genetic algorithm

Portfolio selection has been recognized as one of the most significant and challenging problems in financial engineering since Markowitz’s pioneering work on the mean-variance model. This problem centers on the optimal allocation of wealth across a set of assets to maximize returns while minimizing investment risk. While the basic Markowitz mean-variance framework is theoretically elegant and foundational, it has faced criticism from investment practitioners due to its reliance on unrealistic assumptions that limit its practical applicability. Specifically, the traditional model assumes perfect market conditions and neglects real-world constraints, such as the need to limit the number of assets in a portfolio (cardinality), which can significantly reduce its practical applicability. To address these limitations, this paper extends the mean-variance portfolio selection model by incorporating cardinality and floor-ceiling (quantity) constraints. The cardinality constraint ensures that the portfolio includes a specified number of assets, while the floor-ceiling constraint regulates the allocation to each asset, restricting it within predefined bounds. These added constraints transform the classical quadratic optimization problem into a mixed-integer quadratic problem, which necessitates the use of approximation algorithms such as metaheuristic algorithms for efficient and feasible solutions. Although numerous metaheuristic algorithms have been employed to tackle this problem, genetic algorithms have gained prominence due to their balance between solution quality and computational efficiency. However, the standard genetic algorithm is not without its shortcomings, particularly when handling the complexity of constrained portfolio optimization. To overcome these limitations, we propose a novel crossover operator designed to enhance the performance of the genetic algorithm.

Nowadays, one of the major concerns of investors is choosing a realistic stock portfolio and making proper decisions according to an individual's utility level. It is essential to consider two conflicting goals of return and risk for profitability; as a result, balancing the above goals has been identified as an investment concern. This paper modifies and optimizes a multi-objective and multi-period stock portfolio considering cone constraints and uncertain and stochastic discrete decisions. Non-dominated Sorting Genetic Algorithm-II (NSGA-II) was used to solve the model due to the issue's complexity. Two objective functions in the model could be explained by maximizing expected returns and minimizing investment risk. The Pareto chart of the problem was drawn, which allows investors to make decisions based on various levels of risk. Another result obtained in this study is calculating the percentage of optimal amounts assigned to each asset, providing a base for investors to avert investing in unsuitable assets and incurring losses. Finally, a sensitivity analysis of essential parameters was performed, which is critical in this issue. According to the results, increasing the number of problem constraints provides a base for the model reaction, and the optimal percentage allocated to each asset varies. Therefore, this prioritizes restrictions in different situations and according to the investors' choice.

In this paper, a new type of Vehicle Routing Problem (VRP) in the valuable commodity transportation industry is modeled considering the route risk constraint. The proposed model has two objective functions for risk minimization. In the first objective function, three concepts are presented, which are 1) the vehicle does not travel long distances in the first three moves because it carries more money, 2) to serve the same branch on two consecutive days, at the same time, and 3) the bow should not be repeated in two consecutive days. This reduces the possibility of determining a fixed pattern for the service and increases its security. In the second objective function, the risk is a function of the amount of money, the probability of theft, and the probability of its success. Two different meta-heuristic algorithms have been used to solve the proposed model, including the genetic and ant colony optimization algorithms. In computational testing, the best parameter settings are determined for each component, and the resulting configurations are compared in the best possible settings. The validity of the answers of the algorithms has been investigated by generating different problems in various dimensions and using the real information of Shahr Bank. The results show that the genetic algorithm provides better results compared to the ant colony algorithm, with an average of 0.93% and a maximum of 1.87% difference from the optimal solution.

The main goal of the research is production timing with the approach of meta-heuristic algorithms. First, the mathematical model of the production schedule was presented, and then the model was solved with the genetic algorithm. All types of genetic operators were considered at this stage, and an attempt was made to achieve better answers by choosing appropriate methods. The result of applying these targeted selection methods was the rapid convergence of the population. However, this fast convergence did not provide an optimal solution because it quickly converged all the people of the population to a local optimal solution and did not allow the algorithm to search more of the solution space. Therefore, contrary to expectations, the targeted selection methods without a suitable generation method did not improve the algorithm's efficiency. At this stage, the generation methods were considered; the optimal solution for big problems was also obtained by implementing the selection methods. By finding the appropriate generation method, it was observed that even the operators who did not have much ability to see close to optimal solutions succeeded in finding optimal solutions.

