comp.lang.ada
 help / color / mirror / Atom feed
From: Ada novice <posts@gmx.us>
Subject: Re: books on numerical programming in Ada
Date: Sun, 25 Jul 2010 03:41:14 -0700 (PDT)
Date: 2010-07-25T03:41:14-07:00	[thread overview]
Message-ID: <e3d19c7f-8695-46c9-8689-0f37ea00df68@t10g2000yqg.googlegroups.com> (raw)
In-Reply-To: 4c4b35b2$0$2375$4d3efbfe@news.sover.net

On Jul 24, 8:49 pm, "Peter C. Chapin" <pcc482...@gmail.com> wrote:
Would any good book on numerical
> methods be good enough? The translation into some programming language
> (such as Ada) might be an "implementation detail."
>
> Peter

Thanks. Yes I agree that if one learns numerical methods well then
implementing these methods in any programming language would only
require some knowledge of that language. I'm a mechanical engineer and
not a computer guru. A mechanical engineer doesn't usually know about
software engineering methods and at times won't know what data type
would best represent some entity. Records for example are quite seldom
used by people other than computer programmers but yet records can
represent some entities far better than some other data type. In the
book: Ada for Software Engineers by Mordechai Ben-Ari, I saw an
implementation of the Euler method in solving an ODE and I would never
have thought of designing such an implementation. And I'm still
studying it.

So, computer programmers typically will know the best way to code the
solution for a given problem. This is why it's far better to study a
book (dealing with numerical methods) which has been written by a
computer programmer. Though one can expect that the computer
programmer won't go too deep in the numerical algorithms themselves,
one will surely learn a lot on how to think like a computer programmer
when faced with more demanding implementation tasks. And having a good
textbook on numerical analysis is a necessity.

I have seen other books in other programming languages written by
engineers other than in the field of computer science. These books
typically won't contain the most efficient implementation nor will use
the best data types to represent a given entity. Many of engineers
involved with numerical calculations still keep their own Fortran
mentality and when they write a programming book in C for example, you
can easily see how the C codes seem to be a mere line by line
translation of some Fortran codes.


YC



  reply	other threads:[~2010-07-25 10:41 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-07-24 18:17 books on numerical programming in Ada Ada novice
2010-07-24 18:49 ` Peter C. Chapin
2010-07-25 10:41   ` Ada novice [this message]
2010-07-25 19:52     ` Peter C. Chapin
2010-07-26 13:38       ` Ada novice
replies disabled

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox