CIGRE-759 2024 CIGRE Canada Conference & Expo

Winnipeg, MB, Oct. 28 – 31, 2024

A Power System Use Case for Quantum Computing - Optimally Partitioning a

Large Network

H.S. ATTANAYAKA1*, U.D. ANNAKKAGE,

University of Manitoba1
Canada

SUMMARY

Transient stability (TS) is the commonly used simulation platform for analyzing slow
electromechanical transients within a timeframe of around 10 seconds following a contingency
event and is arguably the most suitable method for studying very large, interconnected power
networks.
Near real-time dynamic security assessment (DSA) enables the computation of system stability
for current operating conditions as they occur and the entire DSA cycle must be completed
within a 10–15-minute timeframe. A common practice involves dividing the system into a study
area with detailed modeling and an external area as an equivalent model. A more accurate
method involves splitting the original system into subsystems that are solved concurrently. Yet,
even for smaller systems, the computing time required for partitioning is significantly large,
highlighting the need for more efficient computational methods for much larger systems.
Therefore, one strategy to increase the simulation speed is the distribution of computation
among several CPU-based servers or the utilization of GPUs.
Going beyond classical computer solutions, the concept of quantum computing has risen as a
groundbreaking technology for many challenges including power system problems.
Therefore, in this paper, the application of quantum computing to solve a selected power system
optimization problem will be presented. The discussion will include the comparison of classical
computers and quantum computers, the evolution of quantum computers through time, the
relevance of quantum computing for power system problems, leading technologies of quantum
computing and their progress, and the challenges associated with quantum computing.
Among different commercially available technologies, IBM has introduced the IBM quantum
experience (IQX) which provides anyone with the opportunity to experiment with a quantum
computer for free through a user interface. IBM provides qiskit which is a Python package used
for developing quantum algorithms, running experiments on real devices, and visualizing
results. In this paper, how to access an IBM quantum computer and understand qiskit and its
functions will be presented.

KEYWORD
Quantum computing, Power system, Network partitioning, IBM, Qiskit

*
[email protected]
1. OVERVIEW
1.1. Introduction
Reliable operation and security of the power system have become a crucial task, due to the
highly nonlinear and complicated nature of the modern power system with inverter-based
resources, power electronic-based devices, and electric transportation. The analysis
determining whether the power system adheres to specific security and reliability criteria under
steady state and transient time frames is defined as dynamic security assessment (DSA). The
computation of the study area which involves the speed of computation and comprehensive and
accurate computation capabilities is a key component of DSA. Since the power system
continuously expands, the speed of computation has become a challenge. As a solution to this
issue, splitting the original network into subsystems and the distribution of subsystems among
different processors (multiple CPUs, GPUs, and hybrid GPU-CPU architecture) has gained
remarkable attention in the last couple of years. Yet, exploring other innovative approaches is
always worth a venture.
Meanwhile, inspecting quantum mechanisms to solve the limitation of classical computation
has attracted more and more attention in both academia and industry. But, quantum computing
to address power system challenges is yet to be explored.
Therefore, in this paper, an attempt is made to explain the field of QC and its capability of
solving optimal network splitting problem.
The structure of the paper is as follows. An overview of QC is presented in section 1. Then
several examples of QC in power systems are explained. In section 3, IBM QC services for the
public and how to access their real quantum devices are described. An example of applying QC
to solve an optimization problem (Max-cut problem) using an IBM real device and points of
formulating optimal network splitting method on a noisy quantum device is shown in section
4.
1.2. Quantum Computing
In this section, QC technology development, theory, quantum algorithms, leading technologies,
general applications, and challenges are described.
1.3. Quantum technology development
The idea of quantum computers initially gained the attention in 1980s owing to Richard
Feynman's views on the simulation of quantum effects, the mechanism of quantum algorithms,
and quantum computer functions [1],[2]. In later years, those views contributed to the
development of quantum algorithms including Deutsch–Jozsa, Shor, and Grover. In addition to
the theoretical growth of quantum computers, important milestones of the physical
implementation of quantum computers such as the development of the first commercial
quantum computer and the development of IBM cloud-accessible quantum sources are shown
in Figure 1.

