gnuplot-info — For questions and discussion about gnuplot

Re: [Gnuplot-info] Multiple sources on a common X axis

[BBands <bb...@gm...>]

A one word change should clear that up.

"The command set multiplot places gnuplot in the multiplot mode, in
which several
_independent_ plots are placed on the same page, window, or screen."

    John

On Thu, Jul 28, 2016 at 11:35 AM, Kevin Oberman <rko...@gm...> wrote:

> On Thu, Jul 28, 2016 at 10:38 AM, Hans-Bernhard Bröker <
> HBB...@t-...> wrote:
>
> > Am 28.07.2016 um 11:03 schrieb Robert von Knobloch:
> >
> > > Using multiplot, I would like to be able to align the temperature with
> > > the date/time of the data, so that the correlation between temp and
> data
> > > can be easily viewed.
> >
> > You really should _not_ be using multiplot for that.  Just plot both
> > data sets in a single 'plot' command, and you'll be done with it.
> >
> > I keep wondering how so many people arrive at the belief that they
> > should be using multiplot for so many jobs it's the completely wrong
> > tool for.  Is the concept of
> >
> >         plot 'file1', 'file2'
> >
> > really that hard to pick up from the documentation and demos?
> >
> >
> >
> ------------------------------------------------------------------------------
> >
>
> In a word, yes!
>
> "set multiplot" just looks like the right thing when you want multiple
> plots to one screen or file. The manual makes it look like that is the
> answer:
> "The command set multiplot places gnuplot in the multiplot mode, in which
> several plots are placed on
> the same page, window, or screen."
>
> I was lucky enough to read the manual for "plot" and wee that it took
> multiple files, each with the standard options for style and it just
> worked. But, if you search the manual for "multi", you find "multiplot over
> and over along with the repeated (and correct) statement:
> "Remember to close the output file before next plot unless in multiplot
> mode."  makes it sound like multiplot is the intended tool for placing
> multiple data sets onto a single plot when it is actually intended for a
> similar, but different purpose.
>
> I wish the "set multiplot" section had something near that top pointing out
> that 'plot "file1", "file2", ..." is another, possibly more appropriate
> approach.
>

Re: [Gnuplot-info] Multiple sources on a common X axis

[Kevin O. <rko...@gm...>]

On Thu, Jul 28, 2016 at 10:38 AM, Hans-Bernhard Bröker <
HBB...@t-...> wrote:

> Am 28.07.2016 um 11:03 schrieb Robert von Knobloch:
>
> > Using multiplot, I would like to be able to align the temperature with
> > the date/time of the data, so that the correlation between temp and data
> > can be easily viewed.
>
> You really should _not_ be using multiplot for that.  Just plot both
> data sets in a single 'plot' command, and you'll be done with it.
>
> I keep wondering how so many people arrive at the belief that they
> should be using multiplot for so many jobs it's the completely wrong
> tool for.  Is the concept of
>
>         plot 'file1', 'file2'
>
> really that hard to pick up from the documentation and demos?
>
>
> ------------------------------------------------------------------------------
>

In a word, yes!

"set multiplot" just looks like the right thing when you want multiple
plots to one screen or file. The manual makes it look like that is the
answer:
"The command set multiplot places gnuplot in the multiplot mode, in which
several plots are placed on
the same page, window, or screen."

I was lucky enough to read the manual for "plot" and wee that it took
multiple files, each with the standard options for style and it just
worked. But, if you search the manual for "multi", you find "multiplot over
and over along with the repeated (and correct) statement:
"Remember to close the output file before next plot unless in multiplot
mode."  makes it sound like multiplot is the intended tool for placing
multiple data sets onto a single plot when it is actually intended for a
similar, but different purpose.

I wish the "set multiplot" section had something near that top pointing out
that 'plot "file1", "file2", ..." is another, possibly more appropriate
approach.
--
Kevin Oberman, Part time kid herder and retired Network Engineer
E-mail: rko...@gm...
PGP Fingerprint: D03FB98AFA78E3B78C1694B318AB39EF1B055683

Re: [Gnuplot-info] Multiple sources on a common X axis

[Craig D. <def...@bo...>]

(Mobile)

> On Jul 28, 2016, at 11:38 AM, Hans-Bernhard Bröker <HBB...@t-...> wrote:
> 
> I keep wondering how so many people arrive at the belief that they 
> should be using multiplot for so many jobs it's the completely wrong 
> tool for.  Is the concept of
> 
>    plot 'file1', 'file2'
> 
> really that hard to pick up from the documentation and demos?
> 

Actually, yes, I think it is.  Gnuplot uses a different paradigm than most other plotting package: in essence, one is specifying a plot object (axes, frame, etc.) at a time.  Most other plotting packages instead approach the problem from a procedural standpoint: conceptually, one is issuing instructions for how to manipulate a plot object, until it gets to the final desired state.  

