acquiring standard

0 views
Skip to first unread message

Lane Straatman

unread,
Feb 7, 2007, 2:31:27 PM2/7/07
to
Is the standard available as a .pdf file? LS


glen herrmannsfeldt

unread,
Feb 7, 2007, 2:44:58 PM2/7/07
to
Lane Straatman wrote:

> Is the standard available as a .pdf file? LS

Which standard? As far as I know, the 2003 standard isn't available
free, but the last draft of the standard is. That is close enough
for me. It is called n3501.pdf, google should be able to find it.

Fortran 66 is available as a scanned PDF. The Fortran 77 I know of
is HTML, there might also be a PDF. Others will probably tell you
about 90 and 95.

-- glen

Lane Straatman

unread,
Feb 7, 2007, 3:23:27 PM2/7/07
to

"glen herrmannsfeldt" <g...@ugcs.caltech.edu> wrote in message
news:RuSdnREGZ99WtlfY...@comcast.com...

> Lane Straatman wrote:
>
>> Is the standard available as a .pdf file? LS
>
> Which standard? As far as I know, the 2003 standard isn't available
> free, but the last draft of the standard is. That is close enough
> for me. It is called n3501.pdf, google should be able to find it.
There's an n3501.pdf about nesting biologies and another in chinese, but I
did find what you meant with the keywords "n3501.pdf download fortran". Thx.

> Fortran 66 is available as a scanned PDF. The Fortran 77 I know of
> is HTML, there might also be a PDF. Others will probably tell you
> about 90 and 95.

I don't really have a need to get into the history, as I've only been
looking at fortran for about six months now. I have no legacy code. LS


Bart Vandewoestyne

unread,
Feb 7, 2007, 5:09:25 PM2/7/07
to
On 2007-02-07, Lane Straatman <inv...@invalid.net> wrote:
>
>> Which standard? As far as I know, the 2003 standard isn't available
>> free, but the last draft of the standard is. That is close enough
>> for me. It is called n3501.pdf, google should be able to find it.
>
> There's an n3501.pdf about nesting biologies and another in chinese, but I
> did find what you meant with the keywords "n3501.pdf download fortran". Thx.

Seems like your Google does a worse job than my Google:

http://www.dkuug.dk/jtc1/sc22/open/n3501.pdf

Regards,
Bart

--
"Share what you know. Learn what you don't."

Tom Micevski

unread,
Feb 7, 2007, 7:17:38 PM2/7/07
to
Bart Vandewoestyne wrote:
> On 2007-02-07, Lane Straatman <inv...@invalid.net> wrote:
>>> Which standard? As far as I know, the 2003 standard isn't available
>>> free, but the last draft of the standard is. That is close enough
>>> for me. It is called n3501.pdf, google should be able to find it.
>> There's an n3501.pdf about nesting biologies and another in chinese, but I
>> did find what you meant with the keywords "n3501.pdf download fortran". Thx.
>
> Seems like your Google does a worse job than my Google:
>
> http://www.dkuug.dk/jtc1/sc22/open/n3501.pdf

that file's dated from sep 2002, which is not the latest draft of the
f2003 standard. the file you want is named "04-007.pdf" (may 2004) and
you can find it on the www.j3-fortran.org web site. you can find the f95
final draft there also.

you might also find this thread useful (especially richard maine's comments):
<http://groups.google.com.au/group/comp.lang.fortran/browse_frm/thread/bdbed3abf08af66>

Bart Vandewoestyne

unread,
Feb 8, 2007, 4:11:08 AM2/8/07
to
On 2007-02-08, Tom Micevski <no...@none.au> wrote:
>
> that file's dated from sep 2002, which is not the latest draft of the
> f2003 standard. the file you want is named "04-007.pdf" (may 2004) and
> you can find it on the www.j3-fortran.org web site. you can find the f95
> final draft there also.

OK. Thanks for this information. So to conclude, this is what
the people that don't want to spend a whole bunch of money for
buying standards can use:

Fortran 95:
http://j3-fortran.org/doc/year/97/97-007r2/pdf/97-007r2.pdf

Fortran 2003:
http://www.j3-fortran.org/doc/year/04/04-007.pdf

Right?

Best wishes,

Tom Micevski

unread,
Feb 8, 2007, 5:20:35 PM2/8/07
to
Bart Vandewoestyne wrote:
> On 2007-02-08, Tom Micevski <no...@none.au> wrote:
>> that file's dated from sep 2002, which is not the latest draft of the
>> f2003 standard. the file you want is named "04-007.pdf" (may 2004) and
>> you can find it on the www.j3-fortran.org web site. you can find the f95
>> final draft there also.
>
> OK. Thanks for this information. So to conclude, this is what
> the people that don't want to spend a whole bunch of money for
> buying standards can use:
>
> Fortran 95:
> http://j3-fortran.org/doc/year/97/97-007r2/pdf/97-007r2.pdf
>
> Fortran 2003:
> http://www.j3-fortran.org/doc/year/04/04-007.pdf
>
> Right?

