An online advertisement optimization, which can be represented by a combinatorial optimization problem is performed using D-Wave 2000Q, a quantum annealing machine. To optimize the online advertisement allocation optimization, we introduce a generalized version of the Markowitz mean-variance model which is a basic model of portfolio optimization. The obtained optimization performance using D-Wave 2000Q is higher than that using the greedy method which is a conventional method. Additionally, to conveniently use Ising machines including a quantum annealing machine, new software called PyQUBO is developed. The first half of the paper gives a review of several combinatorial optimization problems and how to represent them using the Ising model or the quadratic unconstrained binary optimization (QUBO) form. We show the results of the online advertisement allocation optimization and the explanation of PyQUBO in the last half of the paper.
An online advertisement optimization, which can be represented by a combinatorial optimization problem is performed using D-Wave 2000Q, a quantum annealing machine. To optimize the online advertisement allocation optimization, we introduce a generalized version of the Markowitz mean-variance model which is a basic model of portfolio optimization. The obtained optimization performance using D-Wave 2000Q is higher than that using the greedy method which is a conventional method. Additionally, to conveniently use Ising machines including a quantum annealing machine, new software called PyQUBO is developed. The first half of the paper gives a review of several combinatorial optimization problems and how to represent them using the Ising model or the quadratic unconstrained binary optimization (QUBO) form. We show the results of the online advertisement allocation optimization and the explanation of PyQUBO in the last half of the paper.
Quantum annealing is getting increasing attention in combinatorial optimization. The quantum processing unit by D-Wave is constructed to approximately solve Ising models on so-called Chimera graphs. Ising models are equivalent to quadratic unconstrained binary optimization (QUBO) problems and maximum cut problems on the associated graphs. We have tailored branch-and-cut as well as semidefinite programming algorithms for solving Ising models for Chimera graphs to provable optimality and use the strength of these approaches for comparing our solution values to those obtained on the current quantum annealing machine, D-Wave 2000Q. This allows for the assessment of the quality of solutions produced by the D-Wave hardware. In addition, we also evaluate the performance of a heuristic by Selby. It has been a matter of discussion in the literature how well the D-Wave hardware performs at its native task, and our experiments shed some more light on this issue. In particular, we examine how reliably the D-Wave computer can deliver true optimum solutions and present some surprising results.
We present PyQUBO, an open-source Python library for constructing quadratic unconstrained binary optimizations (QUBOs) from the objective functions and the constraints of optimization problems. PyQUBO enables users to prepare QUBOs or Ising models for various combinatorial optimization problems with ease thanks to the abstraction of expressions and the extensibility of the program. QUBOs and Ising models formulated using PyQUBO are solvable by Ising machines, including quantum annealing machines. We introduce the features of PyQUBO with applications in the number partitioning problem, knapsack problem, graph coloring problem, and integer factorization using a binary multiplier. Moreover, we demonstrate how PyQUBO can be applied to production-scale problems through integration with quantum annealing machines. Through its flexibility and ease of use, PyQUBO has the potential to make quantum annealing a more practical tool among researchers.
The quantum annealing devices, which encode the solution to a computational problem in the ground state of a quantum Hamiltonian, are implemented in D-Wave systems with more than 2,000 qubits. However, quantum annealing can solve only a classical combinatorial optimization problem such as an Ising model, or equivalently, a quadratic unconstrained binary optimization (QUBO) problem. In this paper, we formulate the QUBO model to solve elliptic problems with Dirichlet and Neumann boundary conditions using the finite element method. In this formulation, we develop the objective function of quadratic binary variables represented by qubits and the system finds the binary string combination minimizing the objective function globally. Based on the QUBO formulation, we introduce an iterative algorithm to solve the elliptic problems. We discuss the validation of the modeling on the D-Wave quantum annealing system.
In this note, we describe an experiment on portfolio optimization using the Quadratic Unconstrained Binary Optimization (QUBO) formulation. The dataset we use is taken from a real-world problem for which a classical solution is currently deployed and used in production. In this work, carried out in a collaboration between the Raiffeisen Bank International (RBI) and Reply, we derive a QUBO formulation, which we solve using various methods: two D-Wave hybrid solvers, that combine the employment of a quantum annealer together with classical methods, and a purely classical algorithm. Particular focus is given to the implementation of the constraint that requires the resulting portfolio’s variance to be below a specified threshold, whose representation in an Ising model is not straightforward. We find satisfactory results, consistent with the global optimum obtained by the exact classical strategy. However, since the tuning of QUBO parameters is crucial for the optimization, we investigate a hybrid method that allows for automatic tuning.
Recently, Ising machines have garnered attention as accelerators for resolving combinatorial optimization problems by converting them into Quadratic Unconstrained Binary Optimization (QUBO) problems. Ising machines have been rapidly developed with various implementations. For instance, there are different Ising machines like quantum annealing machines that use analog quantum circuits, simulated annealing machines that use digital processors like CPU, GPU, and VE, and digital annealing machines that use digital circuits such as CMOS and FPGA. However, the characteristics of Ising machines have not been fully comprehended, because a standardized evaluation method and a commonly used benchmark program for Ising machines have not been established yet. This research focuses on combinatorial clustering as a benchmark program for the evaluation of Ising machines, aiming to investigate the characteristics of Ising machines.
In this work, we introduce a new workflow to solve portfolio optimization problems on annealing platforms. We combine a classical preprocessing step with a modified unconstrained binary optimization (QUBO) model and evaluate it using simulated annealing (classical computer), digital annealing (Fujitsu’s Digital Annealing Unit), and quantum annealing (D-Wave Advantage). Starting from Markowitz’s theory on portfolio optimization, our classical preprocessing step finds the most promising assets within a set of possible assets to choose from. We then modify existing QUBO models for portfolio optimization, such that there are no limitations on the number of assets that can be invested in. Furthermore, our QUBO model enables an investor to also place an arbitrary amount of money into each asset. We apply this modified QUBO to the set of promising asset candidates we generated previously via classical preprocessing. A solution to our QUBO model contains information about what percentage of the whole available capital should be invested into which asset. For the evaluation, we have used publicly available real-world data sets of stocks of the New York Stock Exchange as well as common ETFs. Finally, we have compared the respective annealing results with randomly generated portfolios by using the return, variance, and diversification of the created portfolios as measures. The results show that our QUBO formulation is capable of creating well-diversified portfolios that respect certain criteria given by an investor, such as maximizing return, minimizing risk, or sticking to a certain budget.