be extended to high-dimensional case, such as Abrams–

OPEN Lloyd’s two-qubit gate. By discussing the practicability, we

[Link]/scientificreports expect applications and experimental implementations in
the near future.

Introducition
Quantum operators1 take a basic role in quantum
information processing as building blocks of quantum
circuits and communication encoding. Series of quantum
operations are able to undertake computing tasks2–4,
evolve physical systems1,5 and simulate quantum

Universal quantum
phenomena6,7. A class of unitary quantum operators is
investigated early on for reasons. One reason is that
time-evolutions are naturally unitary in the conventional

simulation of
quantum mechanics. Another reason may due to the
success of Grover’s square-root-accelerated searching
algorithm4 which strengthens the viewpoint that the core

single-qubit
of a algorithm is to design the step-by-step unitary
evolution of the system8. Besides, it is sufficient to apply
series of unitary operations to simulate a variety of
Hermitian systems and related phenomena5, leading
nonunitary scientists to pay more attentions on the products of
unitary operators.
However, unitary operators and their products are not

operators using enough to everything. On one hand, it is not easy to

design new breakthrough quantum algorithms only by
multiplying unitary operations. For example, it is believed

duality quantum that NP-complete problems cannot be solved in

polynomial time by unitary quantum computing, while the
nonunitary operation (such as Abrams–Lloyd’s gate) is

algorithm possible9,10. On the other hand, the requirement of Her

miticity is unnecessary for some novel physical
systems11–16, of which the time-evolution operators are no
longer unitary. Besides, evolutions of open quantum
Chao Zheng systems or dissipative-quantum systems are nonunitary,
which is important in modeling the system-environment
Quantum information processing enhances human’s power interaction. Therefore, it is necessary to introduce
to simulate nature in quantum level and solve complex nonunitary operators into quantum computing and
problem efficiently. During the process, a series of simulation.
operators is performed to evolve the system or undertake Quantum simulation provides an efficient and effective
17
a computing task. In recent year, research interest in non- way to investigate nature by itself . A Hermitian system
Hermitian quantum systems, dissipative-quantum systems can be simulated by the product of a series of unitary
operators1,5,18–20, whereas it is neither suitable for non-
and new quantum algorithms has greatly increased, which
Hermitian systems21–30 nor for dissipative-quantum
nonunitary operators take an important role in. In this systems31–35 of which the evolutions are non-unitary.
work, we utilize the linear combination of unitaries Several efforts have been made to simulate dissipative-
technique for nonunitary dynamics on a single qubit to give quantum systems. For examples, Barreiro et al. experi
explicit decompositions of the necessary unitaries, and mentally implement open-system dynamics through the
simulate arbitrary time-dependent single-qubit nonunitary dissipative map32; Hu et al. propose and demonstrate a
operator F(t) using duality quantum algorithm. We find general quantum algorithm to evolve open quantum by
that the successful probability is not only decided by F(t) simulating Kraus maps33; Rost et al. simulate condensed
34
and the initial state, but also is inversely proportional to matter system ; Viyuela et al. prepare topological
thermal states to simulate a topological insulator open
the dimensions of the used ancillary Hilbert subspace. In a system
general case, the simulation can be achieved in both eight-
and six-dimensional Hilbert spaces. In phase matching
conditions, F(t) can be simulated by only two qubits. We Department of Physics, College of Science, North China
illustrate our method by simulating typical non-Hermitian University of Technology, Beijing 100144, People’s Republic of
systems and single-qubit measurements. Our method can China. email: czheng@[Link]

Scientific Reports | (2021) 11:3960 | [Link]

1
 Content courtesy of Springer Nature, terms of use apply. Rights
reserved
Vol.:(0123456789)
[Link]/scientificreports/

for the topological-Uhlmann-phase measurement35. In recent years,

non-Hermitian quantum mechanics and related systems are
investigated massively. Typical ones among them are PT-
symmetric, anti-PT-symmetric, pseudo-Hermitian, anti-pseudo-
Hermitian systems, and etc. Motivations to investigate non-
Hermitian systems include extending quantum theory, studying
open quantum systems, discovering novel properties and applica
tions. For example, Hermiticity has been seen to be a necessary
condition for a long time to ensure quantum mechanics physically
since it keeps the energy eigenvalues real. However, it is found to
be a sufficient but not essential one for the reality of observables.
The first attractive extension is PT-symmetric quantum
mechanics11–13, and then pseudo-Hermiticity is pointed out to be a
sufficient and necessary condition keeping the spectrum of a
Hamiltonian purely real14–16. Subsequently, both the two classes of
non-Hermitian systems36–61 and their anti
symmetric counterparts62–72 are investigated theoretically and
experimentally to develop non-Hermitian theory and discover
appealing features. Therefore, it is meaningful to investigate how to
realize nonunitary operators in a controllable Hermitian system, so
that non-Hermitian systems can be simulated on both small
quantum devices and near-term quantum computers.
In this work, we utilize the linear combination of unitaries (LCU)
technique73–79 for non-unitary dynamics on a single qubit, and
simulate single-qubit nonunitary operators in subspaces of different-
dimensional Hilbert spaces. The nonunitary operator F(t) can be
i Ht
a time-evolution operator e−( / ) of a non-Hermitian two-state
system, a single-qubit measurement, and etc. Unitary expansion
(UE) of the operator is the key to our method, and we decide four
phase matching conditions in which F(t) can be
simulated by only two qubits. If none of the conditions is met, three
qubits are necessary. Our method simulates nonunitary operators in
an indeterministic way. We find that the successful probability is
not only affected by F(t) and the initial state of the system, but
also inversely proportional to the dimensions of the total used
space. We apply our method to nonunitary time evolution operators
and an arbitrary single-qubit measurement that can be seen as a
nonlinear operation. We also show how to extend our method to
multi-qubit nonunitary operators by simulating the Abrams–Lloyd’s
gate. Finally we analyze the complexities and discuss experimental
implementations in NMR and quantum optical systems.

Unitary expansions of a single-qubit nonunitary operator

In this section, we investigate unitary expansions of an arbitrary
single-qubit nonunitary operator F(t). A unitary expansion with
less terms saves qubit resource and increases the successful
probability as we will show in the next section. Therefore, we aim
at expanding F(t) by less unitary terms and give the criteria.
We write the most general form of a single-qubit nonunitary operator as
f11(t) f12(t)
,
F(t) = (1)
f21(t) f22(t)

of which each element can be a function of time t. F(t) can be

i Ht
the time-evolution operator e−( / ) of a non Hermitian system, a
nonunitary operator applied to a non-Hermitian or open system, and
etc. F(t) may any operator that shrinks, preserves or enlarges the
norm of a quantum-state vector.
In general, F(t) can be expanded by four terms of Pauli matrices σ0, σ1, σ2 and σ3
as

F(t) = f0σ0 + f1σ1 + f2(iσ2) + f3σ3, (2)

where the four UE-parameters are

 f0 = f0(t) = f11 + f22

2 , f1 = f1(t) = f12 + f21

2 , f2 = f2(t) = f12 − f21

2and f3 = f3(t) = f11 − f22

2.
(3)
Noticing that the fk’s (k = 0, 1, 2, 3) are time-dependent complex functions, we rewrite
them as
iθ
fk = fk(t) = (4) fk e k

for convenience, where fk and θk are the norm and phase angle
that may change with time t. The explicit forms of Pauli matrices
are shown below
10 01 0−i .
10
σ0 = (5)
01 , σ1 = 10 , σ2 = i0 , σ3 = 0−1

Three UE-terms. In a general case, we find that F(t) can be

expressed by three UE-terms as
iθ iθ3
F(t) = e (6) 0 f0 σ0 + e |c1|
iθ3
U1 + e |c2|U2.

The three UE-parameters are time-dependent and can be

expressed as complex functions of fk’s (k = 0, 1, 2, 3) in Eq. (3),
which the details are presented in the Supplementary Information.
fk’s can be arbitrary complex functions with no limits on the
norms.

