ABSTRACT
The traditional idea of an experiment involves observation of a system of interest under controlled conditions, with the intent of learning something about that system. The system of interest varies by discipline: engineers and physicists may be interested in systems involving physical material, biologists may focus on living organisms (or collections or components of them), while social scientists may be interested in experiments involving
of Design and Analysis of
the behavior of human beings. In contrast, the system of interest in a computer experiment is often a computer model, usually a mathematical description of a real system of interest, represented as a computer program. While not critical to anything written here, the computer representation is usually necessary due to the complexity of themodel. Experimental goals are often similar to those in traditional experiments. While the computermodelmust, in principle, be fully known, it is generally so complex that a useful understanding of its behavior requires the empirical approach of an experiment.
