Landscape analysis and automated algorithm selection and tuning are still intensely studied topics in multi-objective optimisation. Landscape analysis captures a problem’s structure through numerical features, thereby supply-ing machine-learning models with rich information for making sound configuration decisions. Beyond this problem-centred view, it is also crucial to quantify—independently of the problem being solved—the intrinsic properties of an algorithm’s own components and to analyse how those properties affect performance. One such component is crossover. Widely investigated, the effectiveness of crossover depends on the operator’s ability to mix information, the specific characteristics of the problem, and the population diversity that emerges from the algorithm’s dynam-ics. This work concentrates on binary encodings and introduces a method for exploring the link between crossover features and the performance of a multi-objective evolutionary algorithm on problem subclasses whose variable in-teractions follow either random or nearest-neighbour patterns and whose scales differ. Using regression models, we identify the crossover features that are relevant to performance within each subclass and show how the influential features change as problem size varies.

In this paper, we propose a method for designing organic molecules with high light absorption, suitable for organic thin-film solar cells. By fragmenting existing compounds into sub-compounds and reassembling them, we generate candidate structures in an evolutionary algorithm framework. To evaluate these candidates efficiently, we use Quantum Deep Field (QDF), a deep learning method based on density functional theory, instead of conventional simulations with Gaussian16. This method makes it possible to perform a large-scale evolutionary search within a realistic time frame.

Through multiple experiments with various population sizes and offspring parameters, we identified an opti-mal parameter configuration for our evolutionary algorithm. We then confirmed that the detailed quantum chemical simulations (Gaussian16) show that the best-performing compounds discovered with our method have spectra with higher light absorption than several conventional organic solar cell materials. Furthermore, our analysis shows that expanding the search space and incorporating additional properties (e.g., absorption wavelength) could further en-hance the quality of discovered compounds. These findings highlight the potential of combining deep learning-based quantum approximations with evolutionary computation, opening new avenues for efficient and data-driven design of advanced photovoltaic materials.

Game theory provides a mathematical framework for analyzing interactive situations between multiple decision-makers. In the field of cybersecurity, it is frequently applied to model the interactions between attackers and defenders, with the goal of developing effective defensive strategies. However, the complexity of the models often leads to exact methods being impractical, especially when the players have multiple objectives and continuous decision spaces. Also, intrinsic features of the game may prevent the use of existing numerical methods. In this work, we present a generic framework for approximating Nash equilibria, the model solutions, in continuous games of simultaneous decision with multi-objective players. The framework is based on the decomposition of the decision space into one set of optimization problems for each player, the concurrent solution of these problems using numerical methods such as evolutionary algorithms, and the merging of the Pareto-optimal solutions from different players into a single set of Nash equilibria approximations. We demonstrate the effectiveness of the framework by applying it to an extension of the widely studied FlipIt game, using standard, unmodified evolutionary algorithms. The results show that the resulting approximations improve in diversity and accuracy as the number of optimization problems increases, as prescribed by the framework. This points to the robustness of the proposed method, which shall appear familiar to both the game-theoretic and the evolutionary computation communities. It has the potential to be used as a general-purpose framework for other games.