That paradigmatic difference between gnuplot and, say, matplotlib could probably do with some introductory material inthegeneral overview at the top of the documentation...





> ------------------------------------------------------------------------------
> _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> Membership management via: https://lists.sourceforge.net/lists/listinfo/gnuplot-info
> 

Re: [Gnuplot-info] Multiple sources on a common X axis

[Hans-Bernhard B. <HBB...@t-...>]

Am 28.07.2016 um 11:03 schrieb Robert von Knobloch:

> Using multiplot, I would like to be able to align the temperature with
> the date/time of the data, so that the correlation between temp and data
> can be easily viewed.

You really should _not_ be using multiplot for that.  Just plot both 
data sets in a single 'plot' command, and you'll be done with it.

I keep wondering how so many people arrive at the belief that they 
should be using multiplot for so many jobs it's the completely wrong 
tool for.  Is the concept of

	plot 'file1', 'file2'

really that hard to pick up from the documentation and demos?

[Gnuplot-info] Multiple sources on a common X axis

[Robert v. K. <bo...@en...>]

Hallo,

I am logging certain industrial processes, with the data in a single file.

Noticing some temperature effects, I have added a temperature sensor, 
but this logs, asynchronously, to a second file.

Both files (data and temperature) have timestamps in known fields.

Using multiplot, I would like to be able to align the temperature with 
the date/time of the data, so that the correlation between temp and data 
can be easily viewed.

The temperature has a different start date and period times.

Is this possible?

Many thanks,

Bob von Knobloch

Re: [Gnuplot-info] Having dual Celsius and Fahrenheit scales automatically?

[Ethan M. <eam...@gm...>]

Use the "set link" command (version 5 only)

gnuplot> set link y2 via (y-32)*5./9. inv (y*9./5.)+32
gnuplot> set ytics format "%.0f °F"
gnuplot> set y2tics format "%.0f °C"
gnuplot> set xrange [0:150]
gnuplot> plot x

Imperfect ascii plot using "set term dumb":

        +-------+-------+-------+------+-------+-------+-------+--**

 140 °F +                                                      *** +60 °C

        |                                                  *****   |

        |                                               ****       |

 120 °F +                                            ****          +50 °C

        |                                         ****             |

 100 °F +                                      ****                +40 °C

        |                                   ****                   |

        |                                ***                       +30 °C

  80 °F +                             ***                          +

        |                          ***                             +20 °C

  60 °F +                       ***                                +

        |                   ****                                   |

        |                ****                                      +10 °C

  40 °F +             ****                                         +

        |          ****                                            +0 °C

  20 °F +       ****                                               +

        |    ****                                                  +-10 °C

        | ***                                                      |

   0 °F **------+-------+-------+------+-------+-------+-------+---+

        0       20      40      60     80     100     120     140





On Wed, Jul 27, 2016 at 9:59 AM, Alan Corey <ala...@gm...> wrote:

> I can't find an example of where these are being set on the fly, and
> for outdoor temperatures this is necessary since the ranges won't be
> the same year-round.  After years of wishing I had a way to do this,
> I've now got an Oregon Scientific temperature sensor outdoors with the
> data being received by radio with an rtlsdr dongle, decoded by rtl_433
> and formatted into something Gnuplot can plot by some C I wrote.  See
>
> https://www.raspberrypi.org/forums/viewtopic.php?f=63&t=154451&p=1015069#p1015069
>
> The thermometer setup gives me data in Celsius which I can plot fine,
> but I'd like to add a Fahrenheit scale as Y2 at the right side.  Best
> would be to have Y2 with its own tickmarks based on (Y1 * 1.8) + 32,
> but it would be acceptable to have the same tickmarks with different
> labels.  I can see how I could manually generate all the tickmarks and
> labels within a program, but is there a way to do this in Gnuplot?
> I'm aiming at having all this run by a cron job and generate a daily
> plot.
>
> I think normally I'd want to set y2range based on what yrange gets set
> to automatically.  Or read the Y values at the tickmarks and calculate
> new ones for Y2.  I don't want to plot more data, I just want to
> calculate another set of tickmarks.  And it needs to change every day.
>
>
>   Alan Corey
> --
> Credit is the root of all evil.  - AB1JX
>
>
> ------------------------------------------------------------------------------
> What NetFlow Analyzer can do for you? Monitors network bandwidth and
> traffic
> patterns at an interface-level. Reveals which users, apps, and protocols
> are
> consuming the most bandwidth. Provides multi-vendor support for NetFlow,
> J-Flow, sFlow and other flows. Make informed decisions using capacity
> planning
> reports.http://sdm.link/zohodev2dev
> _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> Membership management via:
> https://lists.sourceforge.net/lists/listinfo/gnuplot-info
>

