In article <firstname.lastname@example.org>, Alberto Moreira <email@example.com> wrote: >firstname.lastname@example.org (Herman Rubin) said:
>>One of the things I work on are computational procedures. >>In a practical situation, knowing how to carry out >>arithmetic, at which I am fast and accurate, is rarely >>helpful. In computational astrophysics, or computational >>biology, or computational economics (I have worked with >>these people), the computations have to be automated.
>Someone has to write that program, eh ? And there's no testing of >machine results without careful hand analysis. Heck, I'm a programmer, >I know. I remember once I spent a whole week trying to find that one >plus sign that should have been a minus in one of my complicated >engineering programs. It's ok to use the machine, until of course >you're in charge of making the machine work correctly !
I have been in that position as well, and I know the problems involved. Even when it comes to automating computations, there are different ways of doing it, and some are worse than others.
"Hand analysis" can be done with computers as well; there are check procedures. One can set up (and these are now available) computer programs which will precisely duplicate the "hand" operations. I am not going to test a computer program by hand if it involves multiplying numbers with hundreds of digits.
The important thing in designing algorithms, and often in using them, is to know what you are doing. If you do so, you are likely to curse those who designed the computer language, and who produced the compiler for it. Doing mathematics, computing, etc., is an art, and teaching it as "science", which really makes it religious ritual, is wrong. -- This address is for information only. I do not claim that these views are those of the Statistics Department or of Purdue University. Herman Rubin, Deptartment of Statistics, Purdue University email@example.com Phone: (765)494-6054 FAX: (765)494-0558