Quadratic Unconstrained Binary Optimization (QUBO) can be seen as a generic language for optimization problems. QUBOs attract particular attention since they can be solved with quantum hardware, like quantum annealers or quantum gate computers running QAOA. In this paper, we present two novel QUBO formulations for k-SAT and Hamiltonian Cycles that scale significantly better than existing approaches. For k-SAT we reduce the growth of the QUBO matrix from O(k) to O(log(k)). For Hamiltonian Cycles the matrix no longer grows quadratically in the number of nodes, as currently, but linearly in the number of edges and logarithmically in the number of nodes.
We present these two formulations not as mathematical expressions, as most QUBO formulations are, but as meta-algorithms that facilitate the design of more complex QUBO formulations and allow easy reuse in larger and more complex QUBO formulations.

Based on the quantum-assisted genetic algorithm (QAGA) [11] and related approaches we introduce several modifications of QAGA to search for more promising solvers on (at least) graph coloring problems, knapsack problems, Boolean satisfiability problems, and an equal combination of these three. We empirically test the efficiency of these algorithmic changes on a purely classical version of the algorithm (simulated-annealing-assisted genetic algorithm, SAGA) and verify the benefit of selected modifications when using quantum annealing hardware. Our results point towards an inherent benefit of a simpler and more flexible algorithm design.

Archetypes are those extreme values of a data set that can jointly represent all other data points. They often have descriptive meanings and can thus contribute to the understanding of the data. Such archetypes are identified using archetypal analysis and all data points are represented as convex combinations thereof. In this work, archetypal analysis is linked with quantum annealing. For both steps, i.e. the determination of archetypes and the assignment of data points, we derive a QUBO formulation which is solved on D-Wave’s 2000Q Quantum Annealer. For selected data sets, called toy and iris, our quantum annealing-based approach can achieve similar results to the original R-package archetypes.

28th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN 2020)

Many industrially relevant problems can be deterministically solved by computers in principle, but are intractable in practice, as the seminal P/NP dichotomy of complexity theory and Cobham’s thesis testify. For the many NP-complete problems, industry needs to resort to using heuristics or approximation algorithms. For approximation algorithms, there is a more refined classification in complexity classes that goes beyond the simple P/NP dichotomy. As it is well known, approximation classes form a hierarchy, that is, FPTAS  PTAS  APX  NPO. This classification gives a more realistic notion of complexity but—unless unexpected breakthroughs happen for fundamental problems like P = NP or related questions— there is no known efficient algorithm that can solve such problems exactly on a realistic computer. Therefore, new ways of computations are sought. Recently, considerable hope was placed on the possible computational powers of quantum computers and quantum annealing (QA) in particular. However, the precise benefits of such a drastic shift in hardware are still unchartered territory to a good extent. Firstly, the exact relations between classical and quantum complexity classes pose many open questions, and secondly, technical details of formulating and implementing quantum algorithms play a crucial role in real-world applications. Guided by the hierarchy of classical optimisation complexity classes, we discuss how to map problems of each class to a quantum annealer. Those problems are the Minimum Multiprocessor Scheduling (MMS) problem, the Minimum Vertex Cover (MVC) problem and the Maximum Independent Set (MIS) problem. We experimentally investigate if and how the degree of approximability influences implementation and run-time performance. Our experiments indicate a discrepancy between classical approximation complexity and QA behaviour: Problems MIS and MVC, members of APX respectively PTAS, exhibit better solution quality on a QA than MMS, which is in FPTAS, even despite the use of preprocessing the for latter. This leads to the hypothesis that traditional classifications do not immediately extend to the quantum annealing domain, at least when the properties of real-world devices are taken into account. A structural reason, why FPTAS problems do not show good solution quality, might be the use of an inequlity in the problem description of the FPTAS problems. Formulating those inequalities on a quantum hardware (mostly done by formulating a Quadratic Unconstrained Binary optimisation (QUBO) problem in form of a matrix) requires a lot of hardware space which makes finding an optimal solution more difficult. Reducing the density of a QUBO is possible by appropriately pruning QUBO matrices. For the problems considered in our evaluation, we find that the achievable solution quality on a real-world machine is unexpectedly robust against pruning, often up to ratios as high as 50% or more. Since quantum annealers are probabilistic machines by design, the loss in solution quality is only of subordinate relevance, especially considering that the pruning of QUBO matrices allows for solving larger problem instances on hardware of a given capacity. We quantitatively discuss the interplay between these factors.

1st International Symposium on Applied Artificial Intelligence (ISAAI’19)