Demonstration of Quantum Energy Teleportation on Superconducting Quantum

Hardware

Kazuki Ikeda1, 2, ∗
1
Co-design Center for Quantum Advantage, Stony Brook University, Stony Brook, New York 11794-3800, USA
2
Center for Nuclear Theory, Department of Physics and Astronomy,
Stony Brook University, Stony Brook, New York 11794-3800, USA
Teleporting physical quantities to remote locations is a remaining key challenge for quantum
information science and technology. Quantum teleportation has enabled the transfer of quantum
information, but teleportation of quantum physical quantities has not yet been realized. Here we report the realization and
observation of quantum energy teleportation on real superconducting quantum hardware. We achieve this by using several

IBM’s superconducting quantum computers.

arXiv:2301.02666v5 [quant-ph] 22

Aug 2023
The results are consistent with the exact solution of the theory and are improved by the mitigation of
measurement error. Quantum energy teleportation requires only local operations and classical
communication. Therefore our results provide a realistic benchmark that is fully achievable with
current quantum computing and communication technologies.
be applied

I. QUANTUM ENERGY TELEPORTATION

∗
While it is fairly widely known that information about [email protected]; [email protected]
quantum states can be transported to remote loca tions to any system capable of QET.
[1–4], it is less well known that quantum state energy Although QET is conceptually similar to QST, here it
can be similarly transmitted, despite its impact and is important to emphasize that, it is classical infor
potential for future applications. Quantum informa tion mation, not energy, that is sent, and the intermediate
transferred by quantum state teleportation (QST) is not subsystem along the channel between the sender
a physical quantity, but energy is a distinct physical (Alice) and receiver (Bob) is not excited by the energy
quantity. Transferring physical quantities to remote loca carriers of the system during the short duration of the
tions is an unexplored area of technology. Quantum En QET pro cess. Thus, the time scale of energy transport
ergy Teleportation (QET) was proposed by Hotta about by QET is much shorter than the time scale of heat
15 years ago and has been studied theoretically for generation in the natural time evolution. Bob can
spin chains [5–7], an ion trap system [8], a quantum extract energy from a system by performing operations
Hall sys tem [9], and other various theoretical systems on his local system based on the classical information
[10, 11]. QET has only recently been experimentally transmitted by Alice. By teleportation of energy or
validated us ing an NMR setup [12]. energy transfer, we mean that Bob can receive energy
The purpose of this paper is to make the much faster (at the speed of light) than the energy that
experimental verification of QET with actual cloud can be transmitted from Alice to Bob in the natural time
quantum comput ers in the most visible way, and to evolution of the system.
establish the optimized quantum circuits that make it In what follows, we explain that QET is a univer sal
possible. We achieved the realization of QET using means of quantum energy extraction mediated by a
some IBM superconducting quantum computers by many-body quantum system. Any non-trivial local oper
applying quantum error mitiga tion [13–15]. The ations, including measurements on the ground state of
quantum hardware we used includes IBM’s quantum a quantum many-body system, give rise to excited
computer ibmq lima, which is available free of charge states, which in turn increase the energy expectation
to everyone in the world. The quantum algorithm used value. Note that the increase in energy is supplied by
in this work is open access to the pub lic [16], where the experimen tal devices. An important property of the
quantum circuit implementation of QET is provided and ground state of a quantum many-body system is that it
real-time information to the latest ma chine properties has entan glement, which brings about local quantum
is accessible. Using the quantum circuits provided in fluctuations in the global ground state. In QET,
this paper, anyone will be able to reproduce the results measurement plays an important role. Local
and QETs of this study. Since all the proper ties of measurement of the quantum state at a subsystem A
quantum computers are publicly available in real time, destroys this ground state en tanglement. At the same
it will be possible for anyone to verify the QET pro from the de
tocol, regardless of whether one owns a quantum time, energy EA vice making the
device or not. The methods we have established can measurement is injected into the entire system. The
 stays around the sub quantum networks [21]. The maximum circuit depth is 6
injected energy EA system A in
and the number of qubits used is 2. Hence, current
the very early stages of time evolution, but operations quantum computers are powerful enough to implement
around A alone cannot extract EA from the system. This QET.
is because information about EA is also stored in Let k, h be positive real numbers. The Hamiltonian of
remote locations other than A due to the en tanglement 2
that exists prior to the measurement [17]. QET is the
protocol that makes this possible by combing LOCC 1/k = O(100) ns. The density matrix ρ
QET after Bob
and conditional operations. Note that, because of the
operates U1(µ) to P0(µ)|g⟩ is
mid-circuit measurement, the time evolution of the
U1(µ)P0(µ)|g⟩ ⟨g| P0(µ)U†1(µ). (6)
ρQET =
post-measurement state is not unitary. Up to this point,
µ∈{−1,1}
no special assumptions about the quantum system
have been used. The crucial property of QET is that it Using ρQET, the expected local energy at Bob’s subsys
can be realized entirely by the general nature of the tem is evaluated as ⟨E1⟩ = Tr[ρQET(H1 + V )], which is
ground state of the entangled quantum many-body
negative in general. Due to the conservation of energy,
system and the universal fact of measurement. Using
EB = −⟨E1⟩(> 0) is extracted from the system by the
this ground state entanglement, the phase diagrams of
many-body systems are reproduced by QET, where device that operates U1(µ) [22]. In this way, Alice and
Alice and Bob’s coordinates are fixed and teleported Bob can transfer the energy of the quantum system by
energy to Bob’s lo cal system reproduces the phase operations on their own local system and classical com
structure [18, 19]. Those findings suggest that global munication (LOCC).
structures such as symmetry, topology, and
long-correlation of many-body systems can be detected
by simple LOCC; one does not need to mea sure all II. QUANTUM CIRCUIT IMPLEMENTATION OF
QUANTUM ENERGY TELEPORTATION
qubits/states, which is advantageous for research in
quantum physics using quantum computers.
We work on the minimal QET model given in [20]. A. Preparation of Ground State
One of the purposes of this paper is to give a quantum
circuit that utilizes QET with real quantum computers The exact ground state |g⟩ is prepared by
and quantum networks. The complete form of quantum
circuits we used for QET is displayed in Fig. 1, which is |g⟩ = CNOT(RY (2θ) ⊗ I)|00⟩, (7)
recently extended to long-range and large-scale
the minimal model is √ √ . The correspond
where θ = − arccos 121 − h h2+k2
ing quantum circuit is shown in Fig. 1 (A).

