INTRODUCTION

Through the years we have been disappointed by the number of occasions that inversion theory has not yielded successful results. The success ratio of inversion theory seems to match that of wildcat drilling. When results are disappointing it is hard to know the cause. Often we cannot confidently eliminate even one of the major possible causes: (1) inadequate data, (2) insufficient theoretical model, (3) failure of numerical techniques, (4) error in coding. To help students overcome the last two causes of frustration, one of us (JFC) developed the draft of an electronic book (TDF) which guides students through several tens of inverse problems using SEP's methods of ``reproducible research''. Basing the book on Fortran 77 has led to some painful defects that should be remedied before moving on to more extensive and realistic practical examples.

The principal difficulty with TDF in Fortran 77 is the way the inverse theory becomes intermingled with the application. Ideally, the geophysicist would simply set up the problem and then invoke a choice of canned mathematical and numerical solvers. Simply invoking external expertise is easy in Fortran where the geophysicist calls an external subroutine such as FFT. Fortran sets roadblocks, however, when the external intelligence, an optimizer for example, should invoke geophysical application code. Oversimplifying, it is easy when new code calls old code but hard when old code must call the new code. Using C++, there have been somewhat successful attempts to build environments isolating geophysics and mathematics at Rice University, at Colorado Mines, and here at Stanford. Considering the difficulties we at SEP have been having building and sustaining a seismic C++ research environment, we contacted Bill Symes at Rice and he confirmed a willingness to cooperate and attempt to gather a community of like minded researchers. Lydia Deng from Colorado Mines and Martin Karrenbach from Karlsruhe University also participated enthusiastically. From this joint effort sprung our report of this conference.