[Gnuplot-info] Having dual Celsius and Fahrenheit scales automatically?

[Alan C. <ala...@gm...>]

I can't find an example of where these are being set on the fly, and
for outdoor temperatures this is necessary since the ranges won't be
the same year-round.  After years of wishing I had a way to do this,
I've now got an Oregon Scientific temperature sensor outdoors with the
data being received by radio with an rtlsdr dongle, decoded by rtl_433
and formatted into something Gnuplot can plot by some C I wrote.  See
https://www.raspberrypi.org/forums/viewtopic.php?f=63&t=154451&p=1015069#p1015069

The thermometer setup gives me data in Celsius which I can plot fine,
but I'd like to add a Fahrenheit scale as Y2 at the right side.  Best
would be to have Y2 with its own tickmarks based on (Y1 * 1.8) + 32,
but it would be acceptable to have the same tickmarks with different
labels.  I can see how I could manually generate all the tickmarks and
labels within a program, but is there a way to do this in Gnuplot?
I'm aiming at having all this run by a cron job and generate a daily
plot.

I think normally I'd want to set y2range based on what yrange gets set
to automatically.  Or read the Y values at the tickmarks and calculate
new ones for Y2.  I don't want to plot more data, I just want to
calculate another set of tickmarks.  And it needs to change every day.


  Alan Corey
-- 
Credit is the root of all evil.  - AB1JX

Re: [Gnuplot-info] Help creating a percentage stacked bar graph

[theozh <th...@gm...>]

Horizontal bar graphs seem to be a recurrent question.
In contrast to vertical bar graphs and histograms, as far as I know
there is no direct command in gnuplot for creating horizontal ones. I do
not know why. It really would be convenient.

A somewhat cumbersome workaround does it with boxxyerrorbars.

If your file "Test.dat", e.g. looks like the following:

# Col1	Col2	Col3	Legend
5	7	12	ItemA
8 	8	14	ItemB
12	24	2	ItemC
3	2	7	ItemD


Make sure that in the following code there is no newline, except after '\'.
The first plot is in absolute values, the second plot in percentage
values. The legend is written as axis tics.


#### horizontal histogram
reset
set multiplot layout 2,1

plot 'Test.dat' \
    u (0):($0):(0):($1):($0-0.4):($0+0.4):ytic(4) axes x1y1 with
boxxyerrorbars fs solid lc rgb "red" notitle,\
    '' u (0):($0):($1):($1+$2):($0-0.4):($0+0.4):ytic(4) axes x1y1 with
boxxyerrorbars fs solid lc rgb "green" notitle,\
    '' u (0):($0):($1+$2):($1+$2+$3):($0-0.4):($0+0.4):ytic(4) axes x1y1
with boxxyerrorbars fs solid lc rgb "blue" notitle,\
	

plot 'Test.dat' \
    u (0):($0):(0):(($1)/($1+$2+$3)):($0-0.4):($0+0.4):ytic(4) axes x1y1
with boxxyerrorbars fs solid lc rgb "red" notitle,\
    '' u
(0):($0):(($1)/($1+$2+$3)):(($1+$2)/($1+$2+$3)):($0-0.4):($0+0.4):ytic(4)
axes x1y1 with boxxyerrorbars fs solid lc rgb "green" notitle,\
    '' u
(0):($0):(($1+$2)/($1+$2+$3)):(($1+$2+$3)/($1+$2+$3)):($0-0.4):($0+0.4):ytic(4)
axes x1y1 with boxxyerrorbars fs solid lc rgb "blue" notitle,\
	
unset multiplot

Re: [Gnuplot-info] How to close the plot window automatically ?

[theozh <th...@gm...>]

if your terminal is wxt with ID 0 and you type the following in the
gnuplot command line

if (1) { plot sin(x); pause 5; set term wxt 0 close }

it plots sin(x), waits for 5 seconds and then closes the plot window.
Is this what you are looking for?