Htot = H0 + H1 + V, (1) Hn = hZn +h2

√ 2 2 B. Step 1: Deposit Energy
h + k , (n = 0, 1) (2)
We use the following projective measurement
V = 2kX0X1 +2k2
√ 2 2 1
h + k . (3) operator P0(µ) = 2(1 + µX0). (8)

The ground state of Htot is

1√ √ 2 2 1√ √ 2 2 Bob’s energy since [X0, V ] =
h + k |00⟩− 2 h + k |11⟩, (4)
|g⟩ = 2 [X0, H1] = 0. Using [P0(µ), V ]
We measure Alice’s X
= 0 and ⟨+|Z |+⟩ = ⟨−|Z |−⟩ =
operator, by which we obtain
0, we find that Alice’s
1 −h 1 +h a state |+⟩ or |−⟩. This
operation does not affect
The constant terms in the Hamiltonians are added
so that the ground state |g⟩ of Htot returns the zero ⟨g| P0(µ)HtotP0(µ)|g⟩ =h2
mean
mean energy to deposit is
energy for all local and global ⟨g| H1 |g⟩ = ⟨g| V |g⟩ = 0. (5) √ 2 2
h + k . (9)
Hamiltonians: ⟨g| Htot |g⟩ = ⟨g| H0 |g⟩ =
⟨E0⟩ = µ∈{−1,1}
However it should be noted that |g⟩ is neither a ground Alice’s operation can be implemented on a quantum
state nor an eigenstate of Hn, V, Hn + V (n = 0, 1). The ⟩ can be calculated with the out
circuit in Fig 1 (A). ⟨E0
essence of QET is to extract negative ground state
put bit-strings 00, 01, 10, 11. Analytical values ⟨E0⟩ and
energy of those local and semi-local Hamiltonians.
results with quantum computers for different pairs of k
The QET protocol is as follows. First, Alice makes a
and h are summarized in Table I.
measurement on her Pauli operator X0 by P0(µ) = 1
2(1 +µX0) and then she obtains either µ = −1 or +1. It
turns out that Alice’s expectation energy is E0 =h2 C. Step 2: Receive Energy
√ 2 2
h +k .
As soon as Alice observes µ ∈ {−1, 1}, she tells her
Via a classical channel, Alice then sends her
measure ment result µ to Bob, who applies an (µ) to his qubit and mea
result to Bob who operates UB
operation U1(µ) to his qubit and measures H1 and V .
sures his energy. Here UB(µ) is U1(µ) = cos ϕI − iµ sin
She tells the result in a time t, which must be much
shorter than the coupling time scale t ≪ 1/k. In our ϕY1 = RY (2µϕ), (10)
experiment, t = O(10) ns and
3

FIG. 1: Quantum gate operations used for quantum energy teleportation. (A) preparation of ground state and Alice’s X0
measurement to deposit her energy. She tells Bob via classical communication whether µ = −1 or µ = +1 was observed. (B)
Bob’s conditional operations to receive energy. He selects an operation U1(+1) or U1(−1) based on µ = +1 or −1, corresponding
to the Maxwell demon operation. (C) Equivalent implementation of Bob’s operations on a quantum computer.
(B). The corresponding quantum circuit is obtained by
removing the second Hadamard gate from the previous
where ϕ obeys circuit Fig. 1 (C). On average the circuit generates the
energy expectation value
2 2
cos(2ϕ) = h + 2k
⟨g| P0(µ)U†1(µ)(H1 + V )U1(µ)P0(µ)|g⟩
2 2 2 2 2
(h + 2k ) + h k (11)

sin(2ϕ) = hk
2 ⟨E1⟩ = µ∈{−1,1}
(h2 + 2k2)2 + h2k . (12)

The average quantum state ρQET eq.(6) is obtained af map µ∈{−1,1} P0(µ)|g⟩ ⟨g| P0(µ) → ρQET is not a uni tary
ter Bob operates U1(µ) to P0(µ)|g⟩. Then the average transformation. Therefore eq. (13) can be negative.
This is in contrast to eq. (A7).
energy Bob measures is
Now let us explain quantum circuits for the QET pro
⟨E1⟩ = Tr[ρQET(H1 + V )] = Tr[ρQETHtot] − ⟨E0⟩, (13) tocol. Since V and H1 do not commute, measurement
on those terms should be done separately. In other
where we used [U1(µ), H1] = 0. It is important that the words, Bob measures V and H1 independently and ob
tains ⟨V ⟩ and ⟨H1⟩ statistically. As Fig. 3 in Appendix A relatively small errors.
shows, ⟨V ⟩ is always negative and ⟨H1⟩ is always posi The time scale of QET can be estimated by
tive. Therefore it is sufficient for Bob to measure only ⟨V comparing the gate time of a single-qubit gate and that
⟩ to receive energy by QET. of a two qubit gate (CNOT gate). As shown in Table. III,
We consider ⟨V (µ)⟩ = ⟨g| P0(µ)U†1(µ)V U1(µ)P0(µ)|g⟩. the gate time of a CNOT gate is between 200-500 ns.
On the other hand, the duration of a single qubit
The quantum circuit to compute ⟨V (µ)⟩ is shown in the
right panel of Fig. 1 (B). It is important to note that, operator is between 20-40 ns 1. Therefore Bob can
since Bob knows µ which contains Alice’s information, extract energy much faster than the time it takes for
he can obtain ⟨V ⟩ = Tr[ρQETV ] by local measurement energy to be transferred from Alice to Bob in the natural
unitary time evolution.
only, although V is not a local operator. Similarly we
can measure H1 in Z-basis as in the left panel of Fig. 1 We also performed a simulation using a simulator
qasm simulator, which can classically emulate gate op
= −1 erations on the same quantum circuits we used for
√ 2 2 quantum computation. We summarize results with ibmq
h + k [hk sin(2ϕ) − (h2 + 2k2)(1 − cos(2ϕ))]. lima, ibmq jakarta and ibm cairo in Table I. The results
(14) using the simulator agreed with the analytical so lution
with high accuracy, confirming that the quantum circuit
If ϕ is small, ⟨E1⟩ is negative. Bob receives energy ⟨EB⟩ was implemented correctly. More experimental results
= −⟨E1⟩ on average. In Appendix B, we per formed are summarized in Table IV in Appendix D. We describe
details of machine properties and experimental
measurement of V (µ) and H1 based on the quan tum
conditions in Table III in Appendix C.
circuit Fig. 1 (B) and summarized data in Table II,
The most significant achievement in this study is the
where numerical values are compared with analytical
val ues given in eq. (14). observation of negative energy ⟨E1⟩ < 0. The value of ⟨V
⟩ that was closest to the exact analysis value was -
0.1079 (h = 1.5, k = 1 with ibmq jakarta), which is about
D. QET on Real Quantum Hardware 76% accurate. As emphasised in Hotta’s origi nal works
[5–11, 20], after Alice observes her X0, no unitary
Here we describe how to implement conditional operation can make ⟨E1⟩ negative (eq. (A7)). In order for
opera tions that may not be natively supported by many Bob to obtain the correct ⟨E1⟩, Alice and Bob must
quan tum computers and quantum devices. In the QET repeat the experiment an enormous number of times,
pro tocol, Bob’s operation must be selected according and the correct value of ⟨V ⟩ and ⟨H1⟩ can be ob tained
to the results of Alice’s measurements, as shown in only when Alice and Bob communicate correctly in the
Fig. 1 (B). Even in environments where conditional quantum circuit in Fig. 1 (C). Distributions of states
statements are not supported, QET can be obtained by a quantum computer ibm cairo are shown
implemented without prob lems through the technique in Fig. 2 (B), where distributions of raw results and error
of deferred measurement. mitigated results are compared with a simu
We can postpone Alice’s measurement until the end
of the circuit, and obtain the same results. The condi
tional operations can be created by a controlled U gate
1
For the latest real-time information of the duration of each op erator
in each machine, please visit [16]
Λ(U) = |0⟩ ⟨0| ⊗ I + |1⟩ ⟨1| ⊗ U and an anti-controlled U 4
gate (X ⊗ I)Λ(U)(X ⊗ I). One would find the equiva lence
between the following two circuits. We use the right lator qasm simulator. We used a simple measurement
circuit enclosed by the orange dashed frame in Fig. 1 error mitigation to determine the effects of
(C). measurement errors. We prepared a list of 4
We performed quantum computation using 6 dif measurement calibra tion circuits for the full Hilbert
ferent types of IBM quantum hardware ibmq lima, ibmq space. Then we immedi ately measured them to obtain
jakarta, ibmq hanoi, ibm cairo, ibm auckland and ibmq the probability distribu tions. Then we applied the
montreal. The properties of each quantum computer calibration matrix to correct the measured results. The
can be seen from Fig. 2. ibmq lima con sists of 5 qubits average measurement fidelity when using each
(Fig. 2 [Left]) and ibmq jakarta has 7 qubits (Fig. 2 quantum computer is summarized in Table III in
[Middle]). ibm cairo is a 27-qubit hardware, and ibmq Appendix C. The histograms of the observed states
hanoi, ibm cairo, ibm auckland and ibmq montreal have showed similar tendencies for all other quantum
the same graph structure as ibm cairo (Fig. 2 [Right]). A computers we used. It can be seen that the histograms
direct CNOT gate can be applied to two qubits obtained by the measurement of H1 agree with the sim
connected at the edge. We can choose two qubits ulator results with good accuracy. The improvement of
placed on the graph of the hardware to perform the values due to measurement error mitigation is also
quantum computation. We conducted the ex periment confirmed by the results in Table I. The observation of V
by choosing two qubits connected at the edge with is of utmost importance in this study. Although the raw
 data from quantum computers deviated from the simula can be tackled in the short term with current quantum
tor results, in some cases, error mitigation improved computing and communica tion technology. A quantum
them enough to observe negative energy expectation device with 2 qubits and a gate depth of 6 would be
values. ready for immediate experi ments. This is expected to
It should also be emphasized that we observed lead to new developments in the use of quantum
negative ⟨V ⟩ for all parameter (k, h) combinations in all memory [23–25]. Furthermore, veri fying QET in a
quantum computers used. As emphasized in Sec. II C, variety of quantum systems and materials beyond the
the amount of energy available to Bob is greater if only minimal model is an important challenge for future
V is observed since ⟨H1⟩ is always positive (Fig. 3). This applications.
would be enough for practical purposes. Note that the QET without limit of distance is also provided [26].
energy that Bob gains becomes smaller when he The combination of QST and QET has been
observes H1. generalized to a simple universal QET protocol on
arbitrarily large scale quantum networks [21]. The ability
to transfer quantum energy over long distances at the
III. IMPLICATIONS FOR OUR REAL WORLD speed of light will bring about a new revolution in
quantum communi cation technology. For example,
there is a long-distance (∼158km) SBU/BNL quantum
Our results provide implications for new quantum
network in Long Island, New York [27]. Various
com munication technologies with respect to different
quantum networks have been developed [28–30].
phases in the short, medium and long term. It is
Realizing QET on a quantum net work, which is
important to note that, like quantum teleportation,
expected to be in practical use around the 2030s, would
energy can also be teleported only by LOCC.
be a milestone toward realizing QET on a worldwide
Reproducing the minimal QET model we used in our
quantum network.
demonstration in a labora tory system is something that
5