Two UE-terms. The number of UE-terms of F(t) can be reduced

further when the phase angles θk’s of the UE-parameters fk(t)’s
(k = 0, 1, 2, 3) satisfy some conditions, which we call them
phase matching conditions. We find four phase
matching conditions, of which the details are presented in
the Supplementary Information. Whichever condition is met, F(t)
can be expressed by two UE-terms as

Scientific Reports | (2021) 11:3960 | [Link]

2
Content courtesy of Springer Nature, terms of use apply. Rights
reserved
Vol:.(1234567890)
[Link]/scientificreports/

Figure 1. Quantum circuit to simulate F(t) by a qubit-qudit

hybrid system. The system is initialized to |0�a|0�e, and the
 circuit is read from left to right. The qubit e can be prepared in an
arbitrary state |ψ�e by Rψ. First,
a single qudit rotation UF assigns the UE-parameters, and then four controlled operations generate
the four
UE-terms. Notice that the first C0−σ0 is unnecessary in practice
but only for showing our theory clearly. Next, a Hadamard in
SU(4) superposes the four UE-terms. Finally, a measurement is
performed on the ancillary qudit to obtain the output |0�a
probabilistically, in which case the nonunitary F ( t) is
successfully simulated.

F(t) = a0V0 + a1V1, (7)

where ak’s and the elements of Vk (k = 0 and 1) are complex

functions of time t; V0 and V1 are in SU(2). The explicit forms of
a0, a1, V0 and V1, varying in different conditions, are give in the
Supplementary Information.

Duality quantum simulation

For the nonunitary operator F(t), it cannot be simulated only by
one qubit in a two-dimensional Hilbert space as a unitary case. We
simulate F(t) in a two-dimensional subspace of a larger Hilbert
space using duality quantum algorithm 73, which enables linear
combinations of unitary operations.
Duality quantum algorithm was proposed in 2002 73 for the first
time and developed fast74–77. Because of the abilities to realize both
the products and linear combinations of unitary operations, it has
become one of the strongest tool in designing quantum
algorithms78. For example, an algorithm based on linear
combinations of unitary operations to simulate Hamiltonian
dynamics in a closed quantum system79 takes advantages over that
based on product formulas. Recently, scientists apply it to design a
full quantum algorithm for quantum chemistry simulation80 .
Based on our unitary expansions in the “Unitary expansions of a
single-qubit nonunitary operator” and duality quantum algorithm,
we investigate quantum simulation of F(t) in a general case and in
phase match- ing conditions using different-
dimensional Hilbert spaces. The hybrid-system protocols show
principle of our method clearly, while pure-qubit protocols enable
experimental implementations on both small quantum devices and
near-term quantum computers.

Eight-dimensional protocol. Recall Eq. (2), F(t) can be

simulated in an eight-dimensional Hilbert space by both a qubit-
qudit hybrid system and three qubits. Although this is not the most
efficient protocol, it is clear to show the principle and how the
unitary expansion of F(t) is linked to the protocol.

Using a qubit–qudit hybrid system. The

hybrid system consists of a work qubit e and an ancillary four-
dimen sional qudit a, constructing the eight-dimensional Hilbert
space. A qudit is a basic building block of high dimensional
quantum computers1,81. A four-dimensional qudit has four
orthogonal logical bases |0�, |1�, |2� and |3�, which can be
realized by a four-state quantum system such as an ultra-cold atom
with four non-degenerate energy levels, a nuclear spin with four
split-levels, a spin-(3/2) particle, and etc. In some quantum
algorithm, qudits take advantages over qubits. For example, it
reaches a higher accuracy to solve the eigenvalue problem using
quantum phase estimation algorithm by qudits82 than by qubits1 (on
page 217–226).
The quantum circuit is shown in Fig. 1. At the begining, the
whole system is initialized to a pure state |0�a|0�e, and then the
work qubit is prepared in a state |ψ� e as needed by a single qubit
rotation Rψ ∈ SU(2). Now we construct the nonunitary operation
F(t), operating the work qubit with the assistance of the ancillary
qudit. First,
 a single-qudit operator UF = ujk ∈ SO(4) (where ujk are the
matrix element and k, j = 1, 2, 3, 4) is applied to the ancillary
qudit. The explicit form of UF is not unique as long as U†FUF =
I4 and the first column vector is
T f fk 2

f0 f1 f2 f3 , where f =
3

1 (8)
k=0

is a normalizing factor. The rest column vectors can be obtained by

the Gram–Schmidt orthogonalization. The effect of UF is to assign
the UE-parameters fk’s (k = 0, 1, 2, 3) to the probabilistic
amplitudes of the ancillary qudit, which is linked to the first column
vector of UF. Second, four controlled gates follow, i.e., 0-controlled
σ0, 1-controlled σ1, 2-controlled iσ2 and 3-controlled σ3, which the
qubit and the qudit are the target and control ones. In practice, the
first controlled gate can be removed because the effect is the same
as I4. The total effect of the controlled operators is to generate the
four UE-terms in Eq. (2) and entangle them with the qudit. Third, a

Scientific Reports | (2021) 11:3960 | [Link]

3
Content courtesy of Springer Nature, terms of use apply. Rights
reserved
Vol.:(0123456789)
[Link]/scientificreports/

Figure 2. Flowchart to simulate F(t) in a quantum computer. The system is initialized in the first
block.
The second block includes UE-parameters assignment, UE-terms generation and superposition.
Finally,
measurements are performed on the ancillary subsystem to
achieve the simulation of F(t) in an indeterministic way.

Hadamard operator H4 ∈ SU(4) is applied to the ancillary qudit,

which superposes the four UE-terms. Now, the whole system
 evolves to a superposition state
. (9)
1 |0�aF(t)|ψ�e + f 3 |k�a|sk�e
k=1
2f

The explicit forms of |sk�e’s are not given since they will be
discarded if |k�a is output (k = 1, 2, 3). At last, a measurement
is performed on the ancillary qudit. If the result outputs |0�e, the
work qubit will be operated by the nonunitary in Eq. (1) to F(t)|
ψ�e (without a normalizing factor). Therefore, we simulate F(t) in
an indeterministic way with a successful probability of
1
2 (10)
e�ψ|F(t) F(t)|ψ�e.
†
4f

The normalizing factor f in Eq. (8) affects the successful probability

in Eq. (10) but doesn’t affect the effect F(t). If the ancillary qudit is
observed in one of the other three states |k�a, the work qubit will
not be operated to F(t)|ψ�e but to |sk�e (k = 1, 2, 3). We will
discard the results and re-initialize the hybrid system to the state |
0�a|0�e. The whole process above is started over until |0�a is
obtained.

Using three qubits. To apply on a small quantum

device or on a near-term quantum computer, we show how to
simulate F(t) using two ancillary qubits and one work qubit. The
flowchart of our quantum computer program is shown in Fig. 2.
There are three main blocks, i.e., system initialization, nonunitary
operation, and measurement.
The quantum circuit is shown in Fig. 3, reading from left to right.
The system is initialized to a pure state |00�a|0�e, and a single-
qubit rotation Rψ prepares the work qubit in an arbitrary state |
ψ�e. The second block is the main one, including three steps of
UE-parameters assignment, UE-terms generation and
superposition. In the first step, two single-qubit operators

1 �� � � 2  �� �
��� �
f3 �2 ��
f3 � f1 �2
f0 � + 2
 ��� � 2 
�� � f0 � + and W2 = 1 �
f1 �2  f0 − f ∗1
�� � 2
��� � 2 ��� �
f2 � + 2 f1 � ��� � �
− f2 � + f0 �2 +

W1 = (11)
f terms of use apply. Rights reserved
f1 f ∗0
are applied to the first and second ancillary qubits,
respectively. Then, a controlled gate

Scientific Reports | (2021) 11:3960 | [Link]

4
021-83521-5 Content courtesy of Springer Nature,
Vol:.(1234567890)
[Link]/scientificreports/
 Figure 3. Quantum circuit for three qubits using the full Hilbert space. The three qubits is
initialized to
|00�a|0�e, and the qubit e can be rotated to an arbitrary state |ψ�e by Rψ further. The first step is
to assign the
four UE-parameters, similar to UF in Fig. 1. Then four controlled-
controlled operators are applied to generate the four UE-terms,
where the first dashed one is unnecessary in practice. Next, two
Hadamard gates are applied to the ancillary qubits to superpose
the four UE-terms. Finally, measurements are performed on the
ancillary qudits to simulate F(t) in an indeterministic way.