[Gnuplot-info] How to close the plot window automatically ?

[Priyadarshi P. <dav...@gm...>]

How to close the plot window automatically after some time period set by me?

Re: [Gnuplot-info] Multi plot alignment

[Robert v. K. <bo...@en...>]

On 14/07/16 14:24, Marek W. Gutowski wrote:
> set format y "% 6.0f"

Many thanks Marek, I would not have though of this approach, it works 
pefectly.


Bob von Knobloch

[Gnuplot-info] Multi plot alignment

[Robert v. K. <bo...@en...>]

Hello,

I am plotting 2 graphs, above each other, using multiplot with a common 
xrange. Plotting is good but, as the Yaxes have different scales (top 
plot 0-10 and lower plot 0-1100) the plots are of different widths (due 
to the different text lengths of the Yics for upper and lower plots. Is 
there a way to align the xscales?

The xscale is date and time.


Many thanks,


Bob von Knobloch.

Re: [Gnuplot-info] Anomalies in Gnuplot (fwd)

[zowie <def...@bo...>]

It looks correctly plotted to me.  

Looking at your .dat file: you’re missing a point on 17-April-2016, then you’ve got two points 
18-April-2016 and 19-April-2016, both with systolic of 146 (correctly plotted on the Y axis 
which goes to 160). Those are correctly rendered as a short horizontal line segment between 
the two data.

After that you’ve got a stay in hospital.

So it looks like your “glitch” isn’t you running home from hospital, it’s you being sick for a day
and breaking the progression.

The gap before 18-April may appear larger because the missing point on 17-April loses you 
*two* line segments: 16-17 and 17-18.  

You might like to plot this chart with histeps or linespoints rather than with lines:  line
plots like this are not as useful for understanding point-by-point progressions of missing or
error-prone data.  For that type of analysis, “histeps” is a better choice: you still get the
eye-pleasing connectivity of “lines”, but a more rigorous and less counterintuitive treatment 
of sampled and missing intervals.

Note that the above does not describe a problem with gnuplot, but rather with the practice of 
drawing plots of noisy (and potentially missing) data by connecting sampled points with straight
lines.

> On Jul 13, 2016, at 5:15 PM, Dave Horsfall <da...@ho...> wrote:
> 
> Am I the only one with these (allegedly minor) problems?
> 
> -- 
> Dave Horsfall DTM (VK2KFU)  "Those who don't understand security will suffer."
> 
> ---------- Forwarded message ----------
> Date: Wed, 1 Jun 2016 10:08:29 +1000 (EST)
> From: Dave Horsfall <da...@ho...>
> To: Gnuplot Mailing List <gnu...@li...>
> Subject: Anomalies in Gnuplot
> 
> Gnuplot 5.0 pl 3, on MacOS 10.11.5
> 
> I'm seeing a couple of glitches in Gnuplot, and I cannot see where I 
> stuffed up.
> 
> Browse over to www.horsfall.org/gptest and view the files therein; they 
> are a cut-down version of a script I use to plot my BP etc (note that my 
> dates use the format d/m/y, as I am not American; later graphs use the 
> more conventional yy/mm/dd, which actually sorts).
> 
> The gap towards the end of April is where I spent a week in hospital, but 
> note the sole dot in the middle; there is no datum for it, and anyway I 
> don't recall dashing home midway through my unintended confinement :-)
> 
> What would be causing it?  It's driving me nuts.
> 
> The other glitch is the datum for 18/4/16 (146), which did not go off the 
> scale but should have (at the time my scale was 0:140, since a BP of over 
> 140 means Trouble), but that's not so important.
> 
> Help, anyone?
> 
> -- 
> Dave Horsfall DTM (VK2KFU)  "Those who don't understand security will suffer."
> 
> ------------------------------------------------------------------------------
> What NetFlow Analyzer can do for you? Monitors network bandwidth and traffic
> patterns at an interface-level. Reveals which users, apps, and protocols are 
> consuming the most bandwidth. Provides multi-vendor support for NetFlow, 
> J-Flow, sFlow and other flows. Make informed decisions using capacity planning
> reports.http://sdm.link/zohodev2dev
> _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> Membership management via: https://lists.sourceforge.net/lists/listinfo/gnuplot-info
> 

[Gnuplot-info] Anomalies in Gnuplot (fwd)

[Dave H. <da...@ho...>]

Am I the only one with these (allegedly minor) problems?

-- 
Dave Horsfall DTM (VK2KFU)  "Those who don't understand security will suffer."