FIG. 2: (A) properties of quantum computers we used. Each graph of qubits corresponds to the layout of the hardware. A direct
CNOT gate can be applied to two qubits connected at the edge. (B) Distribution of states compared with a simulator qasm
simulator and a quantum computer ibm cairo (raw results and mitigated results). Similar tendencies in the histograms were
found for all other quantum computers used.
information and communication technology and
quantum physics. More recently QET has been applied
The realization of a long-range QET will have impor to quantum inter active proof and quantum
tant implications beyond the development of cryptography on a large scale quantum network [31],
and it provides a secure authen tication system with ing on transaction costs, Bob may choose Alice. A lot
zero-knowledge. In addition, QET will help develop the of such game-theoretic situations can be created
quantum economy market. Infor mation and energy are [33–37]. This implies that quantum information
physical, but also economic. Al lowing physical economics (which does not yet exist) will become a
quantities to be traded concretely on the quantum meaningful idea in the future.
network means that a new economic market will be
born [32]. Quantum teleportation is an established
technology and is being developed for practical use. In
Acknowledgement
addition to this, if QET is put to practical use, it will
mean that various quantum resources will be at the dis
posal of us. The expected value of the Hermite I thank David Frenklakh, Adrien Florio, Sebas tian
operator is called energy, but it need not literally be Grieninger, Fangcheng He, Dmitri Kharzeev, Yuta
used only as energy. Teleported energy can be used as Kikuchi, Vladimir Korepin, Qiang Li, Adam Lowe,
energy, as well as for other uses. The ability to teleport Shuzhe Shi, Hiroki Sukeno, Tzu-Chieh Wei, Kwangmin
a concrete physical quantity, energy, means that Yu and Ismail Zahed for fruitful communication and col
quantum informa tion will have added value. In a laboration. After posting to the first version of this paper
quantum market where Alice, Bob, and Charlie exist, if to arXiv on 7 Jan 2023, Mart´ın-Mart´ıne told me about
Bob can get more en his demonstration with an NMR. I thank him for the
ergy from Charlie than from Alice, Bob may prefer to do communication. I thank Megumi Ikeda for providing the
business with Charlie rather than Alice, and he may cartoons. I acknowledge the use of IBM quantum com-
prefer an entangle state with Charlie. However, depend
6
Backend (h, k) = (1, 0.2) (h, k) = (1, 0.5) (h, k) = (1, 1) (h, k) = (1.5, 1)

Analytical ⟨E0⟩ 0.9806 0.8944 0.7071 1.2481

value

qasm 0.9827 ± 0.8941 ± 0.0001 0.7088 ± 1.2437 ±

simulator 0.0031 0.0001 0.0047

ibmq lima error 0.9423 ± 0.8169 ± 0.0032 0.6560 ± 1.2480 ±

mitigated 0.0032 0.0031 0.0047

unmitigated 0.9049 ± 0.8550 ± 0.0032 0.6874 ± 1.4066 ±

0.0017 0.0031 0.0047

ibmq jakarta error 0.9299 ± 0.8888 ± 0.0056 0.7039 ± 1.2318 ±

mitigated 0.0056 0.0056 0.0084

unmitigated 0.9542 ± 0.9089 ± 0.0056 0.7232 ± 1.2624 ±

0.0056 0.0056 0.0083

ibm cairo error 0.9571 ± 0.8626 ± 0.0031 0.7277 ± 1.2072 ±

mitigated 0.0032 0.0031 0.0047

unmitigated 0.9578 ± 0.8735 ± 0.0031 0.7362 ± 1.2236 ±

0.0031 0.0031 0.0047

Analytical ⟨H1⟩ 0.0521 0.1873 0.2598 0.3480

value

qasm 0.0547 ± 0.1857 ± 0.0022 0.2550 ± 0.3487 ±

simulator 0.0012 0.0028 0.0038

ibmq lima error 0.0733 ± 0.1934 ± 0.0032 0.2526 ± 0.3590 ±

mitigated 0.0032 0.0032 0.0047

unmitigated 0.1295 ± 0.2422 ± 0.0024 0.2949 ± 0.4302 ±

0.0053 0.0028 0.0039

ibmq jakarta error 0.0736 ± 0.2018 ± 0.0056 0.2491 ± 0.3390 ±

mitigated 0.0055 0.0056 0.0084

unmitigated 0.0852 ± 0.2975 ± 0.0045 0.3365 ± 0.4871 ±

0.0022 0.0052 0.0073
 ibm cairo error 0.0674 ± 0.1653 ± 0.0031 0.2579 ± 0.3559 ±
mitigated 0.0032 0.0031 0.0047

