersig, Enhancing the sensitivity of frequency and energy splitting detection by using
, and , within some allowed frequency range. To find the bor- exceptional points: Application to microcavity sensors for single-particle detection.
der  values for which synchronization is possible, we maximize Phys. Rev. Lett. 112, 203901 (2014).
the last two terms of Eq. 8 with respect to . Taking the  derivative 25. W. Chen, Ş. K. Özdemir, G. Zhao, J. Wiersig, L. Yang, Exceptional points enhance sensing
in an optical microcavity. Nature 548, 192–196 (2017).
gives ​​A​12​ ​  sin ( +  ) = ​A2​2​ ​  sin ( − )​, which solves for the relative
26. M. P. Hokmabadi, A. Schumer, D. N. Christodoulides, M. Khajavikhan, Non-hermitian ring
phase at the synchronization border. Using this condition and laser gyroscopes with enhanced sagnac sensitivity. Nature 576, 70–74 (2019).
Eq. 8, we find that the maximal ()2 is given by 27. L. Feng, Z. J. Wong, R.-M. Ma, Y. Wang, X. Zhang, Single-mode laser by parity-time

​ ​ ​​[​​ ​ ─1 ​  cos ( +  ) − ​ ─2 ​  cos ( −  ) ​]​​​​  ​

2 symmetry breaking. Science 346, 972–975 (2014).
​A​  ​​ ​A​  ​​
​ ax​
m 28. H. Hodaei, M.-A. Miri, M. Heinrich, D. N. Christodoulides, M. Khajavikhan, Parity-time–
 ​A2​  ​​ ​A1​  ​​
​​     ​​​ (16) symmetric microring lasers. Science 346, 975–978 (2014).

( ​A2​  ​​ ) ( ​A1​  ​​ )
2
​A​  ​​ ​A​  ​​ 2 29. M. Brandstetter, M. Liertzer, C. Deutsch, P. Klang, J. Schöberl, H. E. Türeci, G. Strasser,
= ​​ ​​─ ​  1 ​​ ​​​​  ​  + ​​ ​​─
​  2 ​​ ​​​​  ​ − 2cos 2 K. Unterrainer, S. Rotter, Reversing the pump dependence of a laser at an exceptional
point. Nat. Commun. 5, 4034 (2014).
30. B. Peng, Ş. Özdemir, S. Rotter, H. Yilmaz, M. Liertzer, F. Monifi, C. Bender, F. Nori, L. Yang,
Hence, synchronization is possible in the frequency detuning Loss-induced suppression and revival of lasing. Science 346, 328–332 (2014).
range Eq. 9. 31. P. Peng, W. Cao, C. Shen, W. Qu, J. Wen, L. Jiang, Y. Xiao, Anti-parity–time symmetry
with flying atoms. Nat. Phys. 12, 1139–1145 (2016).
32. Y. Choi, C. Hahn, J. W. Yoon, S. H. Song, Observation of an anti-PT-symmetric exceptional
SUPPLEMENTARY MATERIALS point and energy-difference conserving dynamics in electrical circuit resonators.
Supplementary material for this article is available at https://science.org/doi/10.1126/ Nat. Commun. 9, 2182 (2018).
sciadv.abm7454 33. Y. Li, Y.-G. Peng, L. Han, M.-A. Miri, W. Li, M. Xiao, X.-F. Zhu, J. Zhao, A. Alù, S. Fan,
C.-W. Qiu, Anti–parity-time symmetry in diffusive systems. Science 364, 170–173 (2019).
34. H. Cao, R. Chriki, S. Bittner, A. A. Friesem, N. Davidson, Complex lasers with controllable
REFERENCES AND NOTES coherence. Nat. Rev. Phys. 1, 156–168 (2019).
1. N. Moiseyev, Non-Hermitian Quantum Mechanics (Cambridge Univ. Press, 2011). 35. S. Longhi, E. Pinotti, Anyonic PT symmetry, drifting potentials and non-hermitian
2. C. M. Bender, Making sense of non-hermitian hamiltonians. Rep. Prog. Phys. 70, 947–1018 delocalization. EPL 125, 10006 (2019).
(2007). 36. Y.-P. Gao, Y. Sun, X.-F. Liu, T.-J. Wang, C. Wang, Parity-time-anyonic coupled resonators
3. M. V. Berry, Physics of nonhermitian degeneracies. Czechoslovak J. Phys. 54, 1039–1047 system with tunable exceptional points. IEEE Access 7, 107874–107878 (2019).
(2004). 37. A. Pikovsky, J. Kurths, M. Rosenblum, J. Kurths, Synchronization: A Universal Concept in
4. W. Heiss, The physics of exceptional points. J. Phys. A Math. Theor. 45, 444016 (2012). Nonlinear Sciences (Cambridge Univ. Press, 2003), vol. 12.
5. C. M. Bender, S. Boettcher, Real spectra in non-hermitian hamiltonians having PT 38. J. Ding, I. Belykh, A. Marandi, M.-A. Miri, Dispersive versus dissipative coupling
symmetry. Phys. Rev. Lett. 80, 5243–5246 (1998). for frequency synchronization in lasers. Phys. Rev. Appl. 12, 054039 (2019).
6. A. Mostafazadeh, Pseudo-hermiticity versus PT symmetry: The necessary condition 39. L. Bello, M. C. Strinati, E. G. Dalla Torre, A. Pe’er, Persistent coherent beating in coupled
for the reality of the spectrum of a non-hermitian hamiltonian. J. Math. Phys. 43, 205–214 parametric oscillators. Phys. Rev. Lett. 123, 083901 (2019).
(2002). 40. F. Rogister, K. S. Thornburg Jr., L. Fabiny, M. Möller, R. Roy, Power-law spatial correlations
7. C. M. Bender, D. C. Brody, H. F. Jones, Complex extension of quantum mechanics. Phys. in arrays of locally coupled lasers. Phys. Rev. Lett. 92, 093905 (2004).
Rev. Lett. 89, 270401 (2002). 41. A. Siegman, Lasers (University Science Books, 1986).
8. I. Rotter, A non-hermitian hamilton operator and the physics of open quantum systems. 42. C. Tradonsky, I. Gershenzon, V. Pal, R. Chriki, A. Friesem, O. Raz, N. Davidson, Rapid laser
J. Phys. A Math. Theor. 42, 153001 (2009). solver for the phase retrieval problem. Sci. Adv. 5, eaax4530 (2019).

