In this paper, we present a hybrid genetic-hierarchical algorithm for the solution of the quadratic assignment problem. The main distinguishing aspect of the proposed algorithm is that this is an innovative hybrid genetic algorithm with the original, hierarchical architecture. In particular, the genetic algorithm is combined with the so-called hierarchical (self-similar) iterated tabu search algorithm, which serves as a powerful local optimizer (local improvement algorithm) of the offspring solutions produced by the crossover operator of the genetic algorithm. The results of the conducted computational experiments demonstrate the promising performance and competitiveness of the proposed algorithm.

In this paper, an improved hybrid genetic-hierarchical algorithm for the solution of the quadratic assignment problem (QAP) is presented. The algorithm is based on the genetic search combined with the hierarchical (hierarchicity-based multi-level) iterated tabu search procedure. The following are two main scientific contributions of the paper: (i) the enhanced two-level hybrid primary (master)-secondary (slave) genetic algorithm is proposed; (ii) the augmented universalized multi-strategy perturbation (mutation process)—which is integrated within a multi-level hierarchical iterated tabu search algorithm—is implemented. The proposed scheme enables efficient balance between intensification and diversification in the search process. The computational experiments have been conducted using QAP instances of sizes up to 729. The results from the experiments with the improved algorithm demonstrate the outstanding performance of the new proposed approach. This is especially obvious for the small- and medium-sized instances. Nearly 90% of the runs resulted in (pseudo-)optimal solutions. Three new best-known solutions have been achieved for very hard, challenging QAP instances.

Rationale. Solving combinatorial optimization problems by means of heuristic optimization algorithms is a highly important area of computational intelligence as well as artificial intelligence. Methodology. This paper presents empirical (computational) results obtained through experiments with a descent-ascent (D-A) principle-based heuristic optimization algorithm, primarily designed for solving combinatorial optimization problems. The descent-ascent algorithm—referred to briefly in this way—has its origins in the local search paradigm. A distinctive feature is that the minimization of the objective function of an optimization problem (i.e., descent) is combined with certain perturbations of solutions (i.e., ascents) to avoid a greedy/deterministic search behavior and, at the same time, premature convergence to suboptimal local optima. Results. The experiments conducted with this algorithm and the results obtained demonstrate a relatively high level of algorithmic efficiency in solving the well-known combinatorial problem—the quadratic assignment problem. Practical relevance of the research. The problem is well relevant in such areas as the the green economy, next-generation industry, digital transformation, renewable energy, sustainable logistics systems, and other socio-technical and information management contexts.