unmitigated 0.0905 ± 0.1825 ± 0.0022 0.2630 ± 0.3737 ±

0.0014 0.0027 0.0037

Analytical ⟨V ⟩ -0.0701 -0.2598 -0.3746 -0.4905

value

qasm −0.0708 ± −0.2608 ± −0.3729 ± −0.4921 ±

simulator 0.0012 0.0032 0.0063 0.0038

ibmq lima error −0.0655 ± −0.2041 ± −0.2744 ± −0.4091 ±

mitigated 0.0012 0.0031 0.0063 0.0063

unmitigated −0.0538 ± −0.1471 ± −0.1233 ± −0.2737 ±

0.0011 0.0025 0.0041 0.0046

ibmq jakarta error −0.0515 ± −0.2348 ± −0.3255 ± −0.4469 ±

mitigated 0.0022 0.0056 0.0112 0.0112

unmitigated −0.0338 ± −0.1371 ± −0.0750 ± −0.2229 ±

0.0021 0.0046 0.0075 0.0083

ibm cairo error −0.0497 ± −0.1968 ± −0.2569 ± −0.3804 ±

mitigated 0.0013 0.0031 0.0063 0.0063

unmitigated −0.0471 ± −0.1682 ± −0.1733 ± −0.3089 ±

0.0012 0.0026 0.0038 0.0045

Analytical ⟨E1⟩ -0.0180 -0.0726 -0.1147 -0.1425

value

qasm −0.0161 ± −0.0751 ± −0.1179 ± −0.1433 ±

simulator 0.0017 0.00398 0.0069 0.0054

ibmq lima error 0.0078 ± −0.0107 ± −0.0217 ± −0.0501 ±

mitigated 0.0034 0.0045 0.0071 0.0079

unmitigated 0.0757 ± 0.0950 ± 0.0035 0.1715 ± 0.1565 ±

0.0054 0.0050 0.0060

ibmq jakarta error 0.0221 ± −0.0330 ± −0.0764 ± −0.1079 ±

mitigated 0.0059 0.0079 0.0125 0.0140

unmitigated 0.0514 ± 0.1604 ± 0.0064 0.2615 ± 0.2642 ±

0.0030 0.0091 0.00111

ibm cairo error 0.0177 ± −0.0315 ± 0.0010 ± −0.0245 ±

mitigated 0.0035 0.0044 0.0070 0.0079

unmitigated 0.0433 ± 0.0143 ± 0.0034 0.0897 ± 0.0648 ±

0.0018 0.0047 0.0058

TABLE I: Comparison between analytical values of ⟨E0⟩,⟨H1⟩,⟨V ⟩,⟨E1⟩ and results from IBM’s real quantum computers, ibmq lima,
ibmq jakarta and ibm cairo. We evaluate ⟨E1⟩ = ⟨H1⟩ + ⟨V ⟩. ”error mitigated” means results using measurement error mitigation
and ”unmitigated” corresponds to results without measurement error mitigation.