The main goal of cloud computing is to achieve higher throughput on a large scale. Load balancing is always a challenge and requires a distributive solution. The response time criterion and energy consumption are evaluated by dynamically transferring the local workload from one machine to another or a less commonly used machine. The main purpose of the load balancing algorithm is to improve the response time by distributing the system's total load. Different algorithms are used in load balancing that can have different parameters. The most important features used are desirability and efficiency. In this report, we optimize the execution time in a set of tasks by examining the load balance parameters and using the Firefly algorithm. The proposed algorithm includes the improved firefly model, which is defined as two parts. The innovation of the present study includes improving the performance of the firefly algorithm and reducing the number of searches in this method, and it has been compared with other optimization algorithms from various aspects. The proposed algorithm enhances the firefly model by improving its performance and reducing the number of searches, as compared to other optimization algorithms. The research results show that the proposed method has a better balance in response time and memory than the GA, NSGA-II, and PSO methods. They also show that the load balance in processor efficiency has a growth of 6% compared to the GA, NSGA-II, and PSO.

Management of blood product consumption is a complex and important issue in health systems. Limited blood supply, corruption, special conditions for storage of blood products, and high costs due to losses and lack of blood in medical centers are among the factors affecting the problem. In this study, all three levels of donors, blood collection centers, and customers (hospitals) are considered for modeling the blood supply network in the form of a multi-objective model. Three objectives of the proposed model are: (a) minimizing total costs, (b) minimizing total delivery time of blood units, and (c) minimizing the maximum unmet demand of hospitals in each period. Next, the model used two multi-objective optimization algorithms namely NSGAII and MOPSO algorithms for solving 30 sample problems in different dimensions (small, medium, and large). After solving the sample problems, the efficiency of the two algorithms were compared with each other. According to the results, for the cost objective function and each of its components separately, it can be seen that the values resulted from the NSGA-II algorithm were less than the MOPSO . Finally, a real word data set from the Tehran blood center was used to evaluate the validity of the proposed model.

Today's transportation systems, which largely rely on the combustion of fossil fuels, play a significant role in contributing to energy-related greenhouse gas (GHG) emissions, thereby raising serious concerns about sustainability. As awareness of environmental issues grows, incorporating sustainable practices into logistics, particularly in cross-dock scheduling, is becoming increasingly vital. This paper introduces a sustainable vehicle routing problem (VRP) that integrates cross-docking to enhance decision-making within logistics systems. Beyond purely economic considerations, it emphasizes critical aspects like environmental impacts, notably CO2 emissions, and social factors such as equity among drivers and overall customer satisfaction. To tackle these complex challenges, a metaheuristic approach blending Genetic Algorithms (GA) with mixed integer programming (MIP) is proposed as an effective solution strategy. The method's efficacy is validated through various instances of differing sizes, revealing that the GA yields results with minimal deviation from optimal fitness values in smaller instances. Additionally, a comprehensive real case study is conducted to showcase the model's applicability in practical scenarios and finally, some suggestions for further researches are given. This study not only illustrates the operational benefits of the proposed approach but also underscores the importance of sustainable logistics in mitigating environmental impacts while fostering social equity and enhancing customer experience.

This paper examines the use of hybrid metaheuristic algorithms to optimize order quantity in a single manufacturer-multi-supplier two-level JIT Supply Chain (SC) in the production system. Over the years, production systems have largely been controlled by either Material Requirement Planning (MRP), Just in Time (JIT), or Optimized Production Technology (OPT) paradigm. In the SC environment, traditional material demand planning does not consider the supplier's supply capacity and economic benefits, which is not conducive to the long-term cooperation of upstream and downstream enterprises in the SC. The main goal of this paper is to optimize ordering batches based on MRP and JIT in the SC. There is limited research in designing and optimizing the SC/procurement network. This study is among the first to integrate supplier selection to optimize performance indicators in SC network design, considering the minimization of the total cost of the JIT SC order batch coordination adjustment model. The Bill of Materials (BOM) constraints and MRP formulation principles of product production are followed to minimize the total cost of downstream companies' inventory, transportation, out-of-stock, and SC crashes. The MRP-led SC ordering batch collaborative optimization model is constructed; the manufacturer's main production plan is adjusted to change the procurement plan to obtain supplier supplies according to the scheme, an improved discrete Particle Swarm Optimization (PSO) algorithm and Genetic Algorithm (GA) is designed to solve the model; an example verifies the feasibility of the model. The algorithm's effectiveness is proved by analyzing and comparing the algorithm results.

Uncertainty and variability in demand and supply processes make it difficult for companies to make inventory management decisions. In this study, a model is developed that will provide the maximum service level of a pharmaceutical warehouse under the budget constraint, taking into account stochastic demand. Due to stochastic demand, the chance constraint programming approach is used to achieve the desired service level at different levels. In this study, the problem of a pharmaceutical warehouse that supplies medicines to pharmacies and hospitals is considered as a real-world problem. The model is designed as a dynamic programming model based on periods. Since there are thousands of drugs in the pharmaceutical warehouse, as the number of products increases, it becomes difficult to find the appropriate solution in an acceptable time. The model is first solved as a mixed integer linear programming model in Lingo. A genetic algorithm (GA) approach is then proposed for large-scale problems. The simulation optimization method also applied to the problem and compared with the optimization method and GA. The GA approach yields better results in the shortest time as the number of periods increases. The developed integrated model demonstrated a numerical example in a pharmaceutical warehouse and was solved using three different approaches. This study is of great importance in terms of providing results that will enable managers to decide the amount of items they should keep in their warehouses by using their budgets in the most efficient way. Nine different scenarios have been derived with various chance constraint risk factors and budget values. Scenario analysis has revealed that the budget has a significant impact on the results at a 95% confidence level. If a pharmaceutical warehouse increases its budget by 10%, it can reduce its total annual inventory carrying costs by 70%.

