Reverse quantum annealing approach to portfolio optimization problems

Abstract

We investigate a hybrid quantum-classical solution method to the mean-variance portfolio optimization problems. Starting from real financial data statistics and following the principles of the Modern Portfolio Theory, we generate parametrized samples of portfolio optimization problems that can be related to quadratic binary optimization forms programmable in the analog D-Wave Quantum Annealer 2000Q TM . The instances are also solvable by an industry-established Genetic Algorithm approach, which we use as a classical benchmark. We investigate several options to run the quantum computation optimally, ultimately discovering that the best results in terms of expected time-to-solution as a function of number of variables for the hardest instances set are obtained by seeding the quantum annealer with a solution candidate found by a greedy local search and then performing a reverse annealing protocol. The optimized reverse annealing protocol is found to be more than 100 times faster than the corresponding forward quantum annealing on average.

Leveraging quantum annealing for large MIMO processing in centralized radio access networks

“…There are several improvements over the design we have evaluated here. First, we anticipate that further optimization of |J F |, T a , and s p as well as new QA techniques such as reverse annealing [68] may close the gap to Opt performance. Second, there are changes in QA architecture expected in annealers due this year [21] featuring qubits with 2× the degree of Chimera, 2× the number of qubits and with longer range couplings.…”

“…Our motivation for considering Opt is that it provides a bound to what can be achieved by the methods that seek to optimize machine parameter settings instance by instance [68,70], currently under investigation. With our traces we compute BER as a function of N a using Eq.…”

Noise amplification at spin-glass bottlenecks of quantum annealing: A solvable model

“…Quantum Annealing (QA), a quantum heuristic for approximately solving NP-hard binary optimization problems, is already in commercial use [1][2][3][4][5][6][7] in machine learning and artificial intelligence applications. The algorithm works by mapping Quadratic Unconstrained Binary Optimization (QUBO) problems to the problem of solving for the ground state of a spin glass Hamiltonian.…”

Gradient Descent for Gaussian Processes Variance Reduction

“…Examples include image recognition [15], bayesian network structure learning [16], fault detection and diagnosis [17], training a binary classifier [18] and portfolio optimization [19,20,21]. Moreover, in the context of mathematical logic, Bian et al [22] propose a QUBO formulation to tackle the maxSAT problem, an optimization extension of the well known SAT (boolean satisfiability) problem [23].…”