puters. I was supported by the U.S. Department of En

ergy, Office of Science, National Quantum Information
Science Research Centers, Co-design Center for [1] C. H. Bennett, G. Brassard, C. Cr´epeau, R. Jozsa, A.
Quantum Advantage (C2QA) under Contract Peres, and W. K. Wootters, “Teleporting an unknown
quantum state via dual classical and
No.DESC0012704.
einstein-podolsky-rosen channels,” Phys. Rev. Lett. 70
(Mar, 1993) 1895–1899. https:
 //link.aps.org/doi/10.1103/PhysRevLett.70.1895. [2] A. [14] R. Takagi, S. Endo, S. Minagawa, and M. Gu,
Furusawa, J. L. Sørensen, S. L. Braunstein, C. A. “Fundamental limits of quantum error mitigation,”
References npj Quantum Information 8 (Jan., 2022) 114,
arXiv:2109.04457 [quant-ph].
[15] Z. Cai, R. Babbush, S. C. Benjamin, S. Endo,
W. J. Huggins, Y. Li, J. R. McClean, and T. E.
O’Brien, “Quantum Error Mitigation,” arXiv e-prints
(Oct., 2022) arXiv:2210.00921, arXiv:2210.00921
[quant-ph].
[16] K. Ikeda, “Quantum Energy Teleportation
with Quantum Computers,” Feb., 2023.
Fuchs, H. J. Kimble, and E. S. Polzik, “Unconditional https://github.com/IKEDAKAZUKI/
quantum teleportation,” science 282 no. 5389, (1998)
Quantum-Energy-Teleportation.git.
706–709.
[17] M. Hotta, “Energy entanglement relation for
[3] S. Pirandola, J. Eisert, C. Weedbrook, A. Furusawa, quantum energy teleportation,” Physics Letters A 374
and S. L. Braunstein, “Advances in quantum no. 34, (July, 2010) 3416–3421, arXiv:1002.0200
teleportation,” Nature Photonics 9 no. 10, (Oct., 2015) [quant-ph].
641–652, arXiv:1505.07831 [quant-ph]. [18] K. Ikeda, “Investigating global and topological
7 order of states by local measurement and classical
communication: Study on SPT phase diagrams
[4] S. Takeda, T. Mizuta, M. Fuwa, P. Van Loock, and by quantum energy teleportation,” arXiv e-prints
(Feb., 2023) arXiv:2302.09630,
arXiv:2302.09630
[19] K. Ikeda, “Criticality of quantum energy
[quant-ph]. teleportation at phase transition points in quantum
A. Furusawa, “Deterministic quantum teleportation field theory,”
of photonic quantum bits by a hybrid technique,” Phys. Rev. D 107 (Apr, 2023) L071502. https:
Nature 500 no. 7462, (2013) 315–318. //link.aps.org/doi/10.1103/PhysRevD.107.L071502.
[5] M. Hotta, “A protocol for quantum energy [20] M. Hotta, “Quantum Energy Teleportation: An
distribution,” Physics Letters A 372 no. 35, (2008) Introductory Review,” arXiv e-prints (Jan., 2011)
5671–5676. https://www.sciencedirect.com/science/ arXiv:1101.3954, arXiv:1101.3954 [quant-ph].
article/pii/S0375960108010347. [21] K. Ikeda, “Long-range quantum energy
[6] M. Hotta, “Quantum Energy Teleportation in Spin teleportation and distribution on a hyperbolic
Chain Systems,” Journal of the Physical Society of quantum network,” arXiv e-prints (Jan., 2023)
Japan 78 no. 3, (Mar., 2009) 034001, arXiv:2301.11884,
arXiv:0803.0348 [quant-ph]. arXiv:2301.11884 [quant-ph].
[7] J. Trevison and M. Hotta, “Quantum energy [22] M. Hotta, “Quantum measurement information as
teleportation across a three-spin Ising chain in a a key to energy extraction from local vacuums,”
Gibbs state,” Journal of Physics A Mathematical Phys. Rev. D 78 (Aug, 2008) 045006. https:
General 48 no. 17, (May, 2015) 175302, //link.aps.org/doi/10.1103/PhysRevD.78.045006.
arXiv:1411.7495 [quant-ph]. [23] S. Dhomkar, J. Henshaw, H. Jayakumar, and C.
[8] M. Hotta, “Quantum energy teleportation with A. Meriles, “Long-term data storage in diamond,”
trapped ions,” Phys. Rev. A 80 no. 4, (Oct., 2009) Science Advances 2 no. 10, (2016) e1600911,
042323, arXiv:0908.2824 [quant-ph]. https://www.science.org/doi/pdf/10.1126/sciadv.1600911.
[9] G. Yusa, W. Izumida, and M. Hotta, “Quantum https://www.science.org/doi/abs/10.1126/sciadv. 1600911.
energy teleportation in a quantum hall system,” [24] H. Kurokawa, M. Yamamoto, Y. Sekiguchi, and
Phys. Rev. A 84 (Sep, 2011) 032336. https: H. Kosaka, “Remote entanglement of
//link.aps.org/doi/10.1103/PhysRevA.84.032336. superconducting qubits via solid-state spin
[10] Y. Nambu and M. Hotta, “Quantum energy quantum memories,” Phys. Rev. Appl. 18 (Dec,
teleportation with a linear harmonic chain,” Phys. Rev. 2022) 064039. https://link.aps.
A 82 (Oct, 2010) 042329. https: org/doi/10.1103/PhysRevApplied.18.064039.
//link.aps.org/doi/10.1103/PhysRevA.82.042329. [25] H. P. Specht, C. N¨olleke, A. Reiserer, M.
[11] M. Hotta, “Quantum energy teleportation with an Uphoff, E. Figueroa, S. Ritter, and G. Rempe, “A
electromagnetic field: discrete versus continuous single-atom quantum memory,” Nature 473 no.
variables,” Journal of Physics A: Mathematical and 7346, (2011)
Theoretical 43 no. 10, (Feb, 2010) 105305. https: 190–193.
//dx.doi.org/10.1088/1751-8113/43/10/105305. [12] N. [26] M. Hotta, J. Matsumoto, and G. Yusa, “Quantum
A. Rodr´ıguez-Briones, H. Katiyar, energy teleportation without a limit of distance,”
E. Mart´ın-Mart´ınez, and R. Laflamme, Phys. Rev. A 89 no. 1, (Jan., 2014) 012311,
“Experimental activation of strong local passive arXiv:1305.3955 [quant-ph].
states with quantum information,” Phys. Rev. Lett. [27] D. Du, P. Stankus, O.-P. Saira, M. Flament,
130 (Mar, 2023) 110801. S. Sagona-Stophel, M. Namazi, D. Katramatos, and
https://link.aps.org/doi/10.1103/PhysRevLett. E. Figueroa, “An elementary 158 km long quantum
130.110801. network connecting room temperature quantum
[13] S. Endo, S. C. Benjamin, and Y. Li, memories,” arXiv e-prints (Jan., 2021)
“Practical Quantum Error Mitigation for arXiv:2101.12742, arXiv:2101.12742 [quant-ph].
Near-Future [28] H. J. Kimble, “The quantum internet,” Nature
Applications,” Physical Review X 8 no. 3, (July, 453 no. 7198, (2008) 1023–1030.
2018) 031027, arXiv:1712.09271 [quant-ph]. [29] Y.-A. Chen, Q. Zhang, T.-Y. Chen, W.-Q. Cai,
 S.-K. Liao, J. Zhang, K. Chen, J. Yin, J.-G. Ren, https://www.science.org/doi/pdf/10.1126/science.abg1919.
Z. Chen, et al., “An integrated space-to-ground https://www.science.org/doi/abs/10.1126/science. abg1919.
quantum [31] K. Ikeda and A. Lowe, “Quantum interactive
communication network over 4,600 kilometres,” proofs using quantum energy teleportation,”
Nature 589 no. 7841, (2021) 214–219. arXiv:2306.08242 [quant-ph].
[30] M. Pompili, S. L. N. Hermans, S. Baier, H. K. C. [32] K. Ikeda and S. Aoki, “Theory of quantum
Beukers, P. C. Humphreys, R. N. Schouten, R. F. L. games and quantum economic behavior,”
Vermeulen, M. J. Tiggelman, L. dos Santos Quantum Information Processing 21 no. 1, (2022)
Martins, B. Dirkse, S. Wehner, and R. Hanson, 1–29.
“Realization of a multinode quantum network of [33] K. Ikeda, “Foundation of quantum optimal transport
remote solid-state
qubits,” Science 372 no. 6539, (2021) 259–264,
8

and applications,” Quantum Information Processing 19

no. 1, (Nov., 2019) 25, arXiv:1906.09817 [quant-ph].
[34] K. Ikeda, “Quantum extensive-form games,” Quantum
Information Processing 22 no. 1, (Jan, 2023) 66.
https://doi.org/10.1007/s11128-022-03806-0.
[35] K. Ikeda and S. Aoki, “Infinitely repeated quantum
games and strategic efficiency,” Quantum Information
Processing 20 no. 12, (Dec., 2021) 387,
arXiv:2005.05588 [quant-ph].
[36] K. Ikeda and A. Lowe, “Quantum Protocol for Decision
Making and Verifying Truthfulness amongN-quantum
Parties: Solution and Extension of the Quantum Coin
Flipping Game,” arXiv e-prints (Nov., 2022)
arXiv:2211.02073, arXiv:2211.02073 [quant-ph].
[37] K. Ikeda, “Quantum contracts between Schr¨odinger and
a cat,” Quantum Information Processing 20 no. 9,
(Sept., 2021) 313.
9

Appendix A: Description of the Model

XX
description of the back =
1. Quantum Gates and Measurement
U
Anti-control U=
Here we give a self-contained
U
Now we describe the measurement of quantum
ground knowledge of the text. We use the following opera tors. We measure Z1 and X0X1. Measurement of
one qubit operators whose matrix representations are Z1 is done by the following circuit
given as
,
X= 01 0 −i i 0
(A1)
1 0, Y =

