 |
|
 |
 Toward a Revised NBS |
 Toward a Revised NBS |
2 Existing Scope and
 |
|
|
|
|
Toward a Revised NBS |
Toward a Revised NBS |
2 Existing Scope and
|
Most would agree that the National Bureau of Standards Handbook of Mathematical Functions has been a phenomenal success in the field of mathematical publishing. Issued in June of 1964 by the US Government Printing Office, this large hardback volume of more than 1000 pages has been in print continuously since that date as a government publication1Over 200 copies were sold in a recent one-year period. A paperback version has been photo-offset and sold by Dover Publications since 1965. Although Dover does not reveal sales data, it undoubtedly outsells the government edition many times over. Both complete and abridged editions (with many of the numerical tables removed) have been produced by other commercial publishers, and translations have appeared in languages other than English (for example, Russian). The handbook is cited very frequently in scientific articles that make use of special functions, making it one of the most cited of all mathematical reference works.
What were the reasons for this success? Four might be comprehensiveness, authoritativeness, timeliness and applicability.
First, comprehensiveness. A concentrated effort was made to identify the most important mathematical functions as measured by their value in fields outside mathematics, such as physics and statistics, and to assemble the most important properties of these functions together with algorithms, tables of numerical values, graphical representations and other relevant information. Most of the assembled information had been published previously but it was unified and presented in a consistent format. A serious effort was made to include the latest results from the research literature. In the case of tables of numerical data, accuracy was verified and gaps were filled by new computations.
Authoritativeness was achieved by establishing a supervisory scientific committee with Philip M. Morse of the Massachusetts Institute of Technology as its head. The committee members were A. Erdélyi, M.C. Gray, N.C. Metropolis, J.B. Rosser, H.C. Thacher, Jr., John Todd, C.B. Tompkins and J.W. Tukey. The idea to produce such a reference work originated with Milton Abramowitz of NBS. He developed the overall plan and led the selection of the authors for the various chapters. Irene Stegun, also of NBS, took over as technical editor after the death of Abramowitz. One sign of the authoritative nature of the handbook is the impact it has had on standardizing definitions and notation. All mathematical functions are subject to definitional and notational variations that can be confusing when articles by different authors are compared, especially when different countries or disciplines are involved. By becoming, in effect, the standard reference for special functions in much of the world, the handbook helped reduce this problem.
The appearance of a compendium such as the NBS Handbook was timely in that it appeared at just about the time electronic computers were beginning to make their influence strongly felt in applied mathematics. Thus it served to sum up the state of a highly developed mathematical technology at a time of transition to a new technology. In the earlier technology mathematical problems were considered solved if results could be obtained in terms of elementary or higher functions. Because the properties and behavior of the functions were understood, insight into the nature of the solution was available. This insight could be qualitative or quantitative. Both kinds of insight remain equally valuable today. Qualitative insight comes from viewing the `forest' at the expense of the `trees.' A view of the `forest' coming, for example, from knowledge of the asymptotic behavior or distribution of zeros of functions entering into a solution can be immensely valuable in scientific and engineering applications. On the other hand, in detailed mathematical models of physical processes or engineering structures the `trees,' e.g. floating-point numbers, are unavoidable. Solutions obtained by more direct methods such as quadrature, finite differences and finite elements provide an immense amount of quantitative information, and their role in scientific computing has ascended with the increasing power of computers since 1964. However, these methods provide no qualitative insight, and competent practitioners still recognize the importance of special functions. Three areas can be listed in which they are important today. First, they provide a source of test problems for direct solvers. Second, they are used to make a computation more efficient in some applications when length and time scales vary over many orders of magnitude through the technique known as subgrid modeling. And third, they can be highly effective when solutions are representable in a series of special functions or as an integral transform.
The final reason for the success of the NBS Handbook is its emphasis on applicability. The functions chosen for inclusion were determined by their importance in applied mathematics, physical sciences, engineering and statistics, both as a vehicle for computation and as a means for intuitive understanding. The approximations and numerical methods, including the now obsolete method of interpolation in tables, were oriented toward practical concerns, including the construction and testing of mathematical software. Accordingly, a highly mathematical style with formal definitions, theorems and proofs was avoided while an insistence on mathematical rigor was maintained. Lengthy explanations and mathematical developments were similarly avoided. Thus the style is terse and assumes considerable mathematical capability on the part of the user. The style and choice of content undoubtedly contributed to the broad and continuing appeal of the handbook.
The name we have adopted for the envisioned successor to the NBS Handbook of Mathematical Functions is the NIST Digital Library of Mathematical Functions2. Our approach is to retain the ingredients of the original success while taking advantage of a new technological transition. This transition is characterized (in part) by the use of powerful computers and network communications to disseminate highly technical reference information electronically. Electronic publishing, online databases, and generation of technical data on demand are emerging components of this transition. Through its Mathematical and Computational Sciences Division in conjunction with its Electron and Optical Physics Division and its Standard Reference Data Program, NIST can provide the necessary leadership and some of the necessary technical, computational and financial resources. But outside technical and financial assistance is essential, and toward this end an invitational workshop was held July 28-31, 1997, at NIST. The remainder of this report takes a closer look at the existing handbook and examines a few of the issues involved in developing a modernized and updated version.
|
|