---------- Forwarded message ----------
Date: Wed, 1 Jun 2016 10:08:29 +1000 (EST)
From: Dave Horsfall <da...@ho...>
To: Gnuplot Mailing List <gnu...@li...>
Subject: Anomalies in Gnuplot

Gnuplot 5.0 pl 3, on MacOS 10.11.5

I'm seeing a couple of glitches in Gnuplot, and I cannot see where I 
stuffed up.

Browse over to www.horsfall.org/gptest and view the files therein; they 
are a cut-down version of a script I use to plot my BP etc (note that my 
dates use the format d/m/y, as I am not American; later graphs use the 
more conventional yy/mm/dd, which actually sorts).

The gap towards the end of April is where I spent a week in hospital, but 
note the sole dot in the middle; there is no datum for it, and anyway I 
don't recall dashing home midway through my unintended confinement :-)

What would be causing it?  It's driving me nuts.

The other glitch is the datum for 18/4/16 (146), which did not go off the 
scale but should have (at the time my scale was 0:140, since a BP of over 
140 means Trouble), but that's not so important.

Help, anyone?

-- 
Dave Horsfall DTM (VK2KFU)  "Those who don't understand security will suffer."

Re: [Gnuplot-info] How to link logarithmic axes?

[theozh <th...@gm...>]

> linked axes are not compatible with log scale
> 
That's unfortunate.
I do not have enough insight into gnuplot why it should not be possible
for logscale.
However, for a simple relative "linked" shift of the log axes the
following should be a workaround using the gnuplot variables GPVAL_Y_MIN
and GPVAL_Y_MAX


##### "link" logarithmic axes
reset

Offset = 0.01  # Log y and y2 axes offset

set xrange [-12:2]

set logscale y
set ytics format "%g" nomirror
set grid ytics
set yrange [*:*]

set logscale y2
set y2tics format "%g" nomirror
set y2range [GPVAL_Y_MIN*Offset:GPVAL_Y_MAX*Offset]

plot 10**x axis x1y1, 10**x axis x1y2
replot        # obviously it needs replot to update the GPVAL values?!

Re: [Gnuplot-info] kdensity plot not filled with filledcurves y=0

[Grothausmann, R. Dr. <gro...@mh...>]

Just a note, that "above"/"below" seem not to get honoured for filled kdensity.

On 05/07/16 11:31, Grothausmann, Roman Dr. wrote:
> On 05/07/16 10:03, Grothausmann, Roman Dr. wrote:
>> On 04/07/16 20:17, Ethan Merritt wrote:
>> Understanding, that the work-around is not so suitable for SVGs as it changes
>> the way of plotting I couldn't resist taking a look into the code to find out
>> what is wrong in the first place. I came as far as getting the expected
>> behaviour if y= is specified when applying the changes given by the patch below.
>> However I'm not sufficiently confident in adjusting the code appropriately for
>> cases I'm not familiar with. My guess it that the if statement would need an
>> additional check for kdensity to call need_fill_border and plot_lines.
>
> As far as I understand: smooth_parameter = -1 by default. So possibly this could
> do the trick appropriately:
>
> --- /opt/compilation/gnuplot-5.0.3/src/graphics.c.orig	2016-07-05
> 09:49:40.886329514 +0200
> +++ /opt/compilation/gnuplot-5.0.3/src/graphics.c	2016-07-05 11:26:14.242687554
> +0200
> @@ -763,13 +763,14 @@
>    		plot_boxes(this_plot, Y_AXIS.term_zero);
>    		if (bar_layer == LAYER_FRONT)
>    		    plot_bars(this_plot);
>    		break;
>
>    	    case FILLEDCURVES:
> -		if (this_plot->filledcurves_options.closeto == FILLEDCURVES_ATY1
> +		if (this_plot->smooth_parameter < 0
> +		&&  this_plot->filledcurves_options.closeto == FILLEDCURVES_ATY1
>    		||  this_plot->filledcurves_options.closeto == FILLEDCURVES_ATY2
>    		||  this_plot->filledcurves_options.closeto == FILLEDCURVES_ATR
>    		||  this_plot->filledcurves_options.closeto == FILLEDCURVES_BETWEEN) {
>    		    plot_betweencurves(this_plot);
>    		} else {
>    		    plot_filledcurves(this_plot);
>
>
>

-- 
Dr. Roman Grothausmann

Tomographie und Digitale Bildverarbeitung
Tomography and Digital Image Analysis

Institut für Funktionelle und Angewandte Anatomie, OE 4120
Medizinische Hochschule Hannover
Carl-Neuberg-Str. 1
D-30625 Hannover

Tel. +49 511 532-2900

Re: [Gnuplot-info] How to link logarithmic axes?