I2 0

C1−W = (12)
0W

follows, where

W3 and W3 = 1 f2 − f ∗3
2·

(13) W = W
−1

f2 f3 f ∗2 .
+ f3
2 2

The above operations have a similar total effect as UF in Fig. 1,

assigning UE-parameters fk’s (k =1, 2, 3) to a superposition
state of |00�a, |01�a, |10�a and |11�a.
In the second step, four jointly-controlled gates are applied. The
first dashed C00−σ0 gate in Fig. 3 is done naturally without any
operation, while the last three controlled–controlled gates are
 σ0 000 0 σ1   σ0 000 0 σ0   σ0 000 0 σ0 
  00 , C10−iσ2 =   00 and   00 ,
  C11−σ3 = 
C01−σ1 = (14)
0 0 σ0 0 000 σ0 0 0 iσ2 0 00 0 σ0 0 0 σ0 0 000 σ3

generating the four UE-terms and entangling them with the ancillary qubits.
In the third step, two Hadamard gates operate the two ancillary
qubits, respectively. Now, the whole system evolves to a
superposition state

1  10,11 �  (15)
|00�aF(t)|ψ�e |k�a|skj�e .
2f 
+f kj=01


The UE-terms are superposed in the first term as that in Eq. (2),
while none of the rest three terms is linked to the initial state |ψ�e
by F(t).
Finally, measurements are performed on the ancillary qubits. If
the two ancillary qubits are observed in a state |00�a, the work
qubit will be operated by the nonunitary operation to a final state
F(t)|ψ�e. If either of the two ancillary qubits is measured in |
1�a, the work qubit will not be operated by F(t). The system will
be re-initialized to |00�a|0�e and the whole progress is started
over until |00�a is output. Similar to that using a hybrid system, it is
in an indeterministic way to simulate F(t), and has the same
successful probability as that in Eq. (10).
In fact, it is not necessary to expand F(t) by the Pauli operators
in our protocol. If F(t) can be expanded by other UE-terms, our
protocol is still applicable by replacing the Pauli operators with the
relevant unitary opera tors in the quantum circuit above. On one
hand, it is for convenience to illustrate by using the Pauli operators.
On the other hand, it is more important to provide a method to
reduce the number of UE-terms and thus increase the successful
probability of our method, which is a key factor that whether the
 method has actual meaning to simulate multi-qubit nonunitary
operators.

Six-dimensional protocol. Now we show how to simulate F(t) in

a six-dimensional Hilbert space based on our unitary expansion of
F(t) in Eq. (6). It takes advantages over the previous protocols in
two aspects. On one hand, it saves a two-dimensional subspace
which may be used for other task at the same time. On the other
hand, it has a higher successful probability than that using an eight-
Hilbert space. Both qubit-qutrit hybrid and pure-qubit systems are
able to achieve the simulation with the higher successful probability.

Scientific Reports | (2021) 11:3960 | [Link]

5
Content courtesy of Springer Nature, terms of use apply. Rights
reserved
Vol.:(0123456789)
[Link]/scientificreports/

Figure 4. Quantum circuit for a qubit-qutrit hybrid system. The system is initialized to |0�a|0�e,
and the
qubit e can be prepared in an arbitrary state |ψ�e by Rψ. First, a single qutrit rotation UT1 assigns
the three
UE-parameters, and then three controlled operators generate the three UE terms, where the first
dashed
C0−σ0 is not essential in practice. Next, another single-qutrit
rotation superposes the three UE terms. Finally, a measurement
is performed on the ancillary qudit to simulate the nonunitary
F(t) probabilistically.

Using a qubit- qutrit hybrid system.

We propose for the qubit–qutrit system to show the principle of this
six dimension protocol. A work qubit e and an ancillary qutrit a
compose the hybrid system and constuct a six dimensional Hilbert
space. A qutrit is a three-state quantum system, e.g., three energy
levels of an ultra-cold atom, a nuclear spin with three split-levels, a
spin-1 particle, and etc. Any superposition states of it can be com
bined by the three orthogonal logical bases |0�, |1� and |2 �.
Similar to a qudit, a qutrit is also a building block of high-
dimensional quantum computers.
The quantum circuit is shown in Fig. 4 for illustration, aiming at
simulating F(t)|ψ�e with the assistance of the ancillary qutrit. In
the first block, the whole system is prepared in a pure state |0� a|
0�e, and the work qubit is initialized to a state |ψ�e by Rψ as
needed. The second one is the main block to simulate F(t). First,
a single-qutrit operator UT1 ∈ SU(3),
 1 ��� �
 ∗ f0 �2 + |c2|2
fc∗ f0c
f 0 − 1 
�
2
�

,
|f0|2+|c2|2 2 2
|f0| +|c2|

UT1 = (16)
f |f0|2+|c2|2 c
∗1
� c2

c1 0 − c2 ff ∗ |f0|2
+|c2|
2

0
�

is applied to the ancillary qutrit to assign the UE-parameters f0, c1

and c2 in Eq. (6), where f is the renormalizing factor in Eq. (8).
Second, three controlled operators generate the three UE-terms in
Eq. (6). The first one is dashed because C0−σ0 is equal to the unit
matrix I3 in SU(3). Another two controlled operations,
σ0 0 0 0 U1 0 σ0 0 0 0 σ0 0

C1−U1 = (17)
0 0 σ0 and C2−U2 = ,
0 0 U2

act Uk on the work qubit e when the ancillary qutrit is in state |

k�a (k = 1, 2), where U1 and U2 are that in Eq. (6). At last,
a single qutrit rotation

1√ �
3

 
 3
20 −

�11 1
3
2

UT2 = (18)
1 √ 1
√ 2− 2 √ 2

is applied to the ancillary qutrit a to superpose the UE-terms. Now,

the initial pure state |0�a|ψ�e evolves to a superposition state

1√  |k�a|s′k�e  (19)
3f |0�aF(t)|ψ�e + .
 
f� k=1,2

The first term is linked to F(t), while the rest terms are not.
We measure the qubit-qutrit hybrid system now. If the ancillary
qutrit a is observed in a state |0�a, the work qubit e will evolve
to F(t)|ψ�e that entangled with |0�a. In this case, quantum
simulation of F(t) is successful. Or, if the ancillary qutrit collapses
into state |1�a or |2�a, the results of work qubit |s′1�e and |s
′
2�e will
be discarded. In these two cases, the system will be reset
to the beginning, and then quantum simulation will be restarted until
|0�a is output. Therefore, it is also in an indeterministic way to
simulate F(t), and the successful probability can be calculated as

1
2
e�ψ|F(t) F(t)|ψ�e,
†
3f
(20)

Scientific Reports | (2021) 11:3960 | [Link]

6
Content courtesy of Springer Nature, terms of use apply. Rights
reserved
Vol:.(1234567890)
[Link]/scientificreports/
 Figure 5. Quantum circuit for three qubits using a six-
dimensional subspace. The three qubits is initialized to |00�a|
0�e, and the qubit e is rotated to |ψ�e. The six-dimensional
subspace is prepared and the UE-parameters are assigned in the
first step, taking a similar role as the UT1 in Fig. 4. Then the three
UE-terms are generated by three controlled-controlled operators
in the second step, where the first dashed one is unnecessary in
practice. In the third step, a series of operations is applied to the
ancillary qubits to superpose the three UE-terms as
the effect of UT2 in Fig. 4. Finally, measurements are performed on the ancillary qudits to simulate
F(t) in a
probabilistic way.

depending on both F(t) and |ψ�e. It is 4/3 times that of Eq. (10),
meaning that the successful probability of this six-dimension
protocol is higher than that of the eight-dimension one.

