Scientific computing can also be called computational science or scientific computation. It is mainly the idea of development of mathematical models, use of quantitative analysis techniques, and use of computers for solving scientific problems.
"Scientific computing is the collection of tools, techniques and theories required to solve on a computer the mathematical models of problems in science and engineering." | ||
| --Gene H. Golub and James M. Ortega | ||
In simple words, scientific computing can be described as an interdisciplinary field, as presented in the following diagram:
Scientific computing as an interdisciplinary field
Scientific computing requires knowledge of the subject of the underlying problem to be solved (generally, it will be a problem from a science or engineering domain), a mathematical modeling capability with a sound idea of various numerical analysis techniques, and finally its efficient and high-performance implementation using computing techniques. It also requires application of computers; various peripherals, including networking devices, storage units, processing units, and mathematical and numerical analysis software; programming languages; and any database along with a good knowledge of the problem domain. The use of computation and related technologies has enabled newer applications, and scientists can infer new knowledge from existing data and processes.
In terms of computer science, scientific computing can be considered a numerical simulation of a mathematical model and domain data/information. The objective behind a simulation depends on the domain of the application under simulation. The objective can be to understand the cause behind an event, reconstruct a specific situation, optimize the process, or predict the occurrence of an event. There are several situations where numerical simulation is the only choice, or the best choice. There are some phenomena or situations where performing experiments is almost impossible, for example, climate research, astrophysics, and weather forecasts. In some other situations, actual experiments are not preferable, for example, to check the stability or strength of some material or product. Some experiments are very costly in terms of time/economy, such as car crashes or life science experiments. In such scenarios, scientific computing helps users analyze and solve problems without spending much time or cost.