Figure 1. Evolution of quantum computing [3]

1.4. Theory
The fundamental block or the basic unit of information in a classical computer is a Bit which
can take two possible values either 0 or 1 one at a time. In a quantum computer, the fundamental
block is a qubit, and from an abstract mathematical point of view, two possible states for a qubit
are state |0⟩ and |1⟩. The difference between bits and qubits is that a qubit can be in a state
other than |0⟩ or |1⟩. It is feasible to form linear combinations of states, called superpositions
[4].
|𝜓⟩ = 𝛼|0⟩ + 𝛽|1⟩ (1)
𝛼 and 𝛽 are complex numbers that determine the probability of having the quantum states |0⟩
or |1⟩, respectively. It means that if the state of the qubit is measured, the probability of state
|0⟩ and |1⟩ are found is |𝛼 2 | and |𝛽2 |, respectively[3].Visual representation of a qubit state
space is shown in Figure 2. Any state of a qubit corresponds to a point on the surface of the
Bloch sphere. Note that the information of a qubit state is through a measuring process, which
provides either |0⟩ or |1⟩ as the output depends on the probability |𝛼 2 | and |𝛽2 |.
In QC, there are 2 main concepts which are superposition and entanglement. As stated
previously, superposition is explained as the ability of a qubit to exist in many states at once
[5]. Entanglement refers to the correlation among two or more qubits. This property allows
separate quantum systems to become bound together so that their quantum state becomes a
single collective phenomenon rather than separate ones, and performing any manipulation on
one of them affects the state of the others. In other words, by measuring one entangled qubit,
the properties of its partners can be deduced without measuring. As the output of a quantum
computation is also superposition, obtaining the correct result will be a tricky task.

Figure 2. Bit vs qubit [3]

1.5. Quantum algorithms

Quantum computers can leverage the principles of quantum mechanics such as superposition,
interference, and entanglement, to their advantage using quantum algorithms and perform
computations that classical computers cannot do. The key purpose of quantum algorithms is to
achieve an expedition compared to any possible classical algorithm.
Grover’s algorithm:
This algorithm is developed to search entities in a database and provide a quantum speedup for
the problem of unstructured search. Assume there is a function 𝑓 ∶ {1,2, … . , 𝑁} → {0,1} and
there exists a unique 𝑎 ∈ {1,2, … , 𝑁} with 𝑓(𝑎) = 1. Given query access to the function f, the
unstructured search problem asks us to determine this value 𝑎.
Harrow-Hassidim-Lloyd (HHL) :
In 2009, Aram W. Harrow, Avinatan Hassidim, and Seth Lloyd proposed to solve linear systems
which is one of the most important algorithms.
Shor’s algorithm :
This algorithm is used to factorize very large integers efficiently. In integer factorization, it is
asked to find, for a given integer 𝑁, the set of all prime factors of 𝑁 and there is no efficient
classical algorithm for this problem yet.
Quantum Approximation Optimization Algorithm QAOA:
It is a hybrid quantum-classical algorithm [6] that generates approximate solutions for
combinatorial optimization problems. In other terms, a hybrid algorithm consists of a
parametrized quantum circuit, estimation of cost function using a quantum measurement, and
a classical optimizer to optimize the quantum circuit’s variational parameters. Combinatorial
optimization problems are a class of optimization problems that gives the best solution from a
finite set of possible solutions and it involves choosing a combination of elements from a finite
set to optimize for a particular objective function. A general example of such a problem (max-
cut problem) will be discussed in the later part of the paper.
1.6.General applications
Quantum computing is growing fast with new advances in multiple areas of applications. Both
experts in academia and industry are exploring practical problems that may be solved by
quantum computers effectively. Some of the applications are listed below [7], [8].

