Optima partitions are made from toughened safety glass to BS EN 12600 to comply with Approved Document K of the Building Regulations. This glass is heat soaked, meaning that if it breaks it will break safely into small fragments rather than shards.

In optimization problems, a high-quality global optimal solution usually exists in several regions within the feasible domain, and the quality of local optima is also important. The goal of optimizing methods is to find multiple good global and local optima, while minimizing the search effort. The most commonly used strategies for achieving these goals include evolutionary algorithms with niching techniques and partition-based random search.

Optima Partitions: Redefining Office Spaces with Innovative Design

Partition-based random search methods divide the entire feasible domain into a set of regions and sample solutions in each region. The regions with the best quality are exploited at higher rates, and the search is facilitated by the knowledge that there is a high probability of finding an optimum in these areas. Unfortunately, the quality of these good solutions can deteriorate as the number of samples grows.

In this article, we introduce a new method for improving the quality of optimized solutions by controlling the rate at which promising regions are sampled. The method uses a combination of the partition rate, sampling strategy, neighborhood definition, and stopping criteria to achieve this goal. The results of this approach are compared to those of a conventional random search method in a series of minimization problems.