skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Formal specification is an experimental science

Journal Article · · Programming and Computer Software
OSTI ID:241214
 [1]
  1. Technical Univ., Lyngby (Denmark)

Traditionally, abstract models of large, complex systems have been given in free-form mathematics, combining - often in ad-hoc, not formally supported ways - notions from the disciplines of partial differential equations, functional analysis, mathematical statistics, etc. Such models have been very useful for assimilation of information, analysis (investigation), and prediction (simulation). These models have, however, usually not been helpful in deriving computer representations of the modelled systems - for the purposes of computerized monitoring and control, Computing science, concerned with how to construct objects that can exist within the computer, offers ways of complementing, and in some cases, replacing or combining traditional mathematical models. Formal, model-, as well as property-oriented, specifications in the styles of denotational (respectively, algebraic semantics) represent major approaches to such modelling. In this expository, discursive paper we illustrate what we mean by model-oriented specifications of large, complex technological computing systems. The three modelling examples covers the introvert programming methodological subject of SDEs: software development environments, the distributed computing system subject of wfs`s: (transaction) work flow systems, and the extrovert subject of robots: robotics! the thesis is, just as for mathematical modelling, that we can derive much understanding, etc., from experimentally creating such formally specified models - on paper - and that we gain little in additionally building ad-hoc prototypes. Our models are expressed in a model-oriented style using the VDM specification language Meta-IV In this paper the models only reflect the {open_quotes}data modelling{close_quotes} aspects. We observe that such data models are more easily captured in the model-oriented siyle than in the algebraic semantics property-oriented style which originally was built of the abstraction of operations. 101 refs., 4 figs.

OSTI ID:
241214
Journal Information:
Programming and Computer Software, Vol. 17, Issue 6; Other Information: PBD: Sep 1992; TN: Translated from Programmirovanie; No. 6, 24-43(Nov-Dec 1991)
Country of Publication:
United States
Language:
English