Material science
The characteristics of materials at the molecular and atomic level can be examined with high
accuracy. Therefore, by simulating and analyzing of properties of materials, innovative
materials with improved properties can be designed. Also, the optimal method of synthesizing
materials can be achieved using QC.
Cyber security:
One of the main concerns is the vulnerability of some encryption methods to QC. In contrast,
QC can be employed to develop more advanced and protected encryption. Another advantage
of QC is details of the security level of complex systems can be obtained through simulation
and texting with quantum computers.
Finance
In the field of finance, QC can be used to develop finance models, risk analysis, and optimize
portfolios. Even though conventional computers can perform such tasks, the time it takes is
much longer. It will lead financial institutes to make decisions with accurate data and to get
maximum profit.
Artificial intelligence and machine learning
In machine learning, training machine models require a large set of data and computational
resources, using QC this process can be accelerated. Currently, researchers are interested in
quantum-assisted neural networks in terms of advancing the training and interference of neural
networks.
Pharmaceuticals
By modeling the characteristics of molecules and predicting their interaction with proteins in
the body, the potential of QC can be harnessed. It can lead to the development of new drugs
and better medical treatment.
Weather forecasting and climate modeling
QC can be employed to study natural phenomena such as the carbon cycle, global atmospheric
circulation, and oceanic circulation. Therefore, more accurate predictions of climate change
can be obtained. Also, actions that are required to be undertaken to mitigate severe climate
change scenarios can be decided based on such predictions.

1.7. Leading technologies

Figure 3 shows some of the leading companies and their highest number of bits along with
commercially available software packages developed by them as of 2024. There are two key
technologies currently in the market naming gate-based (in Figure 3, they are marked in black)
and quantum annealing technology (marked in red). In Gate-based technology, quantum
algorithms are implemented using quantum gates, analogous to classical logic gates in classical
computing. Quantum annealing is analogous to simulated annealing which is an optimization
technique that can ‘escape’ local best and converge towards global best and it simply uses
quantum physics to find low-energy states of a problem.
Also, the physical structure of qubits of each technology is mentioned. Superconducting qubits
are made of superconducting materials operating at extremely low temperatures and are
manipulated by microwave pulses [9]. Such type of qubits method exhibits long coherence
times [10]. Trapped ions refer to charged atomic particles that are confined and manipulated
using electromagnetic fields and have high-fidelity [11]. In neutral atom quantum computers,
to trap them in microscopic arrays neutral altra-cooled atoms in optical lattices are controlled
by lasers.
It can be stated that IBM is the corporate giant in the quantum computing industry offering a
cloud-based IBM Quantum Lab platform with public and premium access. (it will be described
later in this paper). Google is another leading QC company that provides open-source packages,
such as Cirq, open-Fermion, TensorFlow quantum [12] for customers to applications with noisy
quantum devices.
It is worth mentioning that having a high number of qubits essentially does not mean that a
particular quantum computer is the best. Because having a lower number of quality qubits with
less noise is much more effective than having a high number of qubits with low-quality qubits
with high noise. Therefore, to determine how advanced a particular quantum technology is,
IBM has designed a matrix called Quantum Volume (QV) based on the number of qubits, the
error rate of qubits connectivity of qubits, and so on. As of April 2024, Quantinuum has the
highest QV which is 220 and number of qubits is 56. ATOM computing has the quantum
computer with the highest number of qubits which is 1225.

Figure 3. Leading companies based on the physical structure of qubits and their number of qubits
1.8.Challenges
Despite all the advantages of quantum computers over classical computers, there are several
challenges associated with quantum computing. Additionally, the current state of QC is the era
of noisy intermediate-scale quantum (NISQ) [13] computing, because qubits in the existing
devices are prone to errors, have short coherent times, and have a limited number of qubits.
a. The primary problem is its vulnerability which is reflected in several ways:
I. Quantum decoherence which is described as the degradation of the information in
a qubit as it interacts with the environment. During the calculation phase of a
quantum calculation, a slight disturbance in the environment (ex: electromagnetic
 interference, imperfections in the hardware, and temperature fluctuations) can
cause the quantum computation to collapse [14].
II. Error correction in quantum computing is significantly more challenging compared
to classical computing as quantum computers are more volatile and difficult to
control.[15].
b. A lack of expert knowledge coupled with pessimistic views on QC due to extensive
technical challenges is a prominent barrier to quantum computing growth [16]. This may
have discouraged some researchers from being involved in QC disciplines[17].