−1 1√ −1
Z= = 2
100 ,H 111 .
The output of the measurement is a bit string b0b1 ∈
{00, 01, 10, 11}. Since the eigenvalues of Z are −1, 1,
We use |0⟩ =10, |1⟩ =01forthe computational basis
we convert the bit string into 1 − 2b1. Let nshot be the
states, which are eigenstates of Z: Z |0⟩ = |0⟩, Z |1⟩ = −
|1⟩. We also work with another ba sis vectors |±⟩ =|0⟩±|1⟩ number of repetitions of the circuit, and countsb0b1 be
√ the number of times b0 and b1 are detected. Therefore
2. They are eignestates of X:
b
X |−⟩ = − |−⟩, X |+⟩ = − |+⟩. Note that |±⟩ are created by counts b
0 1

applying H to |0⟩ and |1⟩; H |0⟩ = |+⟩, H |1⟩ = |−⟩. Those is ob

nshotsis the probability that a bit string b0b1 tained.
are used for measuring Hn, V (n = 1, 2) in the QET
protocol. For example, Alice finds µ = ±1 by ob Then the expectation value of Z1 is computed by the
 formula
nshots. (A5)
serving the eigenvalues ±1 of her local ⟨Z1⟩ = b0,b1
Pauli X operator. The rotation of X, Y,
Z is defined by (1 − 2b1)countsb0b1
H
RX(α) = e−iα2 X, RY (α) = e−iα2Y, RZ(α) = e−iα2 Z. (A2)
We use two-qubit gate operations. In general, a
control U operation Λ(U) is defined by H

Λ(U) = |0⟩ ⟨0| ⊗ I + |1⟩ ⟨1| ⊗ U (A3) and the

As we described previously, H maps |0⟩, |1⟩ to |+⟩,
corresponding diagram is drwan as |−⟩, which are eigenvectors of X. The output of the
measurement is again a bit string b0b1 ∈ {00, 01, 10,
11}. They are converted to the eigenvalues of X0X1 by
control U=
(1 − 2b0)(1 − 2b1). Then the expectation value of X0X1
U
is computed by the formula

One of the most frequently used controlled gates is a (1 − 2b0)(1 − 2b1)countsb0b1

CNOT gate CNOT = Λ(X), whose diagram is especially

Measurement of X0X1 is done by the following circuit

drawn as
⟨X0X1⟩ = b0,b1
nshots. (A6)
CNOT=

Here we describe the details of the model we used.

For more information please refer to Hotta’s original
papers. First, it is important to note that the ground
Using the formula CNOT(a |0⟩ + b |1⟩)|0⟩ = a |00⟩ + b
state of the total Hamiltonian H is not the ground state
|11⟩, it is easy to check
of lo cal operators. For example, V has three
CNOT(RY (2θ) ⊗ I)|00⟩ = cos θ |00⟩ + sin θ |11⟩. (A4) degenerated

2. Some details of the model

√1 √ eq. (7) is indeed the ground state |g⟩.
√
When θ = − arccos 21 − h
, we find that 2, and the ground
2 2
h +k
ground states |−+⟩, |+−⟩, |−+⟩+|+−⟩ state energy of V is −2k +2k2
It is convenient to define an anti-control gate, which that V ’s ground state energy is negative for all k > 0.
is activated when the control bit is in state |0⟩: |1⟩ ⟨1|⊗I This is also true for Hn, whose ground state energy is
+ |0⟩ ⟨0| ⊗ U, whose diagram is drawn as
√ 2 2
h +k . It is important
10

FIG. 3: Heat maps visualizing expectation values ⟨V ⟩ = Tr[ρQETV ] and ⟨H1⟩ = Tr[ρQETH1] by (k, h).
 To understand the non-triviality of the QET protocol,
−h +h2 Hence, QET can be described as a fast energy
√ 2 2 propaga tion protocol.
h +k .
The expected values of ⟨V ⟩ = Tr[ρQETV ]
and ⟨H1⟩ = Tr[ρQETH1] obtained by QET are shown in It is known that the change in entropy before and
after the measurement can be evaluated as follows
Fig. 3.
it is important to note that after Alice’s 2 0, (A7)
measurement, no matter what unitary ≥1 + sin ξ
2 cos3 ξln 1 + cos ξ 1 − cos ξ
operation W1 is performed on Bob’s pµSAB(µ) (A10)
qubit, no energy can be extracted. EB
This can be confirmed by √ 2 2
h + k (A11)
∆SAB = SAB − µ∈{±1}
Tr[ρW Htot] − ⟨E0⟩ = ⟨g| W†1 HtotW1 |g⟩ ≥
 after the energy that Bob would

measurement, ξ = receive is bounded
where µ∈{−1,1}  k
W1. (A8) arctan hand EB is the below by the
where pµ is the amount of energy that difference of entropy:
†
ρW = W 1 probability distribution Bob can receive (EB =
 of µ, SAB(µ) is the −⟨E1⟩ > 0) [20].
P0(µ)|g⟩ ⟨g| P0(µ) entanglement entropy Moreover the maximal
√ 2 2 2 2
U1(µ)EB ≥2 h + k (4 − 3 cos ξ − 2 + cos ξ)∆SAB
The inequality in eq. (A7) is guaranteed by eq. (5).
If Bob does not perform any local system is as follows h2(1 − cos(4kt)) .
operations on his own system max
after Alice’s measurement, ⟨H (t)⟩ = Tr[ρ eitHH e−itH 2 + (1 − cos ξ) ln 2 1−cos ξ
1 M 1 ]= (1 + cos ξ) ln 1+cos ξ
the time evolution of Bob’s (A12)
than the speed of light. Classical communication,
realized by optical com
√ 2 2
2 h + k Appendix B: Simulation of Hotta’s original QET protocol
itH −itH
⟨V (t)⟩ = Tr[ρMe V e ] = 0,
(A9) Hotta’s original QET protocol, which can be imple
mented by Fig. 1 (B) in the main text, does require the
where ρM =µ∈{±1} P0(µ)|g⟩ ⟨g| P0(µ). conditional operations based on a signal µ ∈ {−1, +1}
As a result of the natural time evolution of the sys that Bob receives from Alice. We performed quantum
tem, energy is indeed transferred to Bob’s local system, computation on the equivalent circuit (right quantum cir
but this is no more than energy propagation in the cuit in Fig. 1 (C)) that yielded exactly the same results.
usual sense. In QET, energy is not obtained through Let Λ(U) = |0⟩ ⟨0| ⊗ I + |1⟩ ⟨1| ⊗ U be a controlled U gate.
the nat ural time evolution of the system, but Note that Λ(U(−1)) and (X ⊗ I)Λ(U(+1))(X ⊗ I) commute:
instantaneously as a result of communication. Since we
consider a non relativistic quantum many-body system,
the speed of en ergy propagation is sufficiently slower =
munication, can convey information systems. U(+1) U(−1)
to remote locations much faster than U(−1) U(+1)
the time evolution of physical
11
(h, k) (1,0.1) (1,0.2) (1,0.5) (1,1) (1.5,1)