Using three qubits. The six-dimension protocol still

involves three qubits to simulate F(t)|ψ�e. Instead of using the
full Hilbert space, it uses a three-dimensional subspace of the two
ancillary qubits. Figure 2 can be referred to as the flowchart of this
protocol, which includes system initialization, simulation of
nonunitary operation and measurement. Although the scheme is
similar to that of the eight-dimension protocol, they are completely
dif
ferent in detail, which can be seen from the quantum circuit in Fig. 5.
At the beginning, the three qubits are initialized to a pure state |
00�a|0�e, and Rψ rotates the work qubit to |ψ�e. The two
ancillary qubits take a similar role as the ancillary qutrit of the hybrid
system in the previous subsec tion. In the middle block, the
nonunitary F(t) is simulated by three steps: space preparation and
UE-parameters assignment, UE-terms generation, and UE-terms
superposition. In the first step, a three-dimensional subspace of the
ancillary qubits is prepared. Meanwhile, the three UE-parameters in
Eq. (6) are assigned. In detail, the two ancillary qubits are
swapped, and then two single-qubit operators R1 and −σ3 are
applied to the first and second ancillary qubits, respectively. The
explicit form of R1 is
 ��� � 2 
f0 � + |c2|2 ,
 c1 − 1 
��� � 2
f0 � + |c2|2 c∗1

R1 = (21)
f

and f are that in Eqs. (6) and (8). Next, a

where f0, c1, c2
controlled-NOT gate is performed on the ancillary sub
system, which
the first and the second qubits are the target and control ones
respectively. After another single qubit rotation
c∗2 f0

(22) R2 = 1
f0 2
+ |c2| 2
−f ∗
0 c2

is applied to the first ancillary qubit, the two ancillary qubits are
of the ancil
swapped again. Now, the basis |11�a lary subsystem is
deleted, and the rest dimensions |00�a, |01�a and |10�a are
left, constructing a six-dimensional subspace together with the
work qubit. Meanwhile, the three UE-parameters in Eq. (6) are
assigned to |00�a|ψ�e, |01�a|ψ�e and |10�a|ψ�e, which is
the key point in this step. The above operations have a similar
effect as UT1 performed on the ancillary qutrit in Fig. 4.
Based on the theory of three UE-terms, the second step aims at
generating the three UE-terms σ0, U1 and U2 in Eq. (6). Because
 C00−σ0 is a trivial unit matrix, only the two jointly-controlled gates
C01−U1 and C10−U2 in Fig. 5 are necessarily in practice, of which
the explicit forms are

  σ0 000 0 U1 0 0 
and C10−U2 =

 
σ0 000 0 σ0 0 0 
.
 
C01−U1 = (23)
0 0 σ0 0 000 σ0 0 0 U2 0 000 σ0

Refer to Eq. (6) and the Supplementary Information for U1 and U2.
Now, the three unitary terms are generated and entangled with the
three bases of the ancillary subspace.
In the third step, the three UE-terms are superposed by
swapping the two ancillary qubits three times with two single-qubit
rotations H2 and R3 in between as shown in Fig. 5, where

Scientific Reports | (2021) 11:3960 | [Link]

7
Content courtesy of Springer Nature, terms of use apply. Rights
reserved
Vol.:(0123456789)
[Link]/scientificreports/

Figure 6. Quantum circuit using two qubits. The system consists

of an ancillary and a work qubits, and the whole system is
initialized to a state |0�a|0�e . The work qubit can be further
rotated to an arbitrary state |ψ�e by Rψ. First, a single-qubit
rotation U is applied to the ancillary qubit to assign the two UE-
parameters. Then, two controlled operations generate the two
UE-terms. Third, a Hadamard superposes the UE-terms. Finally,
the work qubit e will be operated by F(t), if the ancillary qubit is
measured in |0�a.

1√ √ √
3 211− 2
.
R3 = (24)

Now, the whole system evolves to a superposition state

1√  |k�a|s′k�e  (25)
3f |00�aF(t)|ψ�e + ,
 
f � k=01,10

where the UE-terms are superposed in the first term as F(t). The
rest terms in Eq. (25) are not shown explicitly because they are not
linked to F(t)|ψ�e and will be discarded after the measurements.
Finally, measurements are performed on the ancillary qubits. If the
ancillary subsystem outputs a state |00�a, the work qubit will
evolve to F(t)|ψ�e, meaning that the nonunitary operator is
 simulated successfully. If |01�a or |01�a is observed, the process
will be terminated and started over again until |00�a is observed.
Therefore, it is an indeterministic protocol to simulate F(t). The
successful probability is the same as that in Eq. (20). Therefore,
not only a two-dimensional subspace are saved, but also the
successful probability is higher than that in Eq. (10) using the full-
eight dimensions.
We now analyze the complexities for the he 8D- and 6D-
protocols by comparing the two quantum circuits in Figs. 3 and 5.
In the circuit of the 6D-protocol, additional five two-qubit gates are
introduced to prepare the subspace in step 1 and 3, but one
controlled-controlled-σ3, or C2(σ3), is saved in step 2. According to
Ref.1 (on page 182), five or more controlled-gates are necessary to
implement the C2(σ3). Therefore, the globe complexities of the two
circuits to simulate a single-qubit nonunitary gate are nearly the
same, though the local complexities in some steps are different.

Four-dimensional protocol. Although three qubits are able to

simulate single-qubit nonunitary operators universally, less qubits
are preferred in some special cases for two reasons. One is to save
qubit resource and decrease the complexity of quantum algorithm.
The other reason is to increase the successful probability. Based
on our unitary expansion theory of two UE-terms, F(t) can be
expanded as Eq. (7) as long as one of the phase
matching conditions is met, enabling us to simulate it using
only two qubits.
Now, we show our four-dimensional protocol. The system
consists of a work and an ancillary qubits, and the work qubit will
be operated by F(t) with the assistance of the ancillary one. The
quantum circuit is shown in Fig. 6, in which time proceeds from left
to right.
At the beginning, the system is initialized to |0�a|0�e, and the
work qubit is rotated to |ψ�e by Rψ as needed. A single-qubit
rotation
1 a0 − a∗1
,
U = (26)
f a1 a∗0

is applied to the ancillary qubit to assign the UE-parameters a0 and

, respec
a1 in Eq. (7) to |0�a|0�e and |1�a|0�e tively. The
explicit forms of a0 and a1 are calculated for each phase
matching condition (refer to the Supplementary
Information), which depend on F(t) and always satisfy that
|a0|2 + |a1|2.
f = (27)

Then two controlled gates,

V0 0 and C1−V1 = σ0 0 ,

C0−V0 = (28)
0 σ0 0 V1

Scientific Reports | (2021) 11:3960 | [Link] 8

021-83521-5
Content courtesy of Springer Nature, terms of use apply. Rights
reserved
Vol:.(1234567890)
[Link]/scientificreports/

are performed, where the explicit forms of V0 and V1 vary in

phase matching conditions and have been given in
the Supplementary Information. The two controlled gates generate
 the two UE-terms and entangle them with the ancillary qubit. Next,
a Hadamard operation H2 is applied to the ancillary qubit to
superpose the two UE-terms, and the two-qubit system evolves to
a state

1√
2f[|0�aF(t)|ψ�e + |1�a(a0V0 − a1V1)|ψ�e].
(29)

The first term of the entangled state is linked to F(t), while the second term will be
discarded.
Finally, a measurement is performed on the ancillary qubit. If |0�a

is obtained, the work qubit e will be oper

ated by the nonunitary

F(t) with a successful probability of 1

2 (30)
e�ψ|F(t) F(t)|ψ�e,
†
2f

which depends on both the specific operator and the initial state
that it is performed on. If the ancillary qubit is measured in state |
1�a, the result of the work qubit will be discarded. If so, we start
over the simulation until the output |0�a is obtained. Although
F(t) can also be simulated by the above protocols when it meets
the phase matching conditions, the successful
probability using two qubits is higher than that using three qubits.
Therefore, it has benefits to judge whether the phase
matching conditions are met before quantum simulation to
save qubit and increase successful probability, so that the
complexities and difficulties for experimental implementations can
be decreased.

Successful probability and complexity. Successful

probability. From Eqs. (10), (20) and (30), we con clude
that the successful probabilities to simulate a nonunitary F(t) are
affected by three factors. The first one is F(t) itself, and the
second one is the initial state |ψ�e of the work qubit on which
F(t) is performed. The third factor is the dimensions of the total
used Hilbert space. The first two facts can be seen directly from
e�ψ|F(t) F(t)|ψ�e and f, while the third one is implicit.
†