2. Quantum application in power system

Some potential areas of power systems that can be augmented on NISQ devices are unit
commitment, power flow analysis, contingency analysis, and fault analysis.
Alas, experts have not yet fully illustrated the potential of the quantum advantage in solving
large-scale power system problems due to the issue associated with NISQ quantum computers.

2.1. Unit Commitment

Unit commitment is explained as the method of optimally utilizing generation units of available
power plants at a minimum cost to satisfy electricity demand. This problem becomes
burdensome with having different types of generation units in the systems with distinct
generation costs ( especially with invert-based technologies), response time of generator units,
and geographical location of power plants. Therefore, as a solution to the high computational
demand associated with unit commitment, in [18] , employment of quantum approximation
optimization algorithm (QAOA) is presented.
2.2. Optimal power flow
Among different power flow methods, the fast decoupled method is acknowledged as the most
computationally efficient method. To further improve the efficiency, attempts are been made
based on quantum algorithms. In [19], power flow solution based on the fast decupled method
and HHL algorithm is presented and implemented on a noise-free quantum simulator. But, there
is no implementation on real quantum devices and the limitations of the proposed method on
NISQ devices are not mentioned. In [20], HHL-based optimal power flow problem was tested
on an IBM real quantum device. It was observed that the problem converged to the same
solution as the classical fast-decoupled power flow algorithm on the NISQ device. But a
limitation is it takes a considerably large number of iterations to converge which implies that
the noise of real devices negatively impacts the convergence speed.
2.3. Fault analysis
Fault diagnosis is a crucial task in power system studies because of the continuously growing
demand without sufficient improvement in generation capacities. In [21], a hybrid QC-based
deep learning method is implemented using the IEEE-30 bus system as the test system. It says
that the proposed method shows advantages such as faster response time, and scalability without
loss of performance compared to the classical fault diagnosis methods.
2.4. Grid partitioning
In [22] integer slack (IS) and binary expansion method to formulate a grid partitioning problem
as a quantum unconstrained binary optimization problem (QUBP) based on quantum annealing
is proposed. A prospective future task is to formulate this problem on a gate-based quantum
computer (such as IBM, Google).

3. IBM quantum computing

In May 2016, IBM released IBM Quantum Lab; a cloud-based user interface that allows users
to interact with real quantum computers for free. Through this platform, available backends
and their status, information on jobs that are executed, and details of jobs that are in the queue
can be obtained. A person who is not a member of the IBM Q network can access 4 quantum
computers with a total monthly usage of 10 minutes for free. Currently, they are
ibm_sherbrooke, ibm_brisbane, ibm_osaka, and ibm_kyoto and all are 127-qubits devices.
Furthermore, the physical arrangement of each device is illustrated with the error rate of each
qubit. Also, it provides tutorials and courses including basics of quantum information,
fundamentals of quantum algorithms, and navigation with Qiskit.
Qiskit (Quantum Information Science Kit) is a Python package developed by IBM that is used
for developing quantum algorithms, creating noise models, running experiments on real
devices, and visualizing results. It is comprised of 4 modules named Terra (earth), Aer (air),
Ignis (fire), and Aqua (water) each having a specific functionality and role [23].
Terra is at the foundation of all the other elements, hence its translated name, Earth. It contains
all the necessary components required to create quantum circuits, to generate and schedule the
pulses that are sent to operate on the qubits. It can execute quantum circuits, and specify which
device or simulator will execute the circuit. Aer provides a framework that can be used to
develop debugging tools. These tools help replicate the characteristics of a quantum system by
simulating the noise that affects not just the qubit, but also the environment and computations.
Ignis Primarily focuses on errors, typically noise models that affect qubits. It has tools that can
help mitigate the effects of noise and errors, such as providing error correction. Aqua is built
for those who wish to simply use the quantum algorithms by integrating Aqua into their
classical applications as simply as possible. In this case, the user level is a researcher who is
interested in leveraging what is available rather than having to construct everything from the
circuit level up.
4. Application of QC in optimization problems
Though the quantum advantage over classical computations exists for the task of solving
optimization problems, NISQ technology might not be adequate to exhibit that advantage. The
current stage of solving the optimization problem is a hybrid quantum-classical approach.
5. Examples
5.1. Max-cut problem
Max-cut problem is NP-hard and a well-known combinatorial optimization problem [24]. It is
believed that there is no exact classical method to solve max-cut in polynomial time. Some of
the best-known approximation classical algorithms for solving Max-cut with near-optimal
solutions are Goemans and Williamson, greedy algorithms, randomized approximation
algorithms, and local search algorithms [25], [26].
One thing to keep in mind is that as mentioned in [13], the expectation of solving NP-hard
problems using QC is two-fold, (1). It is not expected quantum computers to be able to solve
efficiently worst-case instances of NP-hard problems; (2) It’s conceivable (though not
guaranteed) that quantum devices will be able to find better approximate solutions or find such
approximate solutions faster. The emerging concept of solving optimization problems using
NISQ-era quantum computers is of a hybrid quantum-classical nature. Therefore, in this paper,
an example of solving max-cut problems using QAOA is described (This is not a novel work).
To realize to what extent the obtained solution is optimized an approximation ratio is introduced
[13].
Consider an undirected graph 𝐺 = (𝑉, 𝐸), consists of nodes V (vertices) or (weighted) edges
𝐸. If edges are weighted, then 𝑤𝑖,𝑗 > 0 , 𝑤𝑖,𝑗 = 𝑤𝑗,𝑖 for (𝑖, 𝑗) ∈ 𝐸. By assigning 𝑥𝑖 = 0 or
𝑥𝑖 = 1 to each node, the objective is minimizing the following cost function. The cost function
to be optimized is the sum of the weights of edges connecting the nodes in the two distinct sub-
sets. For such Max-cut problem, QAOA requires n = |V | number of qubits.
In this paper, the steps of utilizing IBM quantum device to solve the Max-cut problem shown
in Figure 4 are described.
 Figure 4. A graph containing 5 edges and 4 nodes

