Computer-Based Modeling in Materials Science

  • PDF / 1,471,934 Bytes
  • 2 Pages / 604.8 x 806.4 pts Page_size
  • 21 Downloads / 205 Views

DOWNLOAD

REPORT


id* •UDDnccn n RD Guest Editor, Brian W. Dodson

In principle, all materials science science research, ranging in scope from problems are easy to solve. One continuum mechanics to atomic-scale describes the system using quantum simulations. mechanical wave functions, solves the The increase in scientific computaSchrodinger equation using the appro- tional capacity which has occurred in priate boundary conditions, and con- the past decade is enormous. While I verts the result into the microscopic was a graduate student at the Univerand/or macroscopic observables of inter- sity of Illinois, the materials research est. Unfortunately, carrying out this laboratory bought a VAX-750 computer, program is usually many orders of mag- then costing about $200,000, to be nitude too difficult. As a result, various shared among perhaps 30 research approximations and simplifications are groups. The acquisition was greeted introduced to obtain descriptions hav- with much enthusiasm since it repreing the minimum complexity to describe sented a major increase in the amount a given class of phenomena. (A simple of computer power routinely available example is the treatment of lattice vibra- to researchers. To illustrate the progress tions in solids using a ball-and-spring made in this short period, I am writing description.) Such a simplified descrip- this article on an IBM-PC equipped with tion is called a model. a co-processor board that provides Due to the complexity of virtually all about twice the computing power of a problems concerning materials, it is fair VAX-750, but that costs less than to say that the use of models is ubiqui- $4,000. This machine can perform simutous in materials science. In many situa- lations considered impractical on a routions, however, models sufficiently tine basis ten years ago. The power of complex to describe the phenomena of such dedicated single-user machines is interest are too difficult to be solved augmented by the proliferation of analytically. The computer has played CRAY-class supercomputers, now stanan essential role in obtaining descrip- dard resources at many research laborations of material behavior based on such tories. A crude estimate indicates that a complex models. Traditionally, the . given investment now produces about amount of computer power available 100 times more computer power than it has restricted such models to idealized did a decade ago. This has had a qualitaproblems, which are then used as proto- tive effect on computer modeling of types for the actual problems. Recently, materials science problems — realistic however, the role of modeling has material descriptions can now be used. expanded to include treatment of In this issue, an effort has been made specific and realistic materials prob- to review the fields in which the advent lems. Development of these new classes of modern computers has had a major of applications has been driven by the impact. The article by L. Davison on tremendous increase in computer recent trends in continuum mechanics power which has becom