Introduction

This document describes a compositional modeling language, CML, that is a general declarative modeling language for logically specifying the symbolic and mathematical properties of the structure and behavior of physical systems. CML is intended to facilitate model sharing between research groups, many of which had long been using similar languages. These languages are based primarily on the language originally defined by Qualitative Process theory [5] and include the languages used for the Qualitative Physics Compiler (QPC) [1, 4], compositional model formulation [3], and the Device Modeling Environment (DME) [7]. CML is an attempt to synthesize and provide a clean redesign of these languages.

CML was designed with the following desired features in mind:

• Support multiple modeling approaches. The language should support a variety of different approaches to representing physical phenomena. In particular, it should enable the definition and use of domain theories that use components, process, bond graphs, kinematic pairs, etc., and also support both relational and object-oriented specification styles.
• Minimal. The syntax of the language should be simple and natural to use.
• Efficient. The language should not be so general as to require impractical inference procedures for predicting behavior. It is for this reason that the language excludes arbitrary disjunction and is approximately horn.
• Readable. Someone familiar with the domain theory should be able to read and easily understand an expression of the domain theory in the language. In particular, the terms should be meaningful to someone who is trained in the domain but not trained in AI or aware of the historical precursors of the language.
• Expressive. The language should support lumped parameter ordinary differential equations that are common in engineering modeling.

There is a natural tension between trying to agree to a common modeling language and supporting future research in this area. Our intention is for this to be an open, evolving language. The current document describes a base language that all of the parties involved will support. Various extensions will then be defined as they naturally arise in the course of research (See section , page ). Writing in the base language should guarantee sharability (although there will no doubt be some limitations on sharability as a function of the class of numeric equations each implementation supports). Thus, an important goal in designing the base language is to support as much sharing as is reasonably possible.

This modeling language specification effort is not formally affiliated with the ARPA Knowledge Sharing Initiative. However, the effort has been strongly influenced by the Knowledge Interchange Format (KIF) being developed in that initiative. [6]; CML is fully translatable to KIF and where possible we have adopted conventions established by KIF. This effort is also not affiliated with the product description standardization effort (PDES) [10]. Integration of CML with the Express modeling language used in PDES is probably desirable, but beyond the scope of the current effort.

CML specifies a set of top-level forms for defining models and an ontology of primitive functions, relations, and constants. Section  gives a brief overview of the typical usage patterns of CML and section  introduces some prerequisite conventions and definitions. Section  specifies the top-level forms while section  specifies the syntax of atomic sentences and the semantics of predefined functions and relations. As much as possible, a summary of the relevant discussions, alternatives considered, and rationale for decisions is included.