Step 1: To find a solution to this problem on a quantum computer, the first step is the derivation
of the objective function 𝐶 (𝑥) based on the problem definition.
𝐶 (𝑥) = ∑ 𝑤𝑖,𝑗 𝑥𝑖 ( 1 − 𝑥𝑗 ) (2)
𝑖,𝑗=1
Step 2: Then the objective function is to be reformulated as a Quantum Unconstrained Binary
Optimization (QUBO) problem (see eqn 3 and 4). QUBO is a mathematical formulation used
to solve combinatorial optimization problems by leveraging quantum computing. In QUBO,
the objective function is formulated in binary variables (0/1).
𝐶 (𝑥) = ∑ 𝑥𝑖 𝑄𝑖,𝑗 𝑥𝑗 + ∑ 𝜇𝑖 𝑥𝑖 𝑤ℎ𝑒𝑟𝑒 𝜇𝑖 = ∑ 𝑤𝑖,𝑗 𝑎𝑛𝑑 𝑄𝑖𝑗 = −𝑤𝑖,𝑗 (3)
𝑖,𝑗=1 𝑖=1 𝑗=1
𝐶 (𝑥) = 𝑥 𝑇 𝑄𝑥 + 𝜇𝑇 𝑥 (4)

Step 3: Next classical inputs should be mapped to a quantum problem. To do that, the problem
is formulated as Hamiltonian problem. A classical Hamiltonian (an operator or matrix) is a
mathematical description of some physical system in terms of its energies and it returns the
energy for a particular state of a system. In gate-based quantum technologies named IBM, and
Google, Hamiltonian is utilized to guide the evolution of the quantum state of qubits toward the
solution.
𝐻𝐶 |𝑥⟩ is the Hamiltonian operator that encodes the cost function, 𝐶(𝑥) (see eqn (5)). It encodes
the cost in a manner such that the minimum expectation value of the operator corresponds to
the maximum number of edges between the nodes in two different groups. The ground state of
the Hamiltonian corresponds to the optimal solution of the optimization problem.
𝐻𝐶 |𝑥⟩ = 𝐶(𝑥)|𝑥⟩ (5)
As shown in eqn (6), each of the optimization variables is mapped into a qubit (to a spin
variable, 𝑧𝑖 ). Then spin operators are promoted to Pauli operators (𝑍𝑖 ).
(1 − 𝑧𝑖 )
𝑥𝑖 = , 𝑧𝑖 ∈ {−1,1} (6)
2

