Re: How to get into Scientific Programming

In article <[EMAIL PROTECTED]>,
   "R. Martin" <[EMAIL PROTECTED]> wrote:
>[EMAIL PROTECTED] wrote:
>> 
>> In article <[EMAIL PROTECTED]>,
>>    [EMAIL PROTECTED] (Russell Martin) wrote:
>> >[EMAIL PROTECTED] wrote in message 
news:<[EMAIL PROTECTED]>...
>> >> In article <[EMAIL PROTECTED]>,
>> >>    [EMAIL PROTECTED] (Old Pif) wrote:
>> >> >[EMAIL PROTECTED] wrote in message
>> news:<[EMAIL PROTECTED]>...
>> >> >>
>> >> >> If you're talking about the percentage of total jobs, I
>> >> >> sure would hope so.  The industry doesn't need a gazillion
>> >> >> interpretations of the FORTRAN^WFortran standard or anything
>> >> >> else to do with computing arithmetic.
>> >> >>
>> >> >
>> >> >Let me recollect the beginning of this discussion. It was not about
>> >> >what industry needs it was about how easy to get the job. So, if you
>> >> >device some measure like, say, #of_position_that_industry_needs/#
>> >> >of_guys_who_believe_they_are_scientific_programmers,
>> >> >the value is somewhere in the vicinity 1/gazillion. And that gives 
you
>> >> >a rough idea how easy it is.
>> >>
>> >> It shouldn't be.  I knew a lot of people who had larger opinions
>> >> of their abilities than reality proved.  My point is that the
>> >> people who wrote the FORTRAN library were not scientific programmers.
>> >
>> >So, are you implying that someone who "just" programs, say, FFTs etc.
>> >for NAG or Mathematica are NOT scientific programmers,
>> 
>> NO, I'm not implying that.  I am stating that the people who did
>> the FORTRAN development and maintenance to be used by scientific
>> programmers were not scientific programmers.  I know they weren't;
>> I was one of them.  It was one of the ways we made money--by selling
>> you the FORTRAN software which consisted of the compiler, the runtime
>> package and the object time system.
>
>OK.  But the only part of the package that I consider to be the
>product of what I would call scientific programming is the mathematical
>functions like sin, tan, etc., and maybe the floating point math,
>anyway.  

These are grey areas.  If it's a function that is so common that
it is included in a general distribution package, I probably
wouldn't expect the code to be written by scientific programmers.
The floating point stuff depends on the CPU architecture.  I hope
you can tell that I think one layer below yours.

> ..The writing of compilers per se are not what I consider to be
>scientific programming.  So while I agree with your point, I think
>it is somewhat moot.  However, I didn't read every contibution to
>this thread closely, so I may have missed some of the thrust of your
>argument.

Oh, but it's not moot for the guy who thinks that he'ld be happier
working in the equivalent of a FORTRAN development group.  That's
the point I've been trying to make, albeit badly.  If the OP 
really thinks that working in a group, that makes the basic
package you guys use for your work, is scientific programming,
he's going to be extremely unhappy.  This work is straight-jacketed.
The developer can't do any old thing he pleases (which is the
impression I got from the OP--he was very unhappy with straight-jacketed
work).  In some aspects of the development, not only are you 
straight-jacketed but your thumbs are duct-taped to opposite walls.

<snip>

>> See?  People who bought our stuff didn't have to do that :-).
>> I doubt that your work involved writing your own compiler
>> and object time system.  You used something that compiler and OTS
>> people wrote.  Then you wrote what had to be customized for your
>> site.
>
>Right, but like I said I don't consider the compiler to be, on the
>whole, the product of the process called scientific programming.

But I got the impression that OP thought it was.

>
>> >
>> >> >
>> >> >> But that's NOT scientific programming.  Using your logic:
>> >> >> Gregory uses screwdrivers; therefore, I'm a scientist because
>> >> >> I used a screwdriver to reattach my cupboard doors.
>> >> >>
>> >> >
>> >> >No, using my logic: Gregory is not a screwdriver maker.
>> >>
>> >> Thus, the screwdriver maker isn't a scientist; they're
>> >> screwdriver makers.
>> >
>> >They are if they use the methods of science to develop new
>> >screwdrivers. :-)
>> 
>> You mean new bits.  The basic tool is still there.
>
>No, I mean any basically new screwdriver which might be developed
>(ala Dr. Who's sonic screwdriver ;-) ).

<GRIN>  It is scientific programming IFF you don't get into
the business of working only on that sonic screwdriver for
commercial sales.  That's how a lot of scientists ended up
retiring from the software development biz.

<snip>
/BAH

Subtract a hundred and four for e-mail.