gnuplot development

On 23.03.2012 23:09, Rafael Reggiani Manzo wrote:

> Someone knows a way to splot the parametric equation (sin(u), cos(u),
> 10) together with two datasets of discrete points?

Just do it.

"help iteration" includes several examples. What, exactly, are you
trying to do?

> I have installed the 4.6 version and I'm looking for
> some examples that show the usage of loops and
> conditionals.
> 
> Can someone please point to specific scripts using these?

Thanks Tait, not too sure what exactly I'm doing yet. Just learning a very
nice tool for visualization.

http://i.imgur.com/pwgXT.gif

Okay, so the for-loop and do-statements are there
and "help if" tells you the whole story in *version 4.6*
AWESOME.

For anyone who might find this useful:

------------------------------->help
if-----------------------------------------

*New syntax:*
*       if (<condition>) { <command>; <command>*
*              <commands>*
*              <commands>*
*       } else {*
*              <commands>*
*       }*
* Old syntax:*
*       if (<condition>) <command-line> [; else if (<condition>) ...; else
...]*
*
*
* This version of gnuplot supports block-structured if/else statements. If
the*
* keyword `if` or `else` is immediately followed by an opening "{", then *
* conditional execution applies to all statements, possibly on multiple
input*
* lines, until a matching "}" terminates the block.  If commands may be
nested.*
*
*
* The old single-line if/else syntax is still supported, but can not be*
* mixed with the new block-structured syntax.  See `if-old`.*

Pau, not sure if this is what you need, but I saw this yesterday so...


Source :
http://en.wikipedia.org/wiki/Wikipedia:How_to_create_charts_for_Wikipedia_articles#gnuplot

So[image: Inline image 1]

Am 18.03.2012 16:07, schrieb Pau:
> Dear all,
> 
> I am trying to find in the internet a table explaining the numbering
> of lt and the type of line and its colour but I am not finding it...
> 
> Is it possible to choose the colour independently of the type of line
> (dashed, solid etc)?
> 


did you read 'help linecolor' ?

re,
 wh

> thanks!
> 
> Pau
> 
> ------------------------------------------------------------------------------
> This SF email is sponsosred by:
> Try Windows Azure free for 90 days Click Here 
> http://p.sf.net/sfu/sfd2d-msazure
> _______________________________________________
> gnuplot-info mailing list
> gnuplot-info@...
> https://lists.sourceforge.net/lists/listinfo/gnuplot-info
>

> I am trying to find in the internet a table explaining the numbering
> of lt and the type of line and its colour but I am not finding it...

Use "test" (as a command, by itself, after setting your terminal/output).

>>>>> "LB" == Leo Butler <l_butler@...> writes:

gnuplot> plot '-'
LB> input data ('e' ends) > -1. 1.
LB> input data ('e' ends) > 0. 0.99999999999999989
LB> input data ('e' ends) > e
LB>   C-c C-c^C

It also spins for me using tip of cvs HEAD, which reports:

  gnuplot 4.7 patchlevel 0

-JimC
-- 
James Cloos <cloos@...>         OpenPGP: 1024D/ED7DAEA6

> gnuplot> plot '-'
> LB> input data ('e' ends) > -1. 1.
> LB> input data ('e' ends) > 0. 0.99999999999999989
> LB> input data ('e' ends) > e
> LB>   C-c C-c^C
>
> It also spins for me using tip of cvs HEAD, which reports:
>
>   gnuplot 4.7 patchlevel 0

I can't replicate this, at least on 4.6.0. What OS, terminal,
and library versions for that terminal are you using?

Leo Butler <l_butler <at> users.sourceforge.net> writes:

> 
> Hi,
> 
> The following bug was noted on the Maxima email list:
> 
> gnuplot> plot '-'
> input data ('e' ends) > -1. 1.
> input data ('e' ends) > 0. 0.99999999999999989
> input data ('e' ends) > e
>   C-c C-c^C
> 
>         G N U P L O T
>         Version 4.4 patchlevel 0
>         last modified March 2010
>         System: Linux 3.2.0-2-686-pae
> 

there should have been the message:

         warning: tick interval too small for machine precision

and internally the interval between the tics is set to
'end-start'.

but in this special case this is not sufficient, because (due
to machine precision) 'end' and 'start' are 1.0 and the step
size is 1.11022e-16
moreover, again due to machine precision.
1.0+1.11022e-16 is 1.0
so the tics generating loop never ends.

would something like this make sense in 'gen_tics()' in 'axis.c'?

for (tic = start; tic <= end; tic += step) {
    if (tic == tic+step) {
        int_warn(NO_CARET, "endless loop when generating tics, stopped");
        return;
    }
...
}

>>>>> "T" == Tait  <gnuplot-info@...> writes:

gnuplot> plot '-'
LB> input data ('e' ends) > -1. 1.
LB> input data ('e' ends) > 0. 0.99999999999999989
LB> input data ('e' ends) > e
LB> C-c C-c^C

>> It also spins for me using tip of cvs HEAD, which reports:

T> I can't replicate this, at least on 4.6.0. What OS, terminal,
T> and library versions for that terminal are you using?

gcc-4.6.2 -march=amdfam10 -O3 (-m64)

The terminal is irrelevant.  It spins even with set term dumb.

Removing a single 9 from 0.99999999999999989 is enough to make it work.

The hang is in gen_tics().

When I interupted the gdb session, it was in gen_tics() at the gprintf()
call on line 1173, just before the /* Range-limited tic placement */
comment.  It reduces to:

  gprintf("1.00000000000000", 50, "%.14f", 1.0, 1);

at the gen_tics() level, but gdb says the format is just "f" when I'm at
the gprintf() level....

It must loop somewhere therein.

-JimC
-- 
James Cloos <cloos@...>         OpenPGP: 1024D/ED7DAEA6

I've found that Gnuplot will not accept as inputs all the time formats
that it supports as outputs---unfortunately.

On Thu, Mar 15, 2012 at 9:04 PM,  <david@...> wrote:
> I've got a situation where I want to plot values per week. I have taken my
> input file and run it through a script to format the time as week/year
> (as per the output of date -"%U/%Y"), however when I try to define the
> input time format as %U/%Y in gnuplot 4.4 I get an error that the file
> does not contain any valid data.
>
> is this a bug, or am I doing something wrong here?
>
> David Lang
>
> ------------------------------------------------------------------------------
> This SF email is sponsosred by:
> Try Windows Azure free for 90 days Click Here
> http://p.sf.net/sfu/sfd2d-msazure
> _______________________________________________
> gnuplot-info mailing list
> gnuplot-info@...
> https://lists.sourceforge.net/lists/listinfo/gnuplot-info



-- 
::::::::::::::::::::::::::::::::::::::::::::
David Lindelöf, Ph.D.
+41 (0)79 415 66 41 or skype:david.lindelof
Better Software for Tomorrow's Cities:
  http://computersandbuildings.com
Follow me on Twitter:
  http://twitter.com/dlindelof

On 15.03.2012 21:04, david@... wrote:
> I've got a situation where I want to plot values per week. I have taken my
> input file and run it through a script to format the time as week/year
> (as per the output of date -"%U/%Y"), however when I try to define the
> input time format as %U/%Y in gnuplot 4.4 I get an error that the file
> does not contain any valid data.

That should stop being a surprise once you realize that the format 
specifiers available for _output_ of time/date data are not the same as 
those available for _input_.  %U be cannot be used to read dates from a 
file.  See "help timefmt".

There are several formats that are documented as 'for output only' but the 
week of year ones are not.

David Lang

On Thu, 15 Mar 2012, David Lindel?f wrote:

> I've found that Gnuplot will not accept as inputs all the time formats
> that it supports as outputs---unfortunately.
>
> On Thu, Mar 15, 2012 at 9:04 PM,  <david@...> wrote:
>> I've got a situation where I want to plot values per week. I have taken my
>> input file and run it through a script to format the time as week/year
>> (as per the output of date -"%U/%Y"), however when I try to define the
>> input time format as %U/%Y in gnuplot 4.4 I get an error that the file
>> does not contain any valid data.
>>
>> is this a bug, or am I doing something wrong here?

On 15.03.2012 18:50, Stefan83 wrote:
>
> Hi all,
>
> Im really new in using gnuplot but I'm excited of the nice look of all the
> diagrams. But one thing I can't handle out.
> I have a logarithmic scale on bothe the x and y axe. For Fitting issues they
> don't start at exact 10^x values. So I want to have the start and end values
> of my axes plotted as well. For example, if my x axe starts with 4000 it
> should be prompt 4x10^3 at under the scale. In the picture you can see the
> original plot.

	set xtics add (4000)

Thanks for your reply! I found your topic on the 'gnuplot in action' forum
and we have indeed the same problem. The answer they gave is however very
clear ('I don't think gnuplot currently has this ability. You might want to
suggest it though on the gnuplot maintainers mailing list.')

Fitting a curve through the data doesn't really work for me, because the
data is not on or close to a single curve or surface and I am not always
interested in the same data points. However, the idea to extract data from
the original database can maybe help me further.

Some other ideas I had:

   - creating a 3D scatter, which is plotted in 2D. The third axis can be
   used to store extra information, which can be derived by rotating the plot.
   - maybe it is possible to use the coordinates of a mouse click to create
   a new plot in which the data point closest to those coordinates is labeled.
   I haven't tried this option yet, but I think it should be possible.

Feel free to send me other ideas, thoughts or experiences with this problem!

Anton



2012/3/15 carminat <carminat@...>

>
> I had the same problem (and asked the same question some time ago "in
> gnuplot
> in action" forum).
> Although I do not really have The Answer, I can tell what I did in my case,
> which I presume is quite typical.
>
> The data I was interested in were more or less on a extremal curve; I firs
> did some transform to stighten the curve, and then I  excluded  data which
> were far away from this locus  (you can do this as follows: use the "set
> table" option together the ternary operator ? to print into a file numbers
> of data lines which are near the curve, then use this information to
> extract
> these lines from your dataset).
>
> Thus you are left with a much smaller file, and plotting with labels is
> more
> reasonable (three digits labels are not so terrible). To do it in the best
> way, you should try to apply some transform so that the "curve" of
> interesting data is plotted as a vertical line (so that labels do not
> overlap).
>
> Of course, I would love to have "invisible labels", which are displayed on
> mouse clicks!
>
>
>
> antong wrote:
> >
> > I create 2D scatter plots from a text file, which works fine. But now I
> > want to retrieve additional information about a specific data point.
> > [...]
> > I thought that it might be possible to retrieve data from a point by
> > moving the mouse over that point. But I'm not sure if that is possible at
> > all. If not or if you have other ideas, pleas let me know!
> >
> > Anton
> >
> >
>
> --
> View this message in context:
> http://old.nabble.com/information-about-a-point-in-scatter-tp33478029p33508157.html
> Sent from the Gnuplot - User mailing list archive at Nabble.com.
>
>
>
> ------------------------------------------------------------------------------
> This SF email is sponsosred by:
> Try Windows Azure free for 90 days Click Here
> http://p.sf.net/sfu/sfd2d-msazure
> _______________________________________________
> gnuplot-info mailing list
> gnuplot-info@...
> https://lists.sourceforge.net/lists/listinfo/gnuplot-info
>