A situation where the set of initial solutions lies near the position of the true optimality (most favourable or desirable solution) by chance can increase the probability of finding the true optimality and significantly reduce the search efforts. In optimisation problems, the location of the global optimum solution is unknown a priori, and initialisation is a stochastic process. In addition, the population size is equally important; if there are problems with high dimensions, a small population size may lie sparsely in unpromising regions, and may return suboptimal solutions with bias. In addition, the different distributions used as position vectors for the initial population may have different sampling emphasis; hence, different degrees of diversity. The initialisation control parameters of population-based metaheuristic algorithms play a significant role in improving the performance of the algorithms. Researchers have identified this significance, and they have put much effort into finding various distribution schemes that will enhance the diversity of the initial populations of the algorithms, and obtain the correct balance of the population size and number of iterations which will guarantee optimal solutions for a given problem set. Despite the affirmation of the role initialisation plays, to our knowledge few studies or surveys have been conducted on this subject area. Therefore, this paper presents a comprehensive survey of different initialisation schemes to improve the quality of solutions obtained by most metaheuristic optimisers for a given problem set. Popular schemes used to improve the diversity of the population can be categorised into random numbers, quasirandom sequences, chaos theory, probability distributions, hybrids of other heuristic or metaheuristic algorithms, Lévy, and others. We discuss the different levels of success of these schemes and identify their limitations. Similarly, we identify gaps and present useful insights for future research directions. Finally, we present a comparison of the effect of population size, the maximum number of iterations, and ten (10) different initialisation methods on the performance of three (3) population-based metaheuristic optimizers: bat algorithm (BA), Grey Wolf Optimizer (GWO), and butterfly optimization algorithm (BOA).

The distributive power of the arithmetic operators: multiplication, division, addition, and subtraction, gives the arithmetic optimization algorithm (AOA) its unique ability to find the global optimum for optimization problems used to test its performance. Several other mathematical operators exist with the same or better distributive properties, which can be exploited to enhance the performance of the newly proposed AOA. In this paper, we propose an improved version of the AOA called nAOA algorithm, which uses the high-density values that the natural logarithm and exponential operators can generate, to enhance the exploratory ability of the AOA. The addition and subtraction operators carry out the exploitation. The candidate solutions are initialized using the beta distribution, and the random variables and adaptations used in the algorithm have beta distribution. We test the performance of the proposed nAOA with 30 benchmark functions (20 classical and 10 composite test functions) and three engineering design benchmarks. The performance of nAOA is compared with the original AOA and nine other state-of-the-art algorithms. The nAOA shows efficient performance for the benchmark functions and was second only to GWO for the welded beam design (WBD), compression spring design (CSD), and pressure vessel design (PVD).

The krill herd (KH) algorithm is a global metaheuristic algorithm that was initially proposed for solving continuous optimization problems. The KH algorithm, since inception, has generated considerable real-world application interests in the research community. The standard algorithm solution implementation steps follow the initialization mechanism, which relies mainly on educated guesses or random initialization solution generation. Therefore, to improve the performance of the KH algorithm, the current study is set to investigate the influence of initializing the KH algorithm with three low-discrepancy sequences, such as the Faure sequence, Sobol sequence, and Van der Corput sequence. These low-discrepancy sequences are known to be more uniformly distributed across the problem search space than the commonly used random number initialization method. The study also evaluates the influence of population size on the performance of the proposed variants of the improved KH algorithms. The experimental results show significant improvements for the enhanced KH algorithms in terms of performance and the quality of solutions obtained; particularly on standard benchmarked high-dimension test problem instances, where the enhanced KH variants outperformed the existing basic KH algorithm for all the test functions evaluated. Similarly, the results for low dimension test cases showed less sensitivity to the initialization schemes, as the performance of our proposed improved scheme was comparable to that of the basic KH algorithm. However, in most cases, as the problem dimension was scaled up, the enhanced KH outperformed the basic KH. Evaluation results based on the population size of the algorithm, revealed that when the number of Krill is set at 25, the Sobol based KH initialization scheme performed better than did the other methods. Although, the Van der Corput and Faure based KH initialization schemes showed similar sensitivity when the dimension was set at 20. As we varied the population size of Krill, it was observed that the performance of the Sobol based KH initialization scheme deteriorated, whereas the other two methods showed superior performance. Overall, the findings from this study revealed that there are significant improvements in the performance of KH algorithm when initialized with low-discrepancy sequences.

Arithmetic optimization algorithm (AOA) is one of the recently proposed population-based metaheuristic algorithms. The algorithmic design concept of the AOA is based on the distributive behavior of arithmetic operators, namely, multiplication (M), division (D), subtraction (S), and addition (A). Being a new metaheuristic algorithm, the need for a performance evaluation of AOA is significant to the global optimization research community and specifically to nature-inspired metaheuristic enthusiasts. This article aims to evaluate the influence of the algorithm control parameters, namely, population size and the number of iterations, on the performance of the newly proposed AOA. In addition, we also investigated and validated the influence of different initialization schemes available in the literature on the performance of the AOA. Experiments were conducted using different initialization scenarios and the first is where the population size is large and the number of iterations is low. The second scenario is when the number of iterations is high, and the population size is small. Finally, when the population size and the number of iterations are similar. The numerical results from the conducted experiments showed that AOA is sensitive to the population size and requires a large population size for optimal performance. Afterward, we initialized AOA with six initialization schemes, and their performances were tested on the classical functions and the functions defined in the CEC 2020 suite. The results were presented, and their implications were discussed. Our results showed that the performance of AOA could be influenced when the solution is initialized with schemes other than default random numbers. The Beta distribution outperformed the random number distribution in all cases for both the classical and CEC 2020 functions. The performance of uniform distribution, Rayleigh distribution, Latin hypercube sampling, and Sobol low discrepancy sequence are relatively competitive with the Random number. On the basis of our experiments’ results, we recommend that a solution size of 6,000, the number of iterations of 100, and initializing the solutions with Beta distribution will lead to AOA performing optimally for scenarios considered in our experiments.