In fact, we can rewrite Eqs. (10), (20) and (30) as a unified equation

dime
1 2 (31)
dim · f · e�ψ|F(t) F(t)|ψ�e,
†

where dim is the dimensions of the total used Hilbert space

including both the ancillary and work subsystems, and dime = 2
is the dimensions of the work subspace on which F(t) is
performed. The normalizing factor f is decided by the elements of
F(t) in Eq. (8). Therefore, the successful probability is proportional
e�ψ|F(t) F(t)|ψ�e and
†
to the dimensions of the work
subspace, and meanwhile is inversely proportional to f 2 and the
dimensions of the total used Hilbert space.
Equation (31) is also suitable for simulating a unitary operator by
one qubit. In this case, both dim and dime are equal to 2.
Since F(t) are unitary in this case, the normalizing factor f = 1
and F(t)†F(t) becomes the unit in SU(2). Substituting the values
into Eq. (31), the successful probability is equal to one. This means
that a single-qubit unitary operator can be simulated by one qubit
in a deterministic way, being accordance with the conventional
quantum mechanics. Therefore, Eq. (31) is correct for both the
unitary and nonunitary cases in which dime = 2 and dim = 2,
4, 6 or 8.
Our method can be generalized to simulate a high-dimensional
 or multi-qubit nonunitary operator and Eq. (31) is valid, which we
will illustrate by the last example in the next section. From Eq. (31),
the dimensions or number of qubits affect the successful
probability by the first fraction dime/dim. Noticing the
definitions of dime and dim, this fraction is equal to the
reciprocal of the used dimensions of ancillary subsystem, say
dima, which depends on the numbers of UE-terms. Given that a
D-dimensional nonunitary operator having D2 matrix ele
ments, a trivial unitary expansion needs about O(D2) UE-terms. In
this case, the successful probability is tending to zero as D → 0.
However, if we can express the operator by k UE-terms (k is
independent and much less than D), the first fraction in Eq. (31) is
equal to a constant of k−1. Therefore, one key point to increase
the successful probability is to express the nonunitary operator by
less UE-terms. We have investigated how to decrease UE
terms in details for single-qubit operators in the previous section,
but it would be more complex for the case of high-dimensional or
multi-qubit operators.

Complexity. To simulate F(t) using qubits, the

complexity of a 2-dimensional protocol is less than that of the
higher-dimensional protocols. The complexities of the 6- and 8-
dimensional protocols are nearly the same in total but different in
local steps. This may be led by the fact that, although the used
dimensions are different, the two ancillary subsystems have the
same dimension in total. Therefore, to simulate a single-qubit
nonunitary operator using qubits, we can conclude that: (1) the
complexity only depend on the number (but not the used
dimensions) of the ancillary qubits; (2) The complexity will be
increased as the number of the ancillary qubits increases. It is
evident that the second conclusion is still hold to simulate higher-
dimensional or multi-qubits nonunitary operators, while the first one
should be investigated carefully in higher-dimension case.
Assuming that qa ancillary qubits are used to simulate a multi-
qubits nonunitary operator F, we now analyze the complexity of
the sub-space preparation by treating this process as an oracle,
Bqa for qa ancillary qubits. One smaller oracle Bqa−1 to assign
q −1
parameters to (qa − 1)-ancillary qubits and 2 a (qa − 1)-qubit-
q −1
controlled gates (C a (F)) are needed to implement Bqa.
(q −1) q −2)
Because a C a (F) can be implemented by three C( a (F)
q −2
gates and two 2-qubit gate, a number of O(12 a ) 2-qubit gates
and O(1) single-qubit gates are necessary. Therefore,

Scientific Reports | (2021) 11:3960 | [Link]

9
Content courtesy of Springer Nature, terms of use apply. Rights
reserved
Vol.:(0123456789)
[Link]/scientificreports/

the complexity should be polynomial in the dimensions of the

q l
ancillary subsystem (O[(2 a ) ], where l is some positive number).

Illustrations
We take time-evolution operators of several non-Hermitian two-
state systems as examples to illustrate our uni tary expansions and
related protocols of quantum simulation. Quantum measurements
can be seen as nonlinear operators in some way. Our method can
be applied to simulate a single-qubit measurement in a four
dimensional Hilbert space without annihilating the measured qubit.
We apply our method to simulate Abrams–Lloyd’s gate, showing
how it is extended to higher-dimensional cases.

Non-Hermitian two-state systems. For some non-Hermitian

 Hamiltonians, a complex extension of the conventional quantum
mechanics is developed by Bender et al.13. The time-evolution
i Ht
operator e−( / ) keeps the unitarity as long as an appropriate
inner product is constructed in the exact PT-symmetric phase.
i Ht
We now simulate e−( / ) for an arbitrary time-independent
two-dimensional H by our method. Assuming the unitary
expansion of H is

H = a0σ0 + a1σ1 + a2(iσ2) + a3σ3, (32)

i/ )Ht
where ak ∈ C (k = 0, 1, 2, 3). If we set a = a21 + a22 + a23 , the operator e−( can be
expanded as

cos(at/ ) · σ0 − i sin (at/ ) 3 ak

e (33) − ia0 t

k=0 .
a · σk

Notice that a is a complex number in general, cos(at/ ) and i sin

iat/ −iat/
(at/ ) have definitions of e ±e /2. The four UE-
parameters are
t/
f0(t) = e (34) −ia0 · cos(at/ ) and fk(t) =

k −ia t/
−ia ae 0 · sin (at/ ) for k = 1, 2, 3. To judge whether

the phase matching conditions are met, the phase

angles of f0(t) and fk(t)’s (k = 1, 2, 3),

ia t/
k −ia t/
e− 0 · cos(at/ ) and θk = arg −ia ae 0 · sin (at/ ) ,

θ0 = arg (35)

should be calculated based on ak’s (k = 0, 1, 2, 3) in Eq. (32). If

one of the phase matching conditions is met, the time-
evolution operator can be simulated by two qubits. Or, three qubits
should be involved to achieve the quantum simulation by the eight-
or six-dimensional protocols, respectively.
To illustrate our theories, we investigate several typical non-
Hermitian two-state systems with PT, anti-PT, P-pseudo and
anti-P-pseudo Hermitian symmetries, of which the Hamiltonians
are

iϕ iϕ iθ , and iθ ,
HPT = se se re su re su
−iϕ , HAPT = i se −iϕ −iθ HAPPH = i −iθ
re
iθ se re re (36)
−iθ iθ −iθ
, HPPH =
re re re

satisfying [PT, HPT ] = 0, {PT, HAPT } = 0, [PT, HPPH] = 0

and {PT, HAPPH} = 0, where P = σ1 is the parity operator and
T is the time-reversal operator having the effect that i → −i.
The first three systems are found to meet the phase
matching condition I, II, and both I and IV, and can be
simulated by two qubits. For the last system, three qubits are
necessary to achieve the simulation in a general phase, while two
qubits are enough in some special cases. This conclusion is in
accordance with a previous work30. Details of the UE-terms and
parameters are presented in the Supplementary Information to
demonstrate the statements above.

Arbitrary single-qubit measurements. Quantum measurement

can be seen as a quantum operation from the view of quantum
information. Therefore, our unitary expansion method and
 simulation protocols can be applied to digitally simulate a single-
qubit measurement in a four dimensional Hilbert space without
annihi lation of the work qubit.
It can be generalized to simulate an arbitrary single-qubit
measurement M. Assuming the two orthogonal eigenstates of M
are |m�� and |m⊥�, we can apply our unitary-expansion
method to construct two nonunitary matrices

M� = |m���m�| and M⊥ = |m⊥��m⊥|. (37)

The effect of the measurement M performed on |ψ�e is equivalent to applying either M or M⊥ to |ψ�e
with a prob
ability of e�m�|ψ�e or e�m⊥|ψ�e, respectively. Notice that M� ± M⊥ are unitary, we
perform a series of unitary
operations of U = H2, 0-controlled M� + M⊥ , 1-controlled M�
− M⊥ and another H2 as quantum circuit in Fig. 6. The initial
state |0�a|ψ�e will evolve to a state

|0�aM�|ψ�e + |1�aM⊥|ψ�e. (38)

We now measure the ancillary qubit. If an output |k�a is obtained, the work qubit will evolve to |k�e
with a prob
ability of e�k|ψ�e (k = m�, m⊥) and be kept for further use.

Scientific Reports | (2021) 11:3960 | [Link]

10
Content courtesy of Springer Nature, terms of use apply. Rights
reserved
Vol:.(1234567890)
[Link]/scientificreports/

Figure 7. Quantum circuit to simulate Abrams–Lloyd’s gate. The system consists of a five-
dimensional
(5D) ancillary qudit and a four-dimensional (4D) work qudit (or two qubits), and is initialized to |0�a|
ψ�e.
UE-parameters are assigned by UN1 ∈ SU(5), and then five controlled operations generate the
related five
UE-terms Nk (k =0–4), which are superposed by UN2 ∈ SU(5). Finally, a measurement is
performed on the
ancillary qudit. If it outputs |0�a, the Abrams–Lloyd’s gate N will be successfully applied to the work
subsystem.