Analytical ⟨E0⟩ 0.9950 0.9806 0.8944 0.7071 1.2481

qasm simulator 0.9929 ± 0.9807 ± 0.8948 ± 0.7067 ± 1.2492 ±
⟨E0⟩ Analytical ⟨V ⟩ 0.0010 0.0010 0.0010 0.0010 0.0015
qasm simulator -0.0193 -0.0701 -0.2598 -0.3746 -0.4905
⟨V ⟩ Analytical −0.0194 ± −0.0682 ± −0.2625 ± −0.3729 ± −0.4860 ±
⟨H1⟩ 0.0057 0.0144 0.0011 0.0521 0.0061 0.1873 0.0063 0.2598 0.0061 0.3480
qasm simulator 0.0136 ± 0.0501 ± 0.1857 ± 0.2550 ± 0.3493 ±
⟨H1⟩ Analytical ⟨E1⟩ 0.0006 0.0011 0.0022 0.0028 0.0038
qasm simulator ⟨E1⟩ -0.0049 -0.0180 -0.0726 −0.1147 -0.1425
−0.0058 ± −0.0181 ± −0.0768 ± −0.1179 ± −0.1367 ±
0.0057 0.016 0.0064 0.0068 0.0072
TABLE II: Comparison between analytical values and numerical values from the quantum circuits with conditional opera tion
(Fig. 1 (B)). Each error corresponds to statistical error of 105shots. We evaluate ⟨E1⟩ = ⟨H1⟩ + ⟨V ⟩.
depth that a quantum computer can successfully
implement. How ever, QV may not be a crucial metric in
this study, since
we are only dealing with 2-qubit, relatively simple quan
Of course, the equivalence of these circuits is
theoreti cally trivial, we used qasm simulator and
executed our simulation based on the left quantum
circuit in Fig. 1 (C), in order to confirm the consistency
between them. Table II summarizes the numerical
results and shows per fect agreement with the
analytical results as well as re sults (Table IV) with the
right circuit in Fig. 1 (C).

Appendix C: Properties of Quantum Hardware

Here we describe more on our experiments with IBM

quantum computers. Graphs of IBM quantum
computers we used are displayed in Fig 4. For
example, the layout of ibmq lima corresponds to (A) in
Fig. 4 and we used the pair of qubits in [0,1] that had
the smallest readout assignment error among all pairs
(Fig. 2 (A) [Left]). We can perform a direct CNOT
operation between qubits connected at the edge. For FIG. 4: Configurations of qubits on graphs: (A) the lay out of
ibmq lima, the CNOT error between [1,2] qubits were ibmq lima which has 5 qubits; (B) the layout of ibmq jakarta
which has 7 qubits; (C) the layout of 27-qubit hardware
0.00510 (Table. III).
including ibmq hanoi, ibm cairo, ibm auckland and ibmq
montreal. A direct CNOT gate can be applied to two qubits
connected at the edge.
Appendix D: Additional results with 6 different quantum
hardware

Here we describe additional results obtained by tum circuits. Errors in quantum computers result from a
some other IBM quantum computers. In the main text combination of various factors, including readout error,
we fo cused on best results with ibmq lima and ibmq CNOT error, etc.. Table IV shows that Alice’s measure
jakarta, but in fact we also experimented with ibmq ments of X0 are relatively accurate in almost all cases.
hanoi, ibm cairo, ibm auckland, ibmq montreal. Table IV With respect to the observation of V , there is a
summarizes the complete lists of the best data we ob deviation from the analytical value. It was confirmed
tained and Table III summarizes the experimental con that the error mitigation improved the results. In this
ditions used for each hardware. In the entire circuit, the study, what is important is that negative expectation
total number N of qubits is 2 and the circuit depth d(N) values ⟨V ⟩ were observed for all cases. It is a
that can be executed is 5 (excluding measurement of V noteworthy achievement that negative energy
) and 6 (including measurement of V ). The quantum expectation values ⟨E⟩ < 0 were observed by error
vol ume is defined by QV = arg maxn≤N min{n, d(n)}2. mitigation. In fact, the histograms of states (Fig. 2 (B))
Therefore quantum computers with QV = 64 are have improved to approach the ex act values,
enough for this work. Here QV is a metric that indicating that all operations were performed correctly.
quantifies the largest random circuit of equal width and
12
Backend ibmq lima ibmq ibm cairo ibm hanoi d ibmq
jakarta

Ntot 5 7 27 27 27 27
Quantum Volume 8 16 64 64 64 128
shots 105 3.2 × 104 105 105 105 3.2 × 104
Measurement fidelity 0.961075 0.924695 0.961935 0.979530 0.979383 0.957484
qubits used [0,1] [3,5] [13,14] [14,16] [14,16] [14,16]
CNOT error 0.00510 0.00665 0.00439 0.01996 0.00570 0.00739
Gate time (ns) 305.778 291.556 220.444 472.889 355.556 355.556
CLOPS 2700 2400 2400 2300 2400 2000
 F t1(µs) 75.67 93.53 146.43 219.15 60.97 129.56
t2(µs) 141.39 41.09 164.29 25.75 150.49 168.53
i
Frequency (GHz) 5.030 5.178 5.282 5.047 5.167 4.961
r Anharmonicity -0.33574 -0.34112 -0.33874 -0.34412 -0.34196 -0.32314
s (GHz) Pauli X error 2.781 × 2.140 × 1.630 × 2.305 × 2.4842 × 1.942 ×
Readout assignment 10−4 10−4 10−4 10−4 10−4 8.100 10−4
t error 1 .960 × 2.440 × 8.500 × 7.400 × × 10−3 1.310 ×
q 10−2 10−2 10−3 10−3 10−2

S t1(µs) 58.03 143.52 94.28 190.07 73.16 83.73

t2(µs) 74.97 59.33 186.99 253.46 183.12 39.92
e
Frequency (GHz) 5.128 5.063 5.044 4.883 4.970 5.086
c Anharmonicity -0.31835 -0.34129 -0.34289 -0.34591 -0.34389 -0.33707
o (GHz) Pauli X error 1.469 × 1.708 × 1.732 × 4.708 × 2.052 × 2.221 ×
Readout assignment 10−4 10−4 10−4 10−4 10−4 10−4
n error 1 .300 × 2.400 × 8.000 × 9.600 × 7.700 × 9.800 ×
d 10−2 10−2 10−3 10−3 10−3 10−3

TABLE III: Machine properties of IBM quantum computers and parameters we used. shots is the number of iterations we
performed for sampling. Average measurement fidelity was computed when preparing a calibration matrix and used for
measurement error mitigation. CNOT error corresponds to the direct CNOT error between two qubits [q0, q1] used. Gate time
corresponds to the gate time between [q0, q1]. The first and second qubits correspond to q0 and q1, respectively. t1 is relaxation
time and t2 is dephasing time. For those who use ibmq lima with an open account, the maximal nshots is 3.2 × 104. CLOPS means
the number of circuit layer operations per second, indicating how many layers of a circuit a quantum processing unit (QPU) can
execute per unit of time.
13
Backend (h, k) = (1, 0.2) (h, k) = (1, 0.5) (h, k) = (1, 1) (h, k) = (1.5, 1)

