Abstract

Dimensional analysis does not adequately model the semantics of measurement data. Distinct concepts such as work and torque are treated as equivalent, while sensibly related concepts such as quantities expressed by weight or volume (tons and bushels of wheat) are rendered incompatible. Dimensional analysis provides no adequate treatment of dimensionless quantities, nor does it differentiate between such concepts as circular angles and rotational angles, or temperature points and temperature intervals.

Preface

The investigation captured in this archival snapshot was conducted sporadically from 1993 to 1995. It was sparked by HP's Computer Systems Division's intent to provide a model of dimensioned data in the OpenODB object-oriented database system, aimed at the needs of POSC (Petroleum Open Systems Consortium). Formal support waned,but several of us remained tantalized by the complexities and anomalies that lay just beneath the surface of such a nominally simple notion as dimensioned data. We met informally for a couple of years, progressing as best we could between the cracks of our "real" jobs, but it eventually became too difficult to sustain any real progress. After one last spurt as an HP Labs Grass-roots Basic Research Project in the summer of 1995, the effort just petered out.

These are the remains, a snapshot of our workbook frozen in mid-flight, with all the lumps and warts of unfinished work in progress. It's disorganized. This document was intended to be a repository from which various "real" documents can be extracted. It provided a framework in which all our stuff could appear, in some logical sequence. It contains material at all stages of completion, and at various levels of presentation for various audiences, from popularly tutorial to academically axiomatic.