Verification is the process of discovering the degree of success in implementing a conceptual model into computer code. Validation is the process of discovering the nature and extent of the correspondence between the computer results and the real world. Accreditation is the determination that a computer program is good enough for the purpose at hand. Generally, data verification and validation are also included in the entire VV&A process. Independent verification and validation (IV&V) is normally a component of VV&A and consists of an outside agent performing some verification and validation. However, the bulk of the verification and validation procedures consists of the internal testing performed by the model builders and the user evaluations that take place each time the model is used.
Because most modern computer models are too complex for complete verification (certainly in one "sitting"), verification is a process that continues throughout the lifetime of the model. On the other hand, the real world is so complex that computer models of the subtler aspects often address areas of the real world that are poorly understood. Accordingly, validation must be an incremental and imprecise process. These two facts make the need for official accreditation more than a formality: at any given time, a model will be imperfectly understood, although its services may be needed. Even for models of well-understood processes, such as projectile trajectories, accreditation may be more than a formality. For example, if computational speed requires an imperfect approximation of the process, it will be required to determine if the known errors are sufficiently small to yield useful results. In all cases, the purpose of the model use is important in the accreditation decision: a model that is acceptible for one use may be totally unacceptible for another use.
Select one of the links below.
HARTLEY
CONSULTING Solving Complex Operational and Organizational Problems Dr. Dean S. Hartley III, Principal |