Re: Plotting Inf

[Daniel J Sebald]

Søren Hauberg wrote:
> Hi,
>   I seem to be experiencing some problems plotting Inf's. The following 
> code shows the problem
>
>   plot(1:5)
>   hold on, plot([3:4, inf]), hold off
>
> After this the gnuplot window hangs. If I press Ctrl-C in the octave 
> prompt after running the above code, I'm able to get gnuplot to repaint 
> its window, but then it just produces an empty plot. Matlab seems to 
> treat Inf just like NaN, i.e. don't plot the point. Octave seems to get 
> NaN's right, so I guess it's only a minor change.

It seems you've discovered a gnuplot and Octave bug.  The Inf is the problem, 
not the hold on/off.  So I just ran the following:

plot([3:4, Inf])
<CNTL-C>
fid = fopen("junk.gp","w");
__go_draw_axes__(gca,fid)
fclose(fid)

which creates a file with:

set origin 0, 0;
set size 1, 1;
set size noratio;
unset title;
unset xlabel;
unset ylabel;
unset zlabel;
set grid noxtics;
set grid noytics;
set grid ztics;
set grid nomxtics;
set grid nomytics;
set grid nomztics;
set xtics;
set ytics;
set ztics;
set xtics;
set ytics;
set ztics;
unset logscale x;
unset logscale y;
unset logscale z;
set style line 1 default;
set style line 1 linecolor rgb "#0000ff" linetype 1 linewidth 0.500000 pointsize 
1.000000;
set xrange [1:3] noreverse;
set yrange [NaN:NaN] noreverse;
set border 431;
unset key;
set style data lines;
plot "-" using ($1):($2) title "" with lines linestyle 1;
1 3
2 4
3 Inf
e

Of all the above code, it is the NaN inside

set yrange [NaN:NaN] noreverse;

that causes a problem.  (I will submit a bug report to gnuplot.)

On the other hand, another question is whether Octave should generate such a 
range [NaN:NaN] which I assume is a result of the Inf as part of the y data.

Dan