1 1 𝑄𝑖,𝑗 𝜇𝑖
𝐻𝐶 = ∑ 𝑄𝑖,𝑗 𝑍𝑖 𝑍𝑗 − ∑ (𝜇𝑖 + ∑ 𝑄𝑖,𝑗 ) 𝑍𝑖 + ( ∑ + ∑ ) (7)
4 2 4 2
𝑖,𝑗=1 𝑖=1 𝑗=1 𝑖,𝑗=1 𝑖=1
Step 4: The next step is to solve the above Hamiltonian in a quantum computer using a QAOA
circuit.
Step 5: Then optimization of the QAOA parameters using a classical optimizer is performed.
Step 6: Next, the solution by running the QAOA circuit with optimal parameters is obtained.
Further explaining steps 4, 5, and 6, a parametrized quantum state is repeatedly prepared and
measured on a quantum computer to estimate its average energy. Then, a classical optimizer,
running in a classical computer, uses such information to decide on the new parameters that are
then provided to the quantum computer. This process is iterated until certain convergence
criteria are met.
Step 7: Finally, the measurement of the qubits is done and the histogram is plotted (see Figure
5). For this example it shows that 00001 and 11110 stand out, which is correct. There are two
solutions because the labeling of the two partitions with '0' or '1' is arbitrary. The results are
mapped into the classical format which is shown in Figure 6.

Figure 5 Output of the Max-cut problem

Figure 6 Output of the Max-cut problem

As a conclusion to this section, it is said that the so-far-best classical algorithm given by
Goemans and Williamson achieved an approximation ratio of 0.951 [26]. In [27] researchers
have stated that they have achieved an approximation ratio of 0.956 using a quantum
computing-based method and said impossible to do better than this.
5.2.Power system test case: optimal network partitioning using a quantum computer
Existing network splitting methods, each having their strengths and limitations are graph
partitioning, optimization-based methods, geographical partitioning, and slow coherency
methods based on regional characteristics. key aspects to be considered when splitting a
network into subsystems are as follows.
a. The number of connections between the subsystems should be minimized.
b. It is desirable to split the preferring weak connections over strong connections.
c. The size of subsystems should be equal or nearly equal.
A possible approach to formulate the network splitting problem as a graph theory problem is
described below.
A power network can be naturally represented as a graph: vertices (nodes) represent buses, and
edges (links) represent electrical connections. QC graph partitioning can be divided into gate-
based technology and quantum annealing technology. The steps of partitioning the graph using
these methods are listed below.
i. Problem definition:
For given a graph consisting of 𝑛 nodes and 𝑣 vertices, partition the nodes into 𝑘 partitions
of equal or nearly equal nodes in each subgraph such that the number of cut edges is
minimized. Cut edge is defined as edges whose endpoints are in distinct subgraphs. Edge
weighted function can be the admittance between buses or the average power flow between
the buses. After the problem definition, the objective function is to be derived.
ii. Reformulation of the objective function as QUBO problem or Isling model:
Ising model, the objective function is formulated in variables (1/−1). Graph problems tend
to be more naturally modeled with 0/1-valued QUBO variables than the −1/+1-valued
Ising variables. Therefore, it is desirable to model using QUBO.
iii. Representation as QUBO Hamiltonian or Ising Hamiltonian:
The objective of this step is to mathematically represent the energy of the state of QUBO or
Isling model. In quantum annealing technologies such as D-wave, the objective is to find the
ground state of the Hamiltonian, representing the optimal solution.
iv. Select the quantum approach
a. Graph partitioning with quantum annealing technology (D-wave technology)
Research work based on this method is described here [28]. Quantum annealers, directly
solve the Ising model/ QUBO Hamiltonian by finding the ground state (lowest energy
state) of the Hamiltonian.
b. Graph partitioning with gate-based quantum technologies (IBM, Google)
As described in example 5.1, QAOA which involves a parameterized quantum circuit
and classical optimization is employed to solve the problem. This method also solves
the Hamiltonian to find the lowest energy state.
v. Map the solution into the classical format

