On Wed, 19 Mar 1997, Konrad Hinsen wrote:
> OOP is just a method of structuring code, no magic to prevent writing
> it. If efficiency demands a special algorithm for each combination
> of a hundred different matrix storage schemes, then these have to
> be written one by one. The best you can expect from OOP is to hide
> this mess from the user.
This pretty much sums up one of the ways in which I view oon. I teach
topics like interval arithmetic and automatic differentiation and sparse
linear solvers to undergraduate students, making use of demonstration
software which is fairly readbale at the top level and hides all the mess
till they are ready for it (if ever!).
We also use similarly structured codes for research students in
optimisation and automatic differentiation.
My route into this sort of software seems different from correspondents
so far. The teaching software I am talking about is in Ada; but we are
currently switching our undergraduate teaching to f90, which is involving
me in a big re-coding chore. For us C++ was not really on the agenda,
for a variety of reasons which are now past debating again for a year or
two at least!
Mike Bartholomew-Biggs
Numerical Optimisation Centre
Mathematics Department
University of Hertfordshire
England
This archive was generated by hypermail 2b29 : Wed Feb 20 2002 - 03:20:05 EST