ABSTRACT
Validation is an essential process that should be undertaken before any computational model is used for “production” purposes. In general, validation helps ensure that the inputs to software are accurate, to the extent that they produce the desired output behavior. While validation has many parallels to verification, verification ensures that the components of the underlying software are correct, whereas validation focuses on ensuring the phenomenological correctness of a model-for instance, by testing that a set of parameters given to a simulation code produce the correct thermodynamic outputs. Thus, one can consider software to be somewhat decoupled from the model. For example, a programmer implementing a database can verify the core functions without needing to validate the specific data that will be stored in the database. This distinction can be summarized as follows:
• Verification: Are we building the system right? Is the code correct? • Validation: Are we modeling the right system? Did we give the code the correct
information?
