Menu
▾
▴
gnuplot-beta
|
From: Martin G. <mg...@in...> - 2007-07-01 12:16:52
|
Hi all, I'm using Gnuplot: Version 4.3 patchlevel 0 last modified June 2007 System: Darwin 8.10.1 And i got rendering problems if i use data-clipping with lines and =20 linespoint render mode plot 'AGdata-test.dat' using 4:($3=3D=3D1?$5:1/0) with lines,\ 'AGdata-test.dat' using 8:($7=3D=3D1?$9:1/0) with lines,\ 'AGdata-test.dat' using 12:($11=3D=3D1?$13:1/0) with lines This does not draw 3 connected 'paths' between the valid value. =20 Instead multiple broken line segments are drawn. plot 'AGdata-test.dat' using 4:($3=3D=3D1?$5:1/0) with points,\ 'AGdata-test.dat' using 8:($7=3D=3D1?$9:1/0) with points,\ 'AGdata-test.dat' using 12:($11=3D=3D1?$13:1/0) with points This in contrast seems do draw all valid point. So i suppose that =20 there is something wrong with lines. Can you please provide me with a patch or workarround to this problem. I attached the data-file for testing. Kind regards Martin Giersich ---------------------------------------------------- Dipl.-Inf. Martin Giersich Universit=C3=A4t Rostock IEF - IFI - MMIS www.informatik.uni-rostock.de/mmis Fon: +49-381-498-7514 Fax: +49-381-498-7522 mar...@in... ---------------------------------------------------- =EF=BF=BC |
|
From: <HBB...@t-...> - 2007-07-14 14:14:51
|
Martin Giersich wrote: > And i got rendering problems if i use data-clipping with lines and > linespoint render mode > > plot 'AGdata-test.dat' using 4:($3==1?$5:1/0) with lines,\ > 'AGdata-test.dat' using 8:($7==1?$9:1/0) with lines,\ > 'AGdata-test.dat' using 12:($11==1?$13:1/0) with lines Guess you've just found out why 1/0 is not the recommended way of spelling out "no valid data", in gnuplot. 0/0, which is well and truly undefined instead of just being infinite, usually works better. |
|
From: Ethan M. <merritt@u.washington.edu> - 2007-07-16 20:54:17
|
On Saturday 14 July 2007 07:15, Hans-Bernhard Br=F6ker wrote:
> Martin Giersich wrote:
>=20
> > And i got rendering problems if i use data-clipping with lines and=20
> > linespoint render mode
> >=20
> > plot 'AGdata-test.dat' using 4:($3=3D=3D1?$5:1/0) with lines,\
> > 'AGdata-test.dat' using 8:($7=3D=3D1?$9:1/0) with lines,\
> > 'AGdata-test.dat' using 12:($11=3D=3D1?$13:1/0) with lines=20
>=20
> Guess you've just found out why 1/0 is not the recommended way of=20
> spelling out "no valid data", in gnuplot. 0/0, which is well and truly=20
> undefined instead of just being infinite, usually works better.
Since you're using the CVS version, you also have the option
of using the symbol NaN ("not a number") directly:
plot 'AGdata-test.dat' using 4:($3=3D=3D1 ? $5 : NaN) with lines
=2D-=20
Ethan A Merritt
|
|
From: Martin G. <mg...@in...> - 2007-07-16 07:46:47
|
On 14.07.2007, at 16:15, Hans-Bernhard Br=C3=B6ker wrote: > Martin Giersich wrote: > >> And i got rendering problems if i use data-clipping with lines and =20= >> linespoint render mode >> plot 'AGdata-test.dat' using 4:($3=3D=3D1?$5:1/0) with lines,\ >> 'AGdata-test.dat' using 8:($7=3D=3D1?$9:1/0) with lines,\ >> 'AGdata-test.dat' using 12:($11=3D=3D1?$13:1/0) with lines > > Guess you've just found out why 1/0 is not the recommended way of =20 > spelling out "no valid data", in gnuplot. 0/0, which is well and =20 > truly undefined instead of just being infinite, usually works better. No change using 0/0. Same broken line drawing. I attached a plot- =20 output of th above plot-command. This rendering problems appear using MacOSX 10.4.... and gnuplot 4.0, =20= 4.2, 4.3. Btw, the official gnuplot-documentation (Thomas Williams & Colin =20 Kelley, "gnuplot - An Interactive Plotting Program", 3 March 2007, =20 available at: http://www.gnuplot.info/docs/gnuplot.pdf) describes at =20 page 25 using "1/0" as a way that "may be used to generate an =20 "undefined" flag, which causes a point to ignored". No other method to do this is refered. If there is another method to do this - besides change of dat-file =20 format - i'd like to know... Regards, Martin ---------------------------------------------------- Dipl.-Inf. Martin Giersich Universit=C3=A4t Rostock IEF - IFI - MMIS www.informatik.uni-rostock.de/mmis Fon: +49-381-498-7514 Fax: +49-381-498-7522 mar...@in... ---------------------------------------------------- =EF=BF=BC |
|
From: Ethan M. <merritt@u.washington.edu> - 2007-07-23 04:03:11
|
On Monday 16 July 2007, Martin Giersich wrote: > Btw, the official gnuplot-documentation (Thomas Williams & Colin > Kelley, "gnuplot - An Interactive Plotting Program", 3 March 2007, > available at: http://www.gnuplot.info/docs/gnuplot.pdf) describes at > page 25 using "1/0" as a way that "may be used to generate an > "undefined" flag, which > causes a point to ignored". No other method to do this is refered. > If there is another method to do this - besides change of dat-file > format - i'd like to know... Starting with version 4.2 there is a pre-defined variable named NaN which may be used for this purpose. |
|
From: Theo H. <th...@ph...> - 2007-07-23 17:48:02
|
Martin Giersich wrote: > > No change using 0/0. Same broken line drawing. I attached a plot- output > of th above plot-command. > > This rendering problems appear using MacOSX 10.4.... and gnuplot 4.0, > 4.2, 4.3. > > Btw, the official gnuplot-documentation (Thomas Williams & Colin > Kelley, "gnuplot - An Interactive Plotting Program", 3 March 2007, > available at: http://www.gnuplot.info/docs/gnuplot.pdf) describes at > page 25 using "1/0" as a way that "may be used to generate an > "undefined" flag, which > causes a point to ignored". No other method to do this is refered. > If there is another method to do this - besides change of dat-file > format - i'd like to know... > gnuplot is behaving exactly as documented. If you plot `with linespoints`, you will see that the expected points are present. However, not all points are joined with lines, because there is missing or invalid data between them. This is usually important information, and so gnuplot indicates this. If you really want to have the lines joined, you can do something like: plot "<awk '$3==1 { print }' AGdata-test.dat" using 4:5 w li You're effectively changing the data file format on the fly to omit the invalid data. gnuplot won't see it, and so will draw lines between the valid data points. THeo |