Analytical ⟨E0⟩ 0.9806 0.894 0.7071 1.2481

value

ibmq lima error 0.9423 ± 0.8169 ± 0.6560 ± 1.2480 ±

mitigated 0.0032 0.0032 0.0031 0.0047

unmitigated 0.9049 ± 0.8550 ± 0.6874 ± 1.4066 ±

0.0017 0.0032 0.0031 0.0047

ibmq jakarta error 0.9299 ± 0.8888 ± 0.7039 ± 1.2318 ±

mitigated 0.0056 0.0056 0.0056 0.0084

unmitigated 0.9542 ± 0.9089 ± 0.7232 ± 1.2624 ±

0.0056 0.0056 0.0056 0.0083

ibm hanoi error 1.0685 ± 0.9534 ± 0.7852 ± 1.3728 ±

mitigated 0.0032 0.0032 0.0031 0.0047
 unmitigated 1.0670 ± 0.9524 ± 0.7809 ± 1.3663 ±
0.0031 0.0031 0.0031 0.0047

ibm cairo error 0.9571 ± 0.8626 ± 0.7277 ± 1.2072 ±

mitigated 0.0032 0.0031 0.0031 0.0047

unmitigated 0.9578 ± 0.8735 ± 0.7362 ± 1.2236 ±

0.0031 0.0031 0.0031 0.0047

ibm auckland error 0.9766 ± 0.8703 ± 0.6925 ± 1.2482 ±

mitigated 0.0032 0.0032 0.0032 0.0047

unmitigated 0.9771 ± 0.8712 ± 0.6931 ± 1.2487 ±

0.0032 0.0032 0.0032 0.0047

ibmq montreal error 0.8774 ± 0.8084 ± 0.6315 ± 1.1449 ±

mitigated 0.0056 0.0056 0.0056 0.0084

unmitigated 0.9036 ± 0.8338 ± 0.6564 ± 1.1819 ±

0.0056 0.0056 0.0056 0.0084

Analytical ⟨H1⟩ 0.0521 0.1873 0.2598 0.3480

value

ibmq lima error 0.0733 ± 0.1934 ± 0.2526 ± 0.3590 ±

mitigated 0.0032 0.0032 0.0032 0.0047

unmitigated 0.1295 ± 0.2422 ± 0.2949 ± 0.4302 ±

0.0053 0.0024 0.0028 0.0039

ibmq jakarta error 0.0736 ± 0.2018 ± 0.2491 ± 0.3390 ±

mitigated 0.0055 0.0056 0.0056 0.0084

unmitigated 0.0852 ± 0.2975 ± 0.3365 ± 0.4871 ±

0.0022 0.0045 0.0052 0.0073

ibm hanoi error 0.1786 ± 0.3256 ± 0.4276 ± 0.5890 ±

mitigated 0.0032 0.0032 0.0032 0.0047

unmitigated 0.2012 ± 0.3427 ± 0.4378 ± 0.6104 ±

0.0019 0.0026 0.0031 0.0042

ibm cairo error 0.0674 ± 0.1653 ± 0.2579 ± 0.3559 ±

mitigated 0.0032 0.0031 0.0031 0.0047

unmitigated 0.0905 ± 0.1825 ± 0.2630 ± 0.3737 ±

0.0014 0.0022 0.0027 0.0037

ibm auckland error 0.1218 ± 0.2004 ± 0.2181 ± 0.3215 ±

mitigated 0.0032 0.0031 0.0032 0.0047

unmitigated 0.1455 ± 0.2205 ± 0.2337 ± 0.3493 ±

0.0017 0.0023 0.0027 0.0038

ibmq montreal error 0.0897 ± 0.1618 ± 0.1921 ± 0.2816 ±

mitigated 0.0056 0.0056 0.0056 0.0084

unmitigated 0.1603 ± 0.2251 ± 0.2454 ± 0.3704 ±

0.0032 0.0041 0.0049 0.0068

Analytical ⟨V ⟩ -0.0701 -0.2598 -0.3746 -0.4905

value

ibmq lima error −0.0655 ± −0.2041 ± −0.2744 ± −0.4091 ±

mitigated 0.0012 0.0031 0.0063 0.0063

unmitigated −0.0538 ± −0.1471 ± −0.1233 ± −0.2737 ±

0.0011 0.0025 0.0041 0.0046
ibmq jakarta error −0.0515 ± −0.2348 ± −0.3255 ± −0.4469 ±
mitigated 0.0022 0.0056 0.0112 0.0112

unmitigated −0.0338 ± −0.1371 ± −0.0750 ± −0.2229 ±

0.0021 0.0046 0.0075 0.0083

ibm hanoi error −0.1136 ± −0.2820 ± −0.3497 ± −0.5512 ±

mitigated 0.0013 0.0031 0.0063 0.0063

unmitigated −0.1061 ± −0.2494 ± −0.2704 ± −0.4767 ±

0.0011 0.0022 0.0034 0.0038

ibm cairo error −0.0497 ± −0.1968 ± −0.2569 ± −0.3804 ±

mitigated 0.0013 0.0031 0.0063 0.0063

unmitigated −0.0471 ± −0.1682 ± −0.1733 ± −0.3089 ±

0.0012 0.0026 0.0038 0.0045

ibm auckland error −0.0138 ± −0.0854 ± −0.0591 ± −0.1887 ±

mitigated 0.0012 0.0032 0.0063 0.0063

unmitigated −0.0113 ± −0.0665 ± −0.0046 ± −0.1412 ±

0.0012 0.0027 0.0044 0.0049

ibmq montreal error −0.0157 ± −0.1207 ± −0.1275 ± −0.1967 ±

mitigated 0.0022 0.0056 0.0112 0.0112

unmitigated −0.0091 ± −0.0764 ± −0.0043 ± −0.0926 ±

0.0021 0.0048 0.0079 0.0089

Analytical ⟨E1⟩ -0.0180 -0.0726 -0.1147 -0.1425

value

ibmq lima error 0.0078 ± −0.0107 ± −0.0217 ± −0.0501 ±

mitigated 0.0034 0.0045 0.0071 0.0079

unmitigated 0.0757 ± 0.0950 ± 0.1715 ± 0.1565 ±

0.0054 0.0035 0.0050 0.0060

ibmq jakarta error 0.0221 ± −0.0330 ± −0.0764 ± −0.1079 ±

mitigated 0.0059 0.0079 0.0125 0.0140

unmitigated 0.0514 ± 0.1604 ± 0.2615 ± 0.2642 ±

0.0030 0.0064 0.0091 0.00111

ibm hanoi error 0.065 ± 0.0034 0.0436 ± 0.0779 ± 1.2481 ± 0.015

mitigated 0.0044 0.0071

unmitigated 0.0950 ± 0.0933 ± 0.1674 ± 1.0566 ± 0.015

0.0022 0.0021 0.0046

ibm cairo error 0.0177 ± −0.0315 ± 0.0010 ± −0.0245 ±

mitigated 0.0035 0.0044 0.0070 0.0079

unmitigated 0.0433 ± 0.0143 ± 0.0897 ± 0.0648 ±

0.0018 0.0034 0.0047 0.0058

ibm auckland error 0.1080 ± 0.1149 ± 0.5877 ± 1.2072 ±

mitigated 0.0034 0.0045 0.0031 0.0047

unmitigated 0.1341 ± 0.154 ± 0.0035 0.6364 ± 1.2236 ±

0.0021 0.0031 0.0047

ibmq montreal error 0.0740 ± 0.0411 ± 0.0645 ± 0.0849 ±

mitigated 0.0060 0.0079 0.0057 0.0140

unmitigated 0.1512 ± 0.1487 ± 0.2411 ± 0.2778 ±

0.0038 0.0063 0.0093 0.0112
TABLE IV: Results by ibmq lima, ibmq jakarta,
ibmq hanoi, ibm cairo, ibm auckland, ibmq montreal.