Journal of Theoretical & Computational Science
Open Access

Recent Trends and Challenges in Computational Optimization and Simulation

Lewis Hamilton^{* *}Correspondence: Lewis Hamilton, Department of Computational and Data Science, Massey University, New Zealand, Email:

Author info »

Description

Modern design practices in engineering and industry incorporate modeling, simulation, and optimization. Over the past few decades, all three components have made incredible progress. However, there are still a lot of difficult problems that need to be solved, and the current trends tend to model and optimize using surrogate-based techniques and algorithms that are inspired by nature. Modern design practices in engineering and industry incorporate computational optimization, modeling, and simulation. The stringent requirements of minimizing environmental impact and carbon footprint necessitate a paradigm shift in scientific thinking and design practice due to the limited resources at hand, the need to minimize costs and energy consumption while maximizing performance, profits, and efficiency. Approximations, on the other hand, are both necessary and a viable option due to the fact that real-world problems typically involve far more complexity and nonlinearity than can be captured by models or optimization tools.

In order to establish a formal relationship between the values of the designable parameters and the performance of the system, it is necessary to rewrite the system's design objectives and behaviours in mathematical terms. This relationship can be represented as a scalar function that can be minimized in some cases, while a set of competing objectives can only be formulated in many others, resulting in a multi-objective optimization problem that is complex. A decision-making process to select the best combination from a feasible set of, typically, non-commensurable, objective sets can occur even if the solution sets to a multi-objective problem can be found. Depending on the utility and/or decision criteria, this choice is not simple. Metaheuristic algorithms that are inspired by nature are taking over from the more conventional approaches that have been used in computational optimization in recent years. Despite the fact that traditional strategies can still play a significant role in solution strategies. However, the development of novel techniques primarily based on swarm intelligence tends to be the primary focus of new studies and research. Particle swarm optimization, cuckoo search, and the firefly algorithm are new algorithms that have gained a lot of popularity. These metaheuristic algorithms are so popular due to their simplicity and ease of implementation as well as their ability to solve a wide range of problems, many of which are highly nonlinear. Nature-inspired algorithms have the advantages of simplicity, flexibility, and ergodicity, which partially fulfill the requirement to deal with nonlinearity in a non-conventional manner. Most of the time, these algorithms are very easy to understand and use, so learning them is easy for new users. As a result, researchers from a variety of backgrounds can use them effectively in their own work. Nature-inspired are also quite adaptable, specifically; these seemingly straightforward algorithms are capable of solving highly complex, high nonlinear optimization issues. This indicates that they are able to locate the genuine global solution in a time frame that is practically acceptable. While gradient-based algorithms, for example, do not exhibit such high ergodicity, the new algorithms inspired by nature do. Surrogate modeling techniques, which would make it possible to create models that are smooth, globally accurate, and computationally inexpensive, have received significant research attention. The most difficult problem, which has not yet been solved, is figuring out how to build models that are both practical and sufficiently accurate while still being computationally efficient. Current trends appear to be moving toward large-scale computation toward parallel computing, grid computing, and cloud computing. With vast computer resources to be tapped for overnight computing, energy consumption may become yet another major issue. Both the speed of computers and the cost of desktop computers have decreased steadily. The main problem is figuring out how to build models that are both computationally efficient and accurate enough to be useful while also being easy to put into practice. This problem hasn't been solved yet. Parallel computing, grid computing, and cloud computing appear to be the current trends in large-scale computation as the speed of computers has steadily increased and the cost of desktop computers has steadily decreased.