Arwas et al., Sci. Adv. 8, eabm7454 (2022) 1 June 2022 7 of 8

SCIENCE ADVANCES | RESEARCH ARTICLE

43. I. Gershenzon, G. Arwas, S. Gadasi, C. Tradonsky, A. Friesem, O. Raz, N. Davidson, Exact 53. V. Evtuhov, A. E. Siegman, A “twisted-mode” technique for obtaining axially uniform
mapping between a laser network loss rate and the classical XY hamiltonian by laser loss energy density in a laser cavity. Appl. Optics 4, 142–143 (1965).
control. Nanophotonics 9, 4117–4126 (2020). 54. J. A. Arnaud, Degenerate optical cavities. Appl. Optics 8, 189–195 (1969).
44. Y. D. Chong, A. D. Stone, General linewidth formula for steady-state multimode lasing 55. C. Tradonsky, V. Pal, R. Chriki, N. Davidson, A. A. Friesem, Talbot diffraction and fourier
in arbitrary cavities. Phys. Rev. Lett. 109, 063902 (2012). filtering for phase locking an array of lasers. Appl. Optics 56, A126–A132 (2017).
45. J. Zhang, B. Peng, Ş. K. Özdemir, K. Pichler, D. O. Krimer, G. Zhao, F. Nori, Y.-x. Liu, S. Rotter,
L. Yang, A phonon laser operating at an exceptional point. Nat. Photonics 12, 479–484 Acknowledgments: We thank S. Rotter, N. Moiseyev, A. Pick, and E. Poem for useful
(2018). discussions. Funding: This work was supported by the Israel Science Foundation
46. M. Fridman, V. Eckhouse, N. Davidson, A. A. Friesem, Effect of quantum noise on coupled (grant numbers 1881/17 and 950/19). O.R. receives the support of the Shlomo and Michla
laser oscillators. Phys. Rev. A 77, 061803 (2008). Tomarin career development chair, the Abramson Family Center for Young Scientists, and the
47. L. Ge, R. El-Ganainy, Nonlinear modal interactions in parity-time (PT) symmetric lasers. Minerva Foundation, with funding from the Federal German Ministry for Education and
Sci. Rep. 6, 24889 (2016). Research. Author contributions: G.A. and S.G. conceived the project. S.G. performed the
48. M. H. Teimourpour, M. Khajavikhan, D. N. Christodoulides, R. El-Ganainy, experiment and analyzed the data with assistance from I.G. All authors discussed the results.
Robustness and mode selectivity in parity-time (PT) symmetric lasers. Sci. Rep. 7, G.A. performed the theoretical analysis and wrote the manuscript with contribution from all
10756 (2017). authors. A.F., N.D., and O.R. supervised the project. Competing interests: The authors declare
49. J. Doppler, A. A. Mailybaev, J. Böhm, U. Kuhl, A. Girschik, F. Libisch, T. J. Milburn, P. Rabl, that they have no competing interests. Data and materials availability: All data needed to
N. Moiseyev, S. Rotter, Dynamically encircling an exceptional point for asymmetric mode evaluate the conclusions in the paper are present in the paper and/or the Supplementary
switching. Nature 537, 76–79 (2016). Materials.
50. M. Nixon, E. Ronen, A. A. Friesem, N. Davidson, Observing geometric frustration
with thousands of coupled lasers. Phys. Rev. Lett. 110, 184102 (2013).
51. D. Leykam, A. Andreanov, S. Flach, Artificial flat band systems: From lattice models Submitted 19 October 2021
to experiments. Adv. Phys. X 3, 1473052 (2018). Accepted 13 April 2022
52. S. Ngcobo, I. Litvin, L. Burger, A. Forbes, A digital laser for on-demand laser modes. Published 1 June 2022
Nat. Commun. 4, 2289 (2013). 10.1126/sciadv.abm7454

Downloaded from https://www.science.org on April 24, 2024

Arwas et al., Sci. Adv. 8, eabm7454 (2022) 1 June 2022 8 of 8