Right.

Lane Straatman

unread,
Feb 8, 2007, 10:02:56 PM2/8/07
to

"Tom Micevski" <no...@none.au> wrote in message
news:45cba232$0$1149$61c6...@un-2park-reader-01.sydney.pipenetworks.com.au...
Right. I have new books at hand. They are a handsome pair by Press,
Teukolsky, Vetterling and Flannery.

Are their works talked about here? LS


Tom Micevski

unread,
Feb 9, 2007, 8:38:07 AM2/9/07
to
Lane Straatman wrote:
> Right. I have new books at hand. They are a handsome pair by Press,
> Teukolsky, Vetterling and Flannery.
>
> Are their works talked about here? LS

sounds like you have numerical recipes in f77 and f90. they are good books.

Gordon Sande

unread,
Feb 9, 2007, 9:00:10 AM2/9/07
to

Sounds more like Numerical Recipes. Try sci.math.num-analysis where
you will find that they are viewed as broad introductions but not
up to "industrial quality" standards. One quickly runs out of polite
bland expressions just as one runs out of simple examples where their
stuff still works well. The C versions appear to viewed with "curiosity"
as they use various hacks to use the base 1 indexing of the Fortran
originals which confuses C programmers no end.


Beliavsky

unread,
Feb 9, 2007, 1:23:04 PM2/9/07
to
On Feb 8, 10:02 pm, "Lane Straatman" <inva...@invalid.net> wrote:

<snip>

> Right. I have new books at hand. They are a handsome pair by Press,
> Teukolsky, Vetterling and Flannery.
>
> Are their works talked about here? LS

I consider discussion of Numerical Recipes in Fortran to be topical
here, although a better forum could be the Numerical Recipes Forum at
http://www.numerical-recipes.com/forum/ . Since Numerical Recipes is
not open source, many newsgroup readers do not have access to it and
cannot answer questions about it, and someone with a question cannot
post a significant chunk of NR code. By contrast, anyone with the
interest can answer a question about public domain Fortran code at
Netlib and other sites.

Lane Straatman

unread,
Feb 9, 2007, 6:01:12 PM2/9/07
to

"Beliavsky" <beli...@aol.com> wrote in message
news:1171045383.9...@m58g2000cwm.googlegroups.com...
Thanks all for replies, and I'm glad that I asked the question before
lumbering ahead. In retrospect, I should have said the titles, "Numerical
Recipes in Fortran 90" and "Numerical Recipes in Fortran, second edition."

I believe that in order to get anything off the ground, one needs to use a
module named nrtype and nrutil, which look to be ten to twenty pages in the
back of the book of proprietary stuff. Does this make it all more trouble
than its worth? LS


glen herrmannsfeldt

unread,
Feb 9, 2007, 6:31:06 PM2/9/07
to
Lane Straatman wrote:
(snip)

> Thanks all for replies, and I'm glad that I asked the question before
> lumbering ahead. In retrospect, I should have said the titles, "Numerical
> Recipes in Fortran 90" and "Numerical Recipes in Fortran, second edition."

I usually use it more as a reference book than as a black box
full of code. They cover many things reasonably well, yet readable
even if you haven't studied that particular type of problem before.

> I believe that in order to get anything off the ground, one needs to use a
> module named nrtype and nrutil, which look to be ten to twenty pages in the
> back of the book of proprietary stuff. Does this make it all more trouble
> than its worth? LS

Well, it won't be a module in the Fortran 77 version. I think there is
an error message routine called by many other programs. You can always
rewrite that part of the program to do something else in case of errors.
The routines are also available on floppy disk to save typing.

If you want black box code, (that is, run but don't look inside), netlib
might be a better choice. If you want to see how it works, possibly
modify it for your situation, and have a readable explanation for the
algorithms used, I believe NR is a good choice.

-- glen

Tom Micevski

unread,
Feb 9, 2007, 9:10:11 PM2/9/07
to

although, you can download pdfs of the entire books from the nr web site.

Brooks Moses

unread,
Feb 10, 2007, 4:35:26 AM2/10/07
to
Gordon Sande wrote:
> Sounds more like Numerical Recipes. Try sci.math.num-analysis where
> you will find that they are viewed as broad introductions but not
> up to "industrial quality" standards. One quickly runs out of polite
> bland expressions just as one runs out of simple examples where their
> stuff still works well. The C versions appear to viewed with "curiosity"
> as they use various hacks to use the base 1 indexing of the Fortran
> originals which confuses C programmers no end.

The C version is, however, IMO more useful to Fortran programmers than
the Fortran version -- since the license on the code prohibits most use
in anything where one might want to distribute the source, but the
authors specifically point out in the introduction that they consider
reimplementation of the algorithms to be a valid way "around" that, it's
much more convenient to have something non-Fortran to work from. :)

I will say that when I needed a decent multidimensional optimizer a
couple of years ago, the algorithm that I got out of Numerical Recipes
was far better than anything I could come up with myself on short
notice, and certainly served my purpose. Sometimes a broad introduction
is what the job calls for.

