Re: [Swarm-Modelling] Floating point arithmetic

[Russell Standish]

On Wed, May 04, 2005 at 10:47:09AM +0100, James Marshall wrote:
> This probably belongs to comp.lang.advocacy.java, but I think Java is 
> much easier to teach to inexperienced programmers (and I teach first 
> year undergraduate computer scientists), and hence easier to use in my book.
> Syntactic richness in C++ translates to more opportunities for the 
> beginner to shoot themselves in the foot in my opinion! Java helps catch 
> errors early on, either during programming, or when they first occur 
> during execution. This is definitely why I chose Java when faced with 
> having to get non-programmers to implement their own model code.

I don't actually dispute the claim that Java is an easier language for
students to learn programming. I'm not in the business of teaching
programming, so I have no opinion on that subject. My own experience
was learning to program using Pascal, almost 25 years ago. Needless to
day, I have not used Pascal in anger since about 1987...

As for your book, I cannot argue with your choice of language - that
is _so_ audience dependent.


> And, as I say, I'm not sure the performance differences are so vast to 
> outweigh the advantages I've outlined above, but I don't have numbers 
> handy to back this up... I have to bow to your expertise that there is a 
> difference, but how big is it really?

To be quite honest, I don't know, since I've never used Java in a
comparative way. What I can say, is that over 15 years of high
performance computing experience, I have only had 1 user (yes that's
right, count 'em on 1 finger) who asked for Java. And that person was
(I believe) experimenting with some of the Java Grande stuff, ie it
was not "serious work".

Of course the wags could say that Java is so efficient, you can get
done on your workstation what requires an entire supercomputer running
C++ :)

> I really don't understand why you're panning Java as being useless for 
> research... I think you might be seriously underestimating the number of 
> journal papers built upon Java models. I believe Java's not a bad choice 

What I'm saying is that Java is not used in High Performance
Computing. A field of computational science that is still being mostly
performed on workstations is not really mature. It is still an
exploratory, fringe activity. I think this comment still holds for
agent based modelling as a field, something I would like to see
change. For the time being, Java doesn't seem to be the answer.

> for parallel models either... I would have thought RMI would have been a 
> pretty handy feature, and I also believe there's a Java implementation 
> of MPI, although I haven't used either myself.

Perhaps, but we don't see anyone using it.


-- 
*PS: A number of people ask me about the attachment to my email, which
is of type "application/pgp-signature". Don't worry, it is not a
virus. It is an electronic signature, that may be used to verify this
email came from me if you have PGP or GPG installed. Otherwise, you
may safely ignore this attachment.

----------------------------------------------------------------------------
A/Prof Russell Standish                  Phone 8308 3119 (mobile)
Mathematics                                    0425 253119 (")
UNSW SYDNEY 2052                         address@hidden             
Australia                                http://parallel.hpc.unsw.edu.au/rks
            International prefix  +612, Interstate prefix 02
----------------------------------------------------------------------------

pgpQf0hQTCh4m.pgp
Description: PGP signature