6. Conclusion
In this paper, an overview of quantum computing is presented. A use case that describes how
to harness the potential of QC for the power system network splitting method is described.
A few conclusions that are derived based on the findings are mentioned below.
Network splitting is an NP-hard problem. But as this is the NISQ era, obtaining quantum
advantage for splitting a large power network will be difficult. For large-scale power systems,
a large number of qubits are required. With the increase of qubits, the error rate also further
increases, which may influence the outcome. Also, it will be inevitable to adopt effective error
strategies for error correction.
However, the commercially available technologies (IBM, D-wave, Google) provide QC
services (platforms, software packages) for researchers to experiment on real quantum devices.
Unfortunately, the waiting time to submit a job to a public IBM real quantum is considerably
high (minimum of 7 hours).
Furthermore, selecting a suitable quantum approach (gate-based technology or quantum
annealing) for optimal network partitioning problems is an open discussion. In terms of the
availability of resources (software packages, tutorials, examples, and courses), and the
navigability of the platform, IBM can be considered as the leading technology. But, in the
literature, the approach the researchers acquired is quantum annealing with D-wave technology.
Table 2 shows the existing literature on grid partitioning based on the graph theory method and
remarks.
 Table 1
Literature Technology Remarks
[28] • The D-Wave 2X machine • Test case: They have mentioned that
• 1095 qubits large graphs in the Graph Partitioning
• It is able to solve graphs of Archive are used.
limited size (⁓ 45 nodes)
[22] • The Advantage quantum • Test case: PEGASE 89-bus system
computer of D-wave • The proposed method is compared with
technology the spectral clustering (classical
• 5000+ qubits algorithms ) and quantum annealing
method proposed in [28] .
• It says this method shows good
robustness compared to spectral
clustering and the time cost is 53% that
of the method proposed in [28].

