MathGroup Archive 1998

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

Search the Archive

Re: Mathematica frustrations...

  • To:
  • Subject: [mg11168] Re: Mathematica frustrations...
  • From: carlos@mars.Colorado.EDU (Carlos A. Felippa)
  • Date: Wed, 25 Feb 1998 03:31:42 -0500
  • Organization: University of Colorado, Boulder
  • References: <6cgf9k$>

In article <6cgf9k$> writes:
>I have recently started learning Mathematica.  I'm trying to  use it for
>some data analysis and plotting, but so far I'm not having much luck. 
>There are a few specific things  that  are  troubling me,  and  I'd
>really appreciate hearing any advice or suggestions to help me work
>around these problems.
>One "feature" I find extremely frustrating is the  numbering  and
>evaluation   order  of  equations  in  notebooks.   When  I  edit
>something early in a notebook, I would like  Mathematica  to  re-
>evaluate the  rest  of  the  notebook,  and  always  perform  the
>notebook  evaluation in a straight top-to-bottom order.  Is there any
>way to set an option to get  it  to  behave  that  way?   For things 
>like  "%"  references  to  previous  results,  having the definition of
>"previous" depend on what  you  edited  last  seems crazy.   I guess I
>prefer the Mathcad way, in which evaluation is always in a well-defined
>order determined by page layout.

The use of % is a legacy from the original In-Out days of interactive
command languages like Basic or Unix. It makes no sense for a Notebook
front end.  Notebooks are best organized in a procedural style, as 
collections of modules. What is still lacking in Mathematica is the 
ability to compile modules and complete cells to machine language and
to form a relocatable object library that can be loaded and executed at
high speed.  This will have to wait until (1) a typing system  is
introduced, and (2) all module symbols are made local by default and
placed in dynamic stacks. 

In fact it might be useful to get rid altogether of the In[ ] and Out[ ]
that still clutter cell boundaries since the numbers that appear
therein serve no useful purpose.   The symbol % could be reserved for
some other  purpose such as inline comment marker, or some kind of

>Another  problem  is  that  Mathematica  seems  to   reset   plot
>parameters  at  random, particularly after erroneous edits.  I am using
>the  MultipleListPlot  and  ListPlot  functions  to  create graphs.  
>After  I  get  the  graph sized how I like it (with the mouse), I will
>sometimes try changing a  few  parameters  in  the graph.   If  I  make
>a typo and Mathematica can't understand the command, my graph will jump
>to some random different size once  I get the error fixed.

No only plot parameters but also output cell formatting & display
parameters  are often (but not always) forgotten after errors.  That
could be fixed  in some future version.  But the best solution would be
to permit the input cell to specify those parameters, for example
Print[expr//MatrixForm, Background->Yellow,Width->300 TruePt,
Font->TimesItalics, TextColor->Red,FontSize->12 Pt].  This guarantees

>Finally, I'm about to give up on getting reasonable Legends on my plots.
In my experience the best way to get legends and labels is to display 
textstrings as plot objects.  Their position can be offset from that of
other graphical objects.  Precise plot sizing, however, remains
elusive because no absolute size specs are available for many
objects.  That may be contrasted with say TeX, which has a nice
two-tier <dim> system: pt, which is affected by magnification, 
and truept which is not.

>Does  the  PlotLegend  option work for MultipleListPlot? Sometimes it
>will show a Legend for me, and sometimes it won't.
>FYI, I'm using Mathematica 3.0 on Linux.

  • Prev by Date: Re: how to pull real numbers out of a linear funktion ?
  • Next by Date: lattice definition: help
  • Prev by thread: Re: Mathematica frustrations...
  • Next by thread: typesetting problem in Mathematica3 / NT