** Next:** Test case: solving some
** Up:** KRYLOV SUBSPACE ITERATIVE METHODS
** Previous:** A basic solver program

I'd like to digress from our geophysics-mathematics themes
to explain why Fortran 90 has been a great step forward
over Fortran 77.
All the illustrations in this book were originally computed in F77.
Then module
`solver_tiny`
was simply a subroutine.
It was not one module for the whole book, as it is now,
but it was many conceptually identical subroutines,
dozens of them, one subroutine for each application.
The reason for the proliferation was that F77 lacks the ability of F90
to represent operators as having two ways to enter,
one for science and another for math.
On the other hand, F77 did not require the half a page
of definitions that we see here in F90.
But the definitions are not difficult to understand,
and they are a clutter that we must see once and never again.
Another benefit is that the book in F77 had no easy way to switch
from the `cgstep` solver to other solvers.

** Next:** Test case: solving some
** Up:** KRYLOV SUBSPACE ITERATIVE METHODS
** Previous:** A basic solver program
Stanford Exploration Project

4/27/2004