References

[1] R. P. Feynman, “Simulating physics with computers,” International Journal of

Theoretical Physics, vol. 21, no. 6–7, 1982, doi: 10.1007/BF02650179.
[2] P. Benioff, “The computer as a physical system: A microscopic quantum mechanical
Hamiltonian model of computers as represented by Turing machines,” J Stat Phys, vol.
22, no. 5, 1980, doi: 10.1007/BF01011339.
[3] S. Golestan, M. R. Habibi, S. Y. Mousazadeh Mousavi, J. M. Guerrero, and J. C.
Vasquez, “Quantum computation in power systems: An overview of recent advances,”
Energy Reports, vol. 9. 2023. doi: 10.1016/j.egyr.2022.11.185.
[4] M. a. Nielsen and I. L. Chuang, Quantum Computation and Quantum Information: 10th
Anniversary Edition. 2011. doi: 10.1017/CBO9780511976667.
[5] A. Giani and Z. Eldredge, “Quantum Computing Opportunities in Renewable Energy,”
SN Comput Sci, vol. 2, no. 5, 2021, doi: 10.1007/s42979-021-00786-3.
[6] “Quantum approximate optimization algorithm | IBM Quantum Learning.” Accessed:
Jun. 09, 2024. [Online]. Available: https://learning.quantum.ibm.com/tutorial/quantum-
approximate-optimization-algorithm
[7] F. Bova, A. Goldfarb, and R. G. Melko, “Commercial applications of quantum
computing,” EPJ Quantum Technology, vol. 8, no. 1. 2021. doi: 10.1140/epjqt/s40507-
021-00091-1.
[8] J. H. Markna, T. P. Palatia, S. Gohel, and B. Kataria, “Unveiling Advanced
Computational Applications in Quantum Computing: A Comprehensive Review,”
International Journal of Advanced Nano Computing and Analytics, vol. 2, no. 2, 2023,
doi: 10.61797/ijanca.v2i2.284.
[9] “What is a qubit? | IBM.” Accessed: Jun. 03, 2024. [Online]. Available:
https://www.ibm.com/topics/qubit
[10] F. Lu, “Several Ways to Implement Qubits in Physics,” in Journal of Physics:
Conference Series, 2021. doi: 10.1088/1742-6596/1865/2/022007.
[11] “What is Trapped Ions.” Accessed: Jun. 03, 2024. [Online]. Available:
https://www.quera.com/glossary/trapped-ions
[12] H. Liu and W. Tang, “Quantum computing for power systems: Tutorial, review,
challenges, and prospects,” Electric Power Systems Research, vol. 223. Elsevier Ltd,
Oct. 01, 2023. doi: 10.1016/j.epsr.2023.109530.
[13] J. Preskill, “Quantum computing in the NISQ era and beyond,” Quantum, vol. 2, 2018,
doi: 10.22331/q-2018-08-06-79.
[14] A. Banafa, Introduction to Quantum Computing. 2023. doi: 10.1201/9781003440239.
[15] P. Kaye, R. Laflamme, and M. Mosca, An Introduction to Quantum Computing. 2006.
doi: 10.1093/oso/9780198570004.001.0001.
[16] J. Singh and K. S. Bhangu, “Contemporary Quantum Computing Use Cases: Taxonomy,
Review and Challenges,” Archives of Computational Methods in Engineering, vol. 30,
no. 1. 2023. doi: 10.1007/s11831-022-09809-5.
[17] M. Dyakonov, “When will useful quantum computers be constructed? Not in the
foreseeable future, this physicist argues. Here’s why: The case against: Quantum
computing,” IEEE Spectr, vol. 56, no. 3, 2019, doi: 10.1109/MSPEC.2019.8651931.
[18] S. Koretsky et al., “Adapting Quantum Approximation Optimization Algorithm
(QAOA) for Unit Commitment,” in Proceedings - 2021 IEEE International Conference
on Quantum Computing and Engineering, QCE 2021, 2021. doi:
10.1109/QCE52317.2021.00035.
[19] F. Feng, Y. Zhou, and P. Zhang, “Quantum Power Flow,” IEEE Transactions on Power
Systems, vol. 36, no. 4, 2021, doi: 10.1109/TPWRS.2021.3077382.
[20] J. Ø. Brynjar Sævarsson, Spyros Chatzivasileiadis, Hjörtur Jóhannsson, “Quantum
Computing for Power Flow Algorithms: Testing on real Quantum Computers,” in 11th
Bulk Power Systems Dynamics and Control Symposium, Banff, Canada, 2024, p. 8.
[21] A. Ajagekar and F. You, “Quantum computing based hybrid deep learning for fault
diagnosis in electrical power systems,” Appl Energy, vol. 303, 2021, doi:
10.1016/j.apenergy.2021.117628.
[22] D. Wang, K. Zheng, F. Teng, and Q. Chen, “Quantum Annealing With Integer Slack
Variables for Grid Partitioning,” IEEE Transactions on Power Systems, vol. 38, no. 2,
2023, doi: 10.1109/TPWRS.2022.3229862.
[23] K. Neha and . A., “Quantum programming: Working with IBM’S qiskit tool,” The
Scientific Temper, vol. 14, no. 01, 2023, doi: 10.58414/scientifictemper.2023.14.1.11.
[24] “2.1 The Max-Cut Problem”.
[25] M. X. Goernans and D. P. Williamson, “878-Approximation algorithms for MAX CUT
and MAX 2SAT,” in Proceedings of the Annual ACM Symposium on Theory of
Computing, 1994. doi: 10.1145/195058.195216.
[26] M. X. Goemans and D. P. Williamson, “Improved Approximation Algorithms for
Maximum Cut and Satisflability Problems Using Semidefinite Programming,” Journal
of the ACM (JACM), vol. 42, no. 6, 1995, doi: 10.1145/227683.227684.
[27] Y. Hwang, J. Neeman, O. Parekh, K. Thompson, and J. Wright, “Unique Games hardness
of Quantum Max-Cut, and a conjectured vector-valued Borell’s inequality,” in
Proceedings of the Annual ACM-SIAM Symposium on Discrete Algorithms, 2023. doi:
10.1137/1.9781611977554.ch48.
[28] H. Ushijima-Mwesigwa, C. F. A. Negre, and S. M. Mniszewski, “Graph partitioning
using quantum annealing on the D-wave system,” in ITiCSE-WGR 2017 - Proceedings
of the 2017 ITiCSE Conference on Working Group Reports, 2018. doi:
10.1145/3149526.3149531.