Hub networks play a crucial role in optimizing transportation flow and reducing overall costs by efficiently connecting origins and destinations through strategically placed hub nodes. The decision of hub location carries significant long-term implications and necessitates consideration of various factors within an uncertain environment. This paper addresses the hub arc location problem in hub networks, considering setup costs, isolated hubs, and uncertain flows between nodes. To tackle this challenge, a two-stage stochastic programming model is formulated to incorporate the uncertainty in flow volumes. Additionally, a robust optimization approach is proposed to enhance the resilience of hub location decisions against uncertain scenarios. The problem is solved using a tailored Genetic algorithm, which achieves optimal solutions with high quality and reasonable computational time. The results demonstrate the effectiveness of the proposed methodology in handling the uncertain nature of the hub location problem, contributing to the advancement of transportation planning and logistics optimization. The findings provide valuable insights for practical applications in real-world scenarios, offering a framework for decision-makers to make informed choices regarding hub network design and location. By integrating uncertainty and robust optimization techniques, this paper offers a comprehensive approach to address complex transportation network problems and improve overall efficiency in transportation systems.

Nowadays, wireless sensor networks (WSNs) are widely used in different sectors. The problem in these networks is the non-rechargeable batteries of these sensors, which limit the lifetime of the network. Therefore, the optimal energy consumption of sensors is an open research topic. In this paper, a new algorithm with the Development of Genetic Algorithm with the Floyd Warshall (DGAFW) has been proposed. Using the proposed DGAFW algorithm, the number of clusters and nodes assigned to each cluster is first determined with the Floyd Warshall algorithm and then the Cluster Head (CH) is selected using fuzzy logic. Finally, the optimal placement of the base station is specified by the combination of the Genetic Algorithm and the Floyd Warshall. The DGAFW algorithm is based on minimizing the distance of sending multi-hop messages. The simulation is carried out in MATLAB 2023a online software. The simulation results obtained from the DGAFW algorithm have been compared based on the distance, the amount of remaining energy in each round, and the number of rounds of network activity in the case where the location of the base station is fixed or randomly determined in each round. The results obtained show that the DGAFW algorithm compared to the case of random base station and fixed station respectively, has 12.7% and 14.3% shorter average message-sending distance in each round, 14.7% and 19.1% more residual energy and also 36% and 48% more rounds of network activity.

Logistics in upstream oil industry is a critical task as rigs need consistent support for ongoing production. In this paper, a multi-period, multi-product and multi-hub routing and scheduling model is presented for offshore logistics problem. As rigs can be served in specific time intervals, time windows constraints are considered in the proposed model. Despite classic VRP models, vessels are not forced to return hubs at the end of duty days. Also, a vessel may leave and return back to hubs several times during the planning horizon. Moreover, the model determines which vessels are applied in each day. In other words, a vessel may be applied in some days and be inactive in other days of planning horizon. To develop a compromise model, fueling issue is considered in the model. As a rig can be supplied by different vessels in real world cases, the proposed model is split delivery. Based on these challenges and contributions, this research deploys an integrated optimization of routing and scheduling of vessels for offshore logistics. This paper deals with a combinatorial optimization model which is NP-hard. Hence, Genetic Algorithm is applied as the solution approach. The average gap between objective functions of GAMS and GA is only 1.18 percent while saving CPU time in GA is much more than GAMS (about 78.16 percent on average). The results confirm the applicability and efficiency of the GA.

The process of transferring money from the treasury to the branches and returning it at specific and limited periods is one of the applications of the Vehicle Routing Problem (VRP). Many parameters affect it, but choosing the right route is the key parameter so that the money delivery process is carried out in a specific period with the least risk. In the present paper, new relationships are defined in the form of three concepts in order to minimize route risk. These concepts are: 1) the vehicle does not travel long routes in the first three movements, 2) a branch is not served at the same hours on two consecutive days, and 3) an arc should not be repeated on two consecutive days. The proposed model with real information received from Bank Shahr has been performed for all branches in Tehran. Because the  VRP is an NP-Hard problem, a genetic algorithm was used to solve the problem. Different issues in various production dimensions were solved with GAMS and MATLAB software to show the algorithm solution quality. The results show that the difference between the genetic algorithm and the optimal solution is an average of 1.09% and a maximum of 1.75%.