Sather Provides Nonproprietary Access to Object‐Oriented Programming

odern scientific com­ puting is placing ever more stringent re­ quirements on the au­ thors of scientific code, and on the languages and programming environments that they use. At one time, most scientists were happy with systems that could efficiently execute relatively simple repetitive operations on fixed-sized arrays. The tools appropriate for that task are woefully inadequate for the challenges posed by the computa­ tional sophistication of recent algor­ ithms for scientific computing. Such challenges take many forms. For example, modern finite­ element code must adaptively change the structure of the underlying grid, dynamically adding and deleting nodes during a computation. Com­ plex geometry algorithms are re­ quired for efficient manipulation and visualization of geometric structures. Modern nonlinear optimization tech­ niques utilize many interacting soft-