MathGroup Archive 2009

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Re: Re: condense axis

  • To: mathgroup at smc.vnet.net
  • Subject: [mg103343] Re: [mg103323] Re: [mg103299] Re: condense axis
  • From: "David Park" <djmpark at comcast.net>
  • Date: Wed, 16 Sep 2009 05:47:49 -0400 (EDT)
  • References: <23668144.1252713896870.JavaMail.root@n11> <h8g0fd$bq3$1@smc.vnet.net> <200909141110.HAA08862@smc.vnet.net> <820DEF16-E99B-43CA-BC42-CDD062578786@me.com> <23978087.1253008544586.JavaMail.root@n11>

I wasn't the other correspondent but certainly it is not out of bounds to
criticize buying a package if one can program it yourself. A lot of people
feel that way. There is nothing in Presentations that one couldn't program
for oneself. The question is whether it is time efficient. (And, of course,
the package tackles many more problems than the broken-axis problem.)

I seriously question whether a reasonably competent Mathematica user could
program and test the ySplitPlot routine in 10 minutes. It has a lot tricky
little things in it like using AbsoluteOptions, understanding the tick data
format, chopping and shifting and using Sequence. Also it is not as general
as seems to be claimed. The splitlines appear to be geared to a specific
plot scale and would probably have to be modified in other cases. It is not
a simple tweak to change it to accommodate different scales for the two
functions - because it is modifying a single set of linear scale ticks.

Then there is the matter of demanding an "exact" solution. We don't know
what an exact solution is because the original poster only asked a generic
question. We don't know what his functions look like, whether he wants
different scales or whether he might demand an actual break in the y axis
and not just a couple of cross lines. In any case, I did in fact produce two
solutions using the existing Presentations.

When I decided we should have nice YAxisBreak and XAxisBreak graphics
routines that a user could use to plug the gap between two tick lines it
took me several hours to program so they would be convenient to use. Then it
took me another hour to plug them into Presentations with Workbench and
write links and documentation Function pages. The buyer of the package has
them handy, gets usage messages and SyntaxInformation and documentation
pages with notes, examples and links to associated routines. And that is
just a small addition to what is in the package. If a user programs these
routines himself (especially in 10 minutes), where is he going to put them?
Will he write all the documentation? Will he find it a year from now and
remember how it works? Is this just a one-off thing? Will he be able to
search MathGroup and find ySplitPlot? 

Mathematica is somewhat torn between two approaches. One approach might be
called the "toolkit" approach. Mathematica provides the tools and it is up
to the user to put them together for a solution. And Mathematica does
provide plenty of tools. The other approach is what I call the "set-piece"
approach. Mathematica provides a set-piece routine that is very powerful,
impressive and does some commonly useful thing. Examples might be Plot3D or
Manipulate. The set-piece routines are great and usually adaptable, but the
more you try to stretch them the more difficult they become and finally they
may peter out. One sees many questions on MathGroup from people trying to go
beyond the set-piece routines and having great difficulty. The toolkit
approach may generally involve more work (because it doesn't make a lot of
automatic choices for you) but it is more intuitive work and for most custom
graphics it is less work than trying to outfox a set-piece routine.

And, in general, it is a matter of development and investigation as to the
best methods to present mathematical and technical ideas using all the
facilities that Mathematica provides. We don't know the best methods and
that is why I believe a toolkit approach is the best approach. The
Presentations package is very much oriented to providing convenient tools
for building up custom presentations of ideas.

As to a $50 cost for a Mathematica package, I have read that good, non Ivy
League, private colleges cost more than $50,000 per year. Assuming that a
student only attends 3/4 of the year, 6 days a week, with 10 hours a day in
actual school work, that comes out to $21/hour that the student is paying
for the privilege of learning within an academic environment. I have to save
him 2 1/2 hours over a year in learning the same material to justify the
package.
 

David Park
djmpark at comcast.net
http://home.comcast.net/~djmpark/  


From: Peter Falloon [mailto:pfalloon at gmail.com] 


Check out SpheroidalPS, SpheroidalEigenvalue and related...

(For the record, the package was -- and still is -- *freely* available  
at http://ftp.physics.uwa.edu.au/pub/Theses/2002/Falloon/ )


It seems I've ruffled a few feathers, so let me repeat what I just  
wrote offline to another correspondent:

"I apologize if I've caused offence here: let me be clear that I'm not  
trying to impugn the package itself. All I'm saying is that if it  
doesn't address the specific question, it seems a little cynical to go  
recommending it?

It would be a different story if a specific implementation using the  
package was demonstrated, and a good argument given as to why this was  
easier than, or superior to, what could be achieved in Mathematica  
alone. If that's the case,
why not do it (my implementation took only about 10 minutes to whip up  
and is reasonably workable)?"

Cheers,
Peter.

On 15 Sep 2009, at 05:52, Syd Geraghty wrote:

> Hi Peter,
>
> I cannot imagine a package that does everything you want for any  
> amount of dollars (even Australian dollars). :-)
>
> The Presentations Package addresses a great many issues of  
> exposition and graphical output and of course split axis graphs are  
> a minor issue.
>
> I believe most of us who use the Presentations Package think it a  
> great value and wish WRI would incorporate David Park's work in the  
> standard distribution of Mathematica.
>
> Some of us have lobbied for that to happen directly with WRI  
> marketing folks and I hope that more Presentations users will add  
> their support to that effort.
>
> This is not an easy process as it appears you have experienced.
>
> In The Mathematica Journal Volume 10, Issue 3, 2007 I recently came  
> across your article:
>
> A Mode-Matching Method for Multichannel Scattering Problems by Peter  
> Falloon.
>
> As a sidebar it was mentioned that:
>
> " Falloon developed a Mathematica package to compute spheroidal  
> wavefunctions, which will be incorporated into the next version of  
> the Mathematica system. "
>
> I have not been able to find any references yet in V7. Que pasa?
>
> Cheers ...  Syd
>
>
>
>
>
> Syd Geraghty B.Sc, M.Sc.
>
> sydgeraghty at mac.com
>
> Mathematica 7.0.1 for Mac OS X x86 (64 - bit) (18th February 2009)
> MacOS X V 10.6 Snow LeopardMacBook Pro 2.33 GHz Intel Core 2 Duo   
> 2GB RAM
>
>
>
>
> On Sep 14, 2009, at 4:10 AM, pfalloon wrote:
>
>> Hi David,
>> Well, with all due respect, I wouldn't recommend paying $50 for a
>> package unless it could do *exactly* what you needed.
>





  • Prev by Date: Wolfram workbench on MacOsX
  • Next by Date: GUIKit: How to use a widget "Tree"?
  • Previous by thread: Re: Re: condense axis
  • Next by thread: Re: Re: condense axis