Abrams–Lloyd’s gate. Our unitary-expansion method can be

extended to simulate multi-qubit nonuni tary operators. Here we
take Abrams–Lloyd’s gate as an example for illustration.
Abrams and Lloyd investigated a nonlinear quantum algorithm9 to determine the existence of an input x
such that f (x) = 1. This algorithm is able to solve NP-complete
problems using polynomial time in theory. The key technique of the
nonlinear algorithm, which includes n input qubits and a flag qubit,
is to iterate a nonlinear transformation to each of the n qubits and
the flag qubit.
 The nonlinear transformation has effects that √12(|00�+|11�)
→ √12(|01�+|11�) , √12(|01�+|10�) → √12(|01�+|11�), and
1 1 , which can be described by a
√ 2(|00�+|10�) → √ 2(|00�+|10�)
two
qubit nonunitary gate (i.e., the Abrams–Lloyd’s gate)
 0 − 110 0 1 01 
   .
N = (39)
0 − 110
0 1 01

We simulate this two-qubit nonunitary gate N as an example of

higher-dimensional extension of our method. We find one unitary
expansion of N is

N = (40) 4

nkNk,
k=0

where the explicit forms of UE-terms Nk and parameters nk (k =

0, 1, 2, 3, 4) are given in the Supplementary Information. A system
composed of two work qubits (or a four-dimensional work qudit) and
a five-dimensional ancillary qudit is able to simulate the nonunitary
gate N. The quantum circuit is shown in Fig. 7. First, the whole
system is initialized to a pure state |0�a|0�e, and the work qubits
are rotated to |ψ�e as needed. The UE-parameters is assigned by
a unitary UN1 ∈ SU(5) of which the first column is

1 T (41)
fN n0 n1 n2 n3 n4 ,

where fN is a normalizing factor, calculated in the Supplementary

Information. The rest column vectors can be obtained by the
Gram–Schmidt orthogonalization. Then, five controlled gates
Ck−Nk’s (k = 0, 1, 2, 3, 4) are applied to generate the five UE-
terms, which the ancillary qudit controls the work qubits. Next,
another unitary operator UN2 ∈ SU(5), of which the the first row is
1√
5 11111 ,
(42)

is applied to the ancillary qudit to superpose the UE-terms. Now, the system evolves to a
state
√ |0�aN|ψ�e + fN |k�a|s k�e
4 ′′ .
5fN
1 k=1

(43)

Finally, a measurement is performed on the ancillary qudit. If it

, the Abrams–Lloyd’s gate is success
outputs |0�a fully applied to
the two work qubits. If the output is not |0�a, the result will be
discarded and the whole process will be started over until |0�a is
obtained.
The successful probability to simulate N is

1
5f 2Ne�ψ|N†N|ψ�e.
(44)

Scientific Reports | (2021) 11:3960 | [Link]

11
Content courtesy of Springer Nature, terms of use apply. Rights
reserved
Vol.:(0123456789)
[Link]/scientificreports/
 Given that the dimensions of the whole system and the two work
qubits are twenty and four respectively, it is also in accordance
with Eq. (31).

Analyze for experimental implementations

From “Duality quantum simulation”, an arbitrary single-qubit
nonunitary operator can be simulated by three qubits in general,
and both the six- and eight-dimensional protocols can be adopted
to realize the quantum simulation. Given that the stability and
controllability of a real quantum system, the six-dimensional
protocol is recommended because its successful probability is
higher than that of the eight-dimensional one. When phase
matching conditions are met, the four-dimensional protocol
is preferred because it not only increase the success
ful probability but also save one qubit, decreasing difficulties for experimental
implementations.
Both nuclear-magnetic-resonance (NMR) and quantum optics
are candidates. For an NMR system, a nuclei of spin-(1/2) take the
role as a qubit. The spatial-averaging method 83 can initialize the
system to a pseudo-pure state |00�a|0�e. Quantum gates are
realized by sequences of magnetic pulses. Since arbitrary quantum
gate can be divided into a series of single- and two-qubit gates 3,
we only need to analyze the two types of quantum gates. A single-
qubit rotation is realized by a series of hard pulses, inducing small
errors. Main errors are induced two-qubit operations, which are
combined by hard pulses and free evolutions of two nuclei in
periods of time22. Therefore, the four-dimensional protocol are
recommended because it contains less two-qubit gates than other
protocols.
Linear quantum optics system is another candidate for
experimental implementation. Both two orthogonal polarized
directions of a photon and two distinguished ways (positions) can
take roles as qubits. We take the polarization qubit as an example.
A single qubit operation can be realized by a series of half-wave
and quarter wave plates84. Although measurement induced
nonlinearity85,86 can realize two-qubit gates in principle, the
efficiency is extremely low especially when there are a lot of two-
qubit gates. Therefore, an NMR system is recommended for
experimental realizations of our protocols.

Conclusions
We utilize the LCU technique for nonunitary dynamics on a single
qubit, and simulate arbitrary time-dependent single-qubit
nonunitary operator F(t) using duality quantum algorithm. We give
explicit decompositions of the necessary unitaries, and minimize
the number of unitary-expansion terms and the relevant operators
for the single-qubit nonunitary evolutions. The successful
probability not only depends on the specific parameters of F(t)
and the initial state to which F(t) is applied, but also it is
proportional to the ratio of the dimensions of the work qubit to the
total used dimensions of the whole system. F(t) can be simulated
in an eight-, six- or four-dimensional Hilbert space, depending on
our unitary-expansion theory and phase matching
conditions. In general, three qubits are enough to simulate arbitrary
F(t) either using a six-dimensional subspace or the full space,
and the successful probability of the former one is higher than that
of the later one. In one of the four phase matching
conditions, we find F(t) can be simulated by only two qubits with a
higher successful probability. Therefore, it is necessary to judge
whether the phase matching conditions are met before
quantum simulation to improve the efficiency.
We illustrate with examples of typical non-Hermitian systems, such as (anti-)PT-symmetric, (anti-) P-
pseudo
Hermitian and a general non-Hermitian Hamiltonians, expanding
the time-evolution operators by unitary terms, checking the
phase matching conditions, and choosing related
protocols of quantum simulation. We also simulate a single-qubit
measurement which can be seen as a nonunitary operation using
our method. Our method can be extended to simulate a single-
qudit or multi-qubit nonunitary operator, and we apply it to the
Abrams–Lloyd’s gate as an example. Quantum simulation of
single-qubit nonunitary operators is able to be implemented by our
protocols at current stage. We will extend our method to simulate
 higher-dimensional nonunitary operator and optimize the UE-terms
in the future.

Received: 17 September 2020; Accepted: 4 February 2021

References
1. Nielsen, M. A. & Chuang, I. L. Quantum Computation
and Quantum Information. 10th anniversary ed.
(Cambridge University Press, Cambridge, 2010).
2. Shor, P. Algorithms for quantum computation: Discrete logarithms and
factoring. In Proceedings of the 35th Annual
Symposium on Fundamentals of Computer
Science 124–134 (IEEE Computer Society Press, Los Alamitos,
California, 1994).
3. Barenco, A. et al. Elementary gates for quantum computation. Phys. Rev. A 52,
3457–3467 (1995).
4. Grover, L. A fast quantum mechanical
algorithm for database searching. In
Proceedings of the 28th Annual ACM
Symposium on Theory of Computing (STOC),
212–219 (1996).
5. Georgescu, I. M., Ashhab, S. & Nori, F. Quantum simulation. Rev. Mod. Phys. 86,
153–185 (2014).
6. Greiner, M., Mandel, O., Esslinger, T., Hansch, T. W. & Bloch, I. Quantum
phase transition from a superfluid to a Mott insulator in a gas of ultracold
atoms. Nature 415, 39–44 (2002).
7. Pearson, J., Feng, G.-R., Zheng, C. & Long, G. L. Experimental quantum
simulation of avian compass in a nuclear magnetic reso nance system.
Sci. China Phys. Mech. Astron. 59, 120312 (2016).
8. Grover, L. K. Quantum computers can search rapidly by using almost any
transformation. Phys. Rev. Lett. 80, 4329–4332 (1998). 9. Abrams,
D. S. & Lloyd, S. Nonlinear quantum mechanics implies polynomial-time
solution for NP-complete and #P problems. Phys. Rev. Lett. 81,
3992–3995 (1998).
10. Terashima, H. & Ueda, M. Nonunitary quantum circuit. Int. J. Quantum Inf. 3,
633–647 (2005).
11. Bender, C. M. & Boettcher, S. Real spectra in non-Hermitian hamiltonians
having PT symmetry. Phys. Rev. Lett. 80, 5243–5246 (1998).
12. Bender, C. M., Boettcher, S. & Meisinger, P. N. PT-symmetric quantum mechanics. J. Math. Phys. 40,
2201–2229 (1999).

