I am an academic researcher and working on optimization. Many of my friends are always asking what my research is about and what I am doing. The term ‘optimization’ is a technical term, and it deserves to be explained in a more academic manner. However, I am going to write this article in a very simple way in order to give a general overview of the optimization for who has no background in computer science. Let me remind you in the first place that I am not a native speaker in English, and please forgive my grammatical errors in the following text.
Broadly speaking, optimization is the process of making better decisions for a given problem. Let us think about an example of preparing and cooking a meal. Your objective is making a delicious one based on your good taste. To prepare a good meal, one should adjust the right amount of each ingredient such as; salt, oil, pepper, etc. If you are an inexperienced cooker then you are probably using the trial and error approach. If you always cook in your daily routine then you know the right amounts of each ingredient. Actually, you are simply optimizing the amount of these to make the meal delicious. You can see amounts of ingredients as decision variables, providing values for those variables as a decision-making process, and making a delicious meal as an objective. We can say that this is a single objective optimization since you have only one objective in this example.
In real life, each of us has many objectives to achieve, and most of them are conflicting with each other. If this sentence sounds complicated, don’t worry, you are going to understand it with the help of the given example next. I am living as a single almost for two years, that is why I am going to give an example about marriage If you are looking for someone to marry, you probably feel always confused about women. You have your own preferences, and based on your preferences you are trying to find the correct ‘one’. There are some alternatives that you have met. One of them is so beautiful but the most foolish one. Another one is best for home works but not beautiful enough. The last one is the most intelligent but worst in cooking. Which one is the best if your objectives are ‘finding the beautiful one’, ‘finding the most intelligent one’, and ‘finding the best one for home works’? I can say that based on your alternatives; your objectives are conflicting with each other. You need to sacrifice in some of your objectives to gain more in others. Therefore, there is no global optimum. If you don’t want to sacrifice, you need to keep looking for new alternatives. In the end, I can easily say that you are going to balance the objectives based on your own preferences. If this is the case, we call this a multiobjective optimization since you have multiple objectives to be balanced.
Now, I am going to give a more serious example which is recently very popular nowadays. As you are aware, COVID-19 is the most spoken topic in the world now. All countries try to stop the spread of the virus. Policymakers (we call ‘decision makers’ in optimization) are trying to find the most effective way to deal with the pandemic. As can be seen from all over the world, there are many aspects that should be taken into account in making decisions. If we try to give some sample objectives, policymakers want to minimize the number of infected people, the number of deaths, and the length of the quarantine. They are also trying to minimize the economic and social crises. To reach these objectives, they decide when to put a city in lockdown and how long, what should be the bed capacity of hospitals, or the number of ventilators, etc. However, if you put a city in lockdown for a long time then there would be some economic problems. It also causes many social problems. Therefore, policymakers need multiobjective optimization tools to find the most balanced solutions. As you can see, some of these objectives are also conflicting. If you want to make the economy better there would be consequences in the number of dead people because of the virus, or vice versa. Preferences of the policymakers become more important, and if they see their resources are enough to deal with some of the above sacrifices, they can make better decisions.
To summarize, optimization means finding the best possible solution for a given problem. If you have hundreds of decision variables and many objectives, it is impossible to find an optimum solution with your own. You need to use some optimization tools that support you the find the solutions. The overall goal is to reach the best solution. However, there is no single optimal solution in multiobjective optimization. The preferences of the decision maker are incorporated into the optimization process to find the most preferred solution by the decision maker. The final solution would be different for different decision makers based on the provided preferences.
I hope this article gives you a general overview and understanding of the optimization. If you are an expert in the field of optimization, please do not hesitate to inform us if there is some misunderstanding in this article. If you like the article, please make your friends know, and share it in social media.
B.A.
Ph.D. Computer Engineer
Keywords: optimization, single-objective optimization, multiobjective optimization, decision-making