- Brooks


--
The "bmoses-nospam" address is valid; no unmunging needed.

Beliavsky

unread,
Feb 11, 2007, 4:55:41 PM2/11/07
to
On Feb 10, 4:35 am, Brooks Moses <bmoses-nos...@cits1.stanford.edu>
wrote:

Translating from C to Fortran the Numerical Recipes codes, which may
not be as robust as some of the public domain Fortran codes at Netlib
and other places, sounds like a waste of time to me.
Besides Googling for Fortran code and browsing Netlib, one can also
look at the Fortran: Source Code section of the Open Directory at
http://dmoz.org/Computers/Programming/Languages/Fortran/Source_Code/ .

John Harper

unread,
Feb 11, 2007, 7:56:18 PM2/11/07
to
In article <1171230941.2...@v45g2000cwv.googlegroups.com>,

Beliavsky <beli...@aol.com> wrote:
>On Feb 10, 4:35 am, Brooks Moses <bmoses-nos...@cits1.stanford.edu>
>wrote:
>> Gordon Sande wrote:
>> > Sounds more like Numerical Recipes. Try sci.math.num-analysis where
>> > you will find that they are viewed as broad introductions but not
>> > up to "industrial quality" standards.
>>
>> I will say that when I needed a decent multidimensional optimizer a
>> couple of years ago, the algorithm that I got out of Numerical Recipes
>> was far better than anything I could come up with myself on short
>> notice, and certainly served my purpose.
>
>Translating from C to Fortran the Numerical Recipes codes, which may
>not be as robust as some of the public domain Fortran codes at Netlib
>and other places, sounds like a waste of time to me.

A major defect of Numerical Recipes codes that I found is that there
isn't enough error-trapping for bad inputs. However, my copy is the
1988 printing, and I don't know if Press et al. have fixed that defect
since. If one is going to translate the codes, or copy them by hand
from the book (OK if you don't distribute them to others according
to page xiii) one might as well include suitable error traps.

-- John Harper, School of Mathematics, Statistics and Computer Science,
Victoria University, PO Box 600, Wellington 6140, New Zealand
e-mail john....@vuw.ac.nz phone (+64)(4)463 5341 fax (+64)(4)463 5045


glen herrmannsfeldt

unread,
Feb 12, 2007, 2:46:47 AM2/12/07
to
John Harper wrote:

(snip)

> A major defect of Numerical Recipes codes that I found is that there
> isn't enough error-trapping for bad inputs. However, my copy is the
> 1988 printing, and I don't know if Press et al. have fixed that defect
> since. If one is going to translate the codes, or copy them by hand
> from the book (OK if you don't distribute them to others according
> to page xiii) one might as well include suitable error traps.

That probably agrees with the way I think they should be used.

That is, not as black boxes, but as research aids. Maybe
as sample implementations of algorithms, which one can use for
comparison, and as a starting point in generating an implementation
of an algorithm. I don't know if that is the way the authors expect
them to be used, but that is how I look at it. I do appreciate the
explanations of the theory behind the algorithms, sometimes more than
the actual code. If I did use the code, I likely would modify things
like error trapping consistent with my needs.

-- glen

Gary Scott

unread,
Feb 12, 2007, 7:42:20 AM2/12/07
to

I tend to think that a public API should follow certain "good practices"
including error trapping and reporting. However, sometimes error
reporting is costly in terms of performance and so it is left out
intentionally. A mainframe graphics API had extensive runtime error
trapping. To improve performance, the user could identify a stub
routine and link that with the application. That had the effect of
no-oping the error trapping and improving performance (I assume there
remained some overhead due to the stub).

>
> -- glen
>


--

Gary Scott
mailto:garylscott@sbcglobal dot net

***** 5 Jan: Back from 7 days in Cozumel! *****

Fortran Library: http://www.fortranlib.com

Support the Original G95 Project: http://www.g95.org
-OR-
Support the GNU GFortran Project: http://gcc.gnu.org/fortran/index.html

If you want to do the impossible, don't hire an expert because he knows
it can't be done.

-- Henry Ford

Lane Straatman

unread,
Feb 12, 2007, 8:43:34 PM2/12/07
to

"Gordon Sande" <g.s...@worldnet.att.net> wrote in message
news:2007020910001016807-gsande@worldnetattnet...
> On 2007-02-08 23:02:56 -0400, "Lane Straatman" said:
>> "Tom Micevski" wrote in message
>>> Bart Vandewoestyne wrote:
I took a look at the forum you mentioned and think it will be more
appropriate for _Numerical Recipes_. I set the follow-up of this to go
there. It'll take a bit before I have a good question for number crunchers.
But when I do, I'll reply to this and have this context for them. I had
been using the german version of this ng to fix up my math. LS


Reply all
Reply to author
Forward
0 new messages