Scientific Reports | (2021) 11:3960 | [Link]

12
Content courtesy of Springer Nature, terms of use apply. Rights
reserved
Vol:.(1234567890)
[Link]/scientificreports/

13. Bender, C. M., Brody, D. C. & Jones, H. F. Complex extension of quantum mechanics. Phys. Rev. Lett.
89, 270401 (2002).
14. Lee, T. D. & Wick, G. C. Negative metric and the unitarity of the S matrix. Nucl. Phys. B 9, 209–
243 (1969).
15. Mostafazadeh, A. Pseudo-hermiticity versus PT symmetry: The necessary
condition for the reality of the spectrum of a non Hermitian hamiltonian. J.
Math. Phys. 43, 205–243 (2002).
16. Konotop, V. V., Yang, J. & Zezyulin, D. A. Nonlinear waves in PT-symmetric systems. Rev. Mod. Phys. 88,
035002 (2016).
17. Feynman, R. Simulating physics with computers. Int. J. Theor. Phys. 21, 467–488 (1982).
18. Gerritsma, R. et al. Quantum simulation of the Dirac equation. Nature 463, 68–
71 (2010).
19. Setia, K., Bravyi, S., Mezzacapo, A. & Whitfield, J. D. Superfast encodings
for fermionic quantum simulation. Phys. Rev. Res. 1, 033033
(2019).
20. Zheng, C., Song, S.-Y., Li, J.-L. & Long, G. L. Direct experimental simulation
of the Yang–Baxter equation. J. Opt. Soc. Am. B 30, 1688–1693
(2013).
21. Gunther, U. & Samsonov, B. F. Naimark-dilated PT-symmetric brachistochrone. Phys. Rev. Lett.
101, 230404 (2008).
22. Zheng, C., Hao, L. & Long, G. L. Observation of a fast evolution in a parity-
time-symmetric system. Philos. Trans. R. Soc. A 371,
20120053 (2013).
23. Zheng, C. & Wei, S. Duality quantum simulation of the Yang–Baxter equation. Int. J. Theor. Phys. 57,
2203–2212 (2018).
24. Wang, H. et al. Experimental simulation of the four-dimensional Yang–
Baxter equation on a spin quantum simulator. Phys. Rev. A 102,
 012610 (2020).
25. Zheng, C. Duality quantum simulation of a general parity-time-symmetric two-level system. EPL 123, 40002
(2018).
26. Wen, J. et al. Experimental demonstration of a digital quantum simulation
of a general PT-symmetric system. Phys. Rev. A 99, 062122 (2019).
27. Zheng, C. Duality quantum simulation of a generalized anti-PT-symmetric two-level system. EPL 126, 30005
(2019).
28. Pinske, J., Teuber, L. & Scheel, S. Holonomic gates in pseudo-Hermitian quantum systems. Phys. Rev. A 100,
042316 (2019).
29. Wen, J. et al. Observation of information flow in the anti-PT-symmetric
system with nuclear spins. NPJ Quantum Inf. 6, 28 (2020). 30.
Zheng, C. et al. Efficient quantum simulation of an anti-P-pseudo-Hermitian
two-level system. Entropy 22, 812 (2020).
31. Breuer, H.-P. & Petruccione, F. The Theory of Open
Quantum Systems. 10th anniversary ed. (Oxford University Press,
Oxford, 2002). 32. Barreiro, J. T. et al. An open-system quantum simulator
with trapped ions. Nature 470, 486–491 (2011).
33. Hu, Z., Xia, R. & Kais, S. A quantum algorithm for evolving open quantum
dynamics on quantum computing devices. Sci. Rep. 10, 3301 (2020).
34. Del Re, L., Rost, B., Kemper, A. F. & Freericks, J. K. Driven-dissipative
quantum mechanics on a lattice: Simulating a fermionic reservoir on a
quantum computer. Phys. Rev. B 102, 125112 (2020).
35. Viyuela, O. et al. Observation of topological Uhlmann phases with superconducting qubits. NJP Quantum
Inf. 4, 10 (2018).
36. Bender, C. M., Brody, D. C. & Jones, H. F. Scalar quantum field theory with a
complex cubic interaction. Phys. Rev. Lett. 93, 251601 (2004).
37. Bender, C. M. Making sense of non-Hermitian hamiltonians. Rep. Prog. Phys. 70,
947–1018 (2007).
38. Bender, C. M., Brody, D. C., Jones, H. F. & Meister, B. K. Faster than
Hermitian quantum mechanics. Phys. Rev. Lett. 98, 040403
(2007).
39. Bender, C. M. & Mannheim, P. D. No-ghost theorem for the fourth-order
derivative Pais–Uhlenbeck oscillator model. Phys. Rev. Lett.
100, 110402 (2008).
40. Bender, C. M., Hook, D. W., Meisinger, P. N. & Wang, Q. H. Complex
correspondence principle. Phys. Rev. Lett. 104, 061601 (2010). 41.
Bittner, S. et al. PT symmetry and spontaneous symmetry breaking in a
microwave billiard. Phys. Rev. Lett. 108, 024101 (2012). 42. Jing, H.
et al. PT-symmetric phonon laser. Phys. Rev. Lett. 113, 053604
(2014).
43. Peng, B. et al. Parity-time-symmetric whispering-gallery microcavities. Nat. Phys. 10,
394–398 (2014).
44. Lu, X.-Y., Jing, H., Ma, J.-Y. & Wu, Y. PT-symmetry-breaking chaos in optomechanics. Phys. Rev. Lett.
114, 253601 (2015).
45. Bender, C. M. PT-symmetric quantum theory. J. Phys. Conf. Ser. 631,
012002 (2015).
46. Liu, Z. P. et al. Metrology with PT-symmetric cavities: Enhanced
sensitivity near the PT-phase transition. Phys. Rev. Lett. 117,
110802 (2016).
47. Bender, C. M. Nonlinear eigenvalue problems and PT-symmetric quantum
mechanics. J. Phys. Conf. Ser. 873, 012002 (2017). 48. Liang, F.,
El-Ganainy, R. & Li, G. Non-Hermitian photonics based on parity-time symmetry.
Nat. Photonics 11, 752–762 (2017). 49. Silva, E. F., Barbosa, A. L. R.
& Ramos, J. G. G. S. Parity and time-reversal symmetry in the Hanbury Brown-
twiss effect. EPL 117, 14001 (2017).
50. Longhi, S. Parity-time symmetry meets photonics: A new twist in non-Hermitian optics. EPL 120, 64001
(2017).
51. El-Ganainy, R. et al. Non-Hermitian physics and PT symmetry. Nat. Phys. 14,
11–19 (2018).
52. Jin, L. & Song, Z. Incident direction independent wave propagation and
unidirectional lasing. Phys. Rev. Lett. 121, 073901 (2018). 53.
Klauck, F. et al. Observation of PT-symmetric quantum interference.
Nat. Photonics 13, 883–887 (2019).
54. Li, J. et al. Observation of parity-time symmetry breaking transitions in a
dissipative Floquet system of ultracold atoms. Nat. Commun. 10,
855 (2019).
55. Jin, L., Wu, H. C., Wei, B.-B. & Song, Z. Hybrid exceptional point created
from type-III Dirac point. Phys. Rev. B 101, 045130 (2020).
56. Bian, Z. et al. Conserved quantities in parity-time symmetric systems. Phys. Rev. Res. 2,
022039(R) (2020).
57. Mostafazadeh, A. Pseudo-Hermiticity versus PT-symmetry III: Equivalence of
pseudo-hermiticity and the presence of antilinear symmetries. J. Math.
Phys. 43, 3944–3951 (2002).
58. Solombrino, L. Weak pseudo-hermiticity and antilinear commutant. J. Math. Phys. 43, 5439–
5445 (2002).
59. Nixon, S. & Yang, J. All-real spectra in optical systems with arbitrary gain-
and-loss distributions. Phys. Rev. A 93, 031802(R) (2016). 60.
Mostafazadeh, A. Time-dependent pseudo-Hermitian hamiltonians and a hidden
geometric aspect of quantum mechanics. Entropy 22, 471 (2020).
61. Chu, Y., Liu, Y., Liu, H. & Cai, J. Quantum sensing with a single-qubit
 pseudo-Hermitian system. Phys. Rev. Lett. 124, 020501 (2020).
62. Ge, L. & Tureci, H. E. Antisymmetric PT-photonic structures with balanced
positive- and negative-index materials. Phys. Rev. A 88, 053810
(2013).
63. Hang, C., Huang, G. & Konotop, V. V. PT symmetry with a system of three-level atoms. Phys. Rev. Lett.
110, 083604 (2013).
64. Antonosyan, D. A., Solntsev, A. S. & Sukhorukov, A. A. Parity-time anti-
symmetric parametric amplifier. Opt. Lett. 40, 4575 (2015). 65. Wu, J.-H.,
Artoni, M. & La Rocca, G. C. Parity-time-antisymmetric atomic lattices without
gain. Phys. Rev. A 91, 033811 (2015). 66. Peng, P. et al. Anti-
parity-time symmetry with flying atoms. Nat. Phys. 12, 1139 (2016).
67. Wang, X. & Wu, J.-H. Optical PT-symmetry and PT-antisymmetry in
coherently driven atomic lattices. Opt. Express 24, 4289–4298
(2016).
68. Yang, F., Liu, Y. C. & You, L. Anti-PT symmetry in dissipatively coupled optical systems. Phys. Rev. A 96,
053845 (2017).
69. Choi, Y., Hahn, C., Yoon, J. W. & Song, S. H. Observation of an anti-PT-
symmetric exceptional point and energy-difference con serving dynamics in
electrical circuit resonators. Nat. Commun. 9, 2182 (2018).
70. Konotop, V. V. & Zezyulin, D. A. Odd-time reversal PT symmetry induced by
an anti-PT-symmetric medium. Phys. Rev. Lett. 120, 123902
(2018).

