This paper presents topology optimization for thermal cloaks expressed by level-set functions and explored using the covariance matrix adaptation evolution strategy (CMA-ES). Designed optimal configurations provide superior performances in steady-state conduction succeed realizing invisibility, despite structures being simply composed of iron aluminum without inhomogeneities caused employing metamaterials. To design cloaks, a prescribed objective function is used to evaluate difference...

We generate optimal topologies in the structural design of bifunctional cloaks manipulating heat flux and direct current, using topology optimization that incorporates both thermal conductivity electrical current. The cloak composed bulk isotropic materials is designed to restrain disturbances caused by an insulated obstacle minimizing difference between cloaked distributions referenced when no present. Our results show presented optimizations provide reproduce undisturbed temperature...

We introduce an acceleration for covariance matrix adaptation evolution strategies (CMA-ES) by means of adaptive diagonal decoding (dd-CMA). This endows the default CMA-ES with advantages separable without inheriting its drawbacks. Technically, we a [Formula: see text] that expresses coordinate-wise variances sampling distribution in DCD form. The can learn rescaling problem coordinates within linear number function evaluations. Diagonal also exploit separability problem, but, crucially,...

By including acoustic-elastic interactions in a topology optimization based on the covariance matrix adaptation evolution strategy, we developed acoustic cloaks of optimal design that render an object unobservable through airborne and water-borne sounds. This strategy helps exploring topologies minimize scattering sounds around made from acrylonitrile butadiene styrene copolymers frequently used as ink 3D printers. applying level set methods, our designed are expressed iso-surfaces...

Various strategies have been proposed to achieve invisibility cloaking, but usually only one phenomenon is controlled by each device. Cloaking an object from two different waves, such as electromagnetic and acoustic a challenging problem, if not impossible, be achieved using transformation theory metamaterials, which are the major approaches in physics. Here, developing topology optimization for controlling both we present multidisciplinary attempt designing biphysical cloaks with...

Many multiobjective evolutionary algorithms rely on the nondominated sorting procedure to determine relative quality of individuals with respect population. In this paper, we propose a new method decrease cost procedure. Our approach is at start algorithm run and update knowledge as population changes. order do efficiently, special data structure called M-front, hold part The M-front uses geometric algebraic properties Pareto dominance relation convert orthogonal range queries into interval...

We propose a novel natural gradient based stochastic search algorithm, VD-CMA, for the optimization of high dimensional numerical functions. The algorithm is comparison-based and hence invariant to monotonic transformations objective function. It adapts multivariate normal distribution with restricted covariance matrix twice dimension as degrees freedom, representing an arbitrarily oriented long axis additional axis-parallel scaling. derive different components show linear internal time...

Safety alignment is an essential research topic for real-world AI applications. Despite the multifaceted nature of safety and trustworthiness in AI, current methods often focus on a comprehensive notion safety. By carefully assessing models from existing safety-alignment methods, we found that, while they generally improved overall performance, failed to ensure specific categories. Our study first identified difficulty eliminating such vulnerabilities without sacrificing model's helpfulness....

We propose a novel variant of the covariance matrix adaptation evolution strategy (CMA-ES) using parameterized with smaller number parameters. The motivation restricted is twofold. First, it requires less internal time and space complexity that desired when optimizing function on high dimensional search space. Second, evaluations to adapt if rich enough express variable dependencies problem. In this paper we derive computationally efficient way update where model richness controlled by an...

Hyperparameter optimization (HPO), formulated as black-box (BBO), is recognized essential for automation and high performance of machine learning approaches. The CMA-ES a promising BBO approach with degree parallelism, has been applied to HPO tasks, often under parallel implementation, shown superior other approaches including Bayesian (BO). However, if the budget hyperparameter evaluations severely limited, which case end users who do not deserve computing, exhausts without improving due...

We propose an approach to saddle point optimization relying only on oracles that solve minimization problems approximately. analyze its convergence property a strongly convex–concave problem and show linear toward the global min–max point. Based analysis, we develop heuristic adapt learning rate. An implementation of developed using (1+1)-CMA-ES as oracle, namely, Adversarial-CMA-ES, is shown outperform several existing approaches test problems. Numerical evaluation confirms tightness...