Scientific Reports | (2021) 11:3960 | [Link]

13
Content courtesy of Springer Nature, terms of use apply. Rights
reserved
Vol.:(0123456789)
[Link]/scientificreports/

71. Chuang, Y.-L., Ziauddin, A. & Lee, R.-K. Realization of simultaneously parity-
time-symmetric and parity-time-antisymmetric susceptibilities along the
longitudinal direction in atomic systems with all optical controls. Opt.
Express 26, 21969–21978 (2018).
72. Li, Y. et al. Anti-parity-time symmetry in diffusive systems. Science 364, 170–173 (2019).
73. Long, G.-L. General quantum interference principle and duality computer. Commun. Theor. Phys.
45, 825–843 (2006).
74. Long, G.-L. & Liu, Y. Duality quantum computing. Front. Comput. Sci.
China 2, 167 (2008).
75. Long, G.-L., Liu, Y. & Wang, C. Allowable generalized quantum gates. Commun. Theor.
Phys. 51, 65–67 (2009).
76. Cui, J., Zhou, T. & Long, G. L. Density matrix formalism of duality quantum
computer and the solution of zero-wave-function paradox. Quantum
Inf. Process. 11, 317–323 (2012).
77. Wei, S.-J. & Long, G.-L. Duality quantum computer and the efficient quantum
simulations. Quantum Inf. Process. 15, 1189–1212 (2016).
78. Shao, C. P., Li, Y. & Li, H. B. Quantum algorithm design: Techniques and
applications. J. Syst. Sci. Complex. 32, 375–495 (2019). 79.
Childs, A. M. & Wiebe, N. Hamiltonian simulation using linear combinations of
unitary operations. Quantum Inf. Comput 12, 901–924 (2012).
80. Wei, S.-J., Li, H. & Long, G.-L. A full quantum eigensolver for quantum
chemistry simulations. Research 2020, 1486935 (2020). 81. Wang, Y.,
Hu, Z., Sanders, B.C. & Kais, S. Qudits and High-
Dimensional Quantum Computing. Preprint at
[Link] abs/2008.00959 (2020).
82. Cao, Y., Peng, S. G., Zheng, C. & Long, G. L. Quantum fourier transform and
phase estimation in qudit system. Commun. Theor. Phys.
55, 790–794 (2011).
83. Cory, D. G., Price, M. D. & Havel, T. F. Nuclear magnetic resonance
spectroscopy: an experimentally accessible paradigm for quantum
computing. Phys. D 120, 82–101 (2008).
84. Hu, S. W., Xue, K. & Ge, M. L. Optical simulation of the Yang–Baxter equation. Phys. Rev. A 78,
022319 (2008).
85. Knill, E., Laflamme, R. & Milburn, G. A. Scheme for efficient quantum
computation with linear optics. Nature 409, 46–52 (2001). 86. Cerf, N. J.,
Adami, C. & Kwiat, P. G. Optical simulation of quantum logic. Phys. Rev.
A 57, R1477(R) (1998).

Acknowledgements
This work was supported by the National Natural Science
Foundation of China Grant No. 11705004, Organiza tion
Department of Beijing Municipal Committee Talents Project No.
2017000020124G050, the Fundamental Research Funds of Beijing
Municipal Education Commission, and NCUT Research Fund for
Key Discipline No. 110052972027/014 and Talents Project.
 Author contributions
C.Z. conceived the ideas, designed the protocol, analyzed the results, and wrote the
manuscript.

Competing interests
The author declares no competing interests.

Additional information
Supplementary Information The online version contains
supplementary material available at [Link]
org/10.1038/s41598-021-83521-5.
Correspondence and requests for materials should be addressed to C.Z.
Reprints and permissions information is available at
[Link]/reprints.
Publisher’s note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original
author(s) and the source, provide a link to the Creative Commons
licence, and indicate if changes were made. The images or other
third party material in this article are included in the article’s
Creative Commons licence, unless indicated otherwise in a credit
line to the material. If material is not included in the article’s
Creative Commons licence and your intended use is not permitted
by statutory regulation or exceeds the permitted use, you will need
to obtain permission directly from the copyright holder. To view a
copy of this licence, visit
[Link]

© The Author(s) 2021

Scientific Reports | (2021) 11:3960 | [Link]

14
Content courtesy of Springer Nature, terms of use apply. Rights
reserved
Vol:.(1234567890)
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center
GmbH (“Springer Nature”). Springer Nature supports a reasonable amount of sharing of research papers by
authors, subscribers and authorised users (“Users”), for small scale personal, non-commercial use provided
that all copyright, trade and service marks and other proprietary notices are maintained. By accessing,
sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use
(“Terms”). For these purposes, Springer Nature considers academic use (by researchers and students) to be
non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a
relevant site licence or a personal subscription. These Terms will prevail over any conflict or ambiguity with
regards to the relevant terms, a site licence or a personal subscription (to the extent of the conflict or
ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used
will apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use
these personal data internally within ResearchGate and Springer Nature and as agreed share it, in an
anonymised way, for purposes of tracking, analysis and reporting. We will not otherwise disclose your
personal data outside the ResearchGate or the Springer Nature group of companies unless we have your
permission as detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is
important to note that Users may not:

1.
use such content for the purpose of providing other users with access on a regular or large scale
basis or as a means to circumvent access control;
2.
use such content where to do so would be considered a criminal or statutory offence in any
jurisdiction, or gives rise to civil liability, or is otherwise unlawful;
3.
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless
explicitly agreed to by Springer Nature in writing;
4.
use bots or other automated methods to access the content or redirect messages
5.
override any security feature or exclusionary protocol; or
6.
share the content in order to create substitute for Springer Nature products or services or a systematic
database of Springer Nature journal content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product
or service that creates revenue, royalties, rent or income from our content or its inclusion as part of a paid for
service or for other commercial gain. Springer Nature journal content cannot be used for inter-library loans
and librarians may not upload Springer Nature journal content on a large scale into their, or any other,
institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated
to publish any information or content on this website and may remove it or features or functionality at our
sole discretion, at any time with or without notice. Springer Nature may revoke this licence to you at any
time and remove access to any copies of the Springer Nature journal content which have been saved. To the
fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users,
either express or implied with respect to the Springer nature journal content and all parties disclaim and
waive any implied warranties or warranties imposed by law, including merchantability or fitness for any
particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by
Springer Nature that may be licensed from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular
basis or in any other manner not expressly permitted by these Terms, please contact Springer Nature at

onlineservice@[Link]