[theozh <th...@gm...>]

> linked axes are not compatible with log scale
> 
That's unfortunate.
I do not have enough insight into gnuplot why it should not be possible
for logscale.
However, for a simple relative "linked" shift of the log axes the
following should be a workaround using the gnuplot variables GPVAL_Y_MIN
and GPVAL_Y_MAX


##### "link" logarithmic axes
reset

Offset = 0.01  # Log y and axes offset

set xrange [-12:2]

set logscale y
set ytics format "%g" nomirror
set grid ytics
set yrange [*:*]

set logscale y2
set y2tics format "%g" nomirror
set y2range [GPVAL_Y_MIN*Offset:GPVAL_Y_MAX*Offset]

plot 10**x axis x1y1, 10**x axis x1y2
replot        # obviously it needs replot to update the GPVAL values?!

Re: [Gnuplot-info] How to link logarithmic axes?

[Ethan M. <eam...@gm...>]

linked axes are not compatible with log scale

On Mon, Jul 11, 2016 at 12:56 PM, theozh <th...@gm...> wrote:
> Hello,
>
> while trying to link logarithmic axes, I get unexpected results.
> The y and y2 axis should be e.g. just shifted by two orders of magnitude.
>
> So, I thought that the following command would do it.
>
> set link y2 via y*0.01 inverse y*100
>
> However, that's not what I get. See copy&paste example below.
> Although, it seems to work on linear scale.
>
> How can I do it with logarithmic axes properly?
> I am missing anything?
> Thanks for any hints.
>
>
> # link logarithmic axes
> reset
> set multiplot layout 2,1
> set xrange [-12:2]
>
> set logscale y
> set ytics format "%g" nomirror
> set grid ytics
> set yrange [1e-12:0.01]
>
> set logscale y2
> set y2tics format "%g" nomirror
>
> set link y2 via y*0.01 inverse y*100
>
> plot 10**x axis x1y1, 10**x axis x1y2
> unset logscale y
> unset logscale y2
> replot
> unset multiplot
>
>
>
>
>
>
> ------------------------------------------------------------------------------
> Attend Shape: An AT&T Tech Expo July 15-16. Meet us at AT&T Park in San
> Francisco, CA to explore cutting-edge tech and listen to tech luminaries
> present their vision of the future. This family event has something for
> everyone, including kids. Get more information and register today.
> http://sdm.link/attshape
> _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> Membership management via: https://lists.sourceforge.net/lists/listinfo/gnuplot-info

[Gnuplot-info] How to link logarithmic axes?

[theozh <th...@gm...>]

Hello,

while trying to link logarithmic axes, I get unexpected results.
The y and y2 axis should be e.g. just shifted by two orders of magnitude.

So, I thought that the following command would do it.

set link y2 via y*0.01 inverse y*100

However, that's not what I get. See copy&paste example below.
Although, it seems to work on linear scale.

How can I do it with logarithmic axes properly?
I am missing anything?
Thanks for any hints.


# link logarithmic axes
reset
set multiplot layout 2,1
set xrange [-12:2]

set logscale y
set ytics format "%g" nomirror
set grid ytics
set yrange [1e-12:0.01]

set logscale y2
set y2tics format "%g" nomirror

set link y2 via y*0.01 inverse y*100

plot 10**x axis x1y1, 10**x axis x1y2
unset logscale y
unset logscale y2
replot
unset multiplot

Re: [Gnuplot-info] kdensity plot not filled with filledcurves y=0

[Ethan M. <eam...@gm...>]

That is almost correct.  The test should be against
(this_plot->plot_smooth) rather than smooth_parameter.
I had a different fix in mind, but this one is better.  Thanks.

On Tue, Jul 5, 2016 at 2:31 AM, Grothausmann, Roman Dr.
<gro...@mh...> wrote:
>
>
> On 05/07/16 10:03, Grothausmann, Roman Dr. wrote:
>> On 04/07/16 20:17, Ethan Merritt wrote:
>> Understanding, that the work-around is not so suitable for SVGs as it changes
>> the way of plotting I couldn't resist taking a look into the code to find out
>> what is wrong in the first place. I came as far as getting the expected
>> behaviour if y= is specified when applying the changes given by the patch below.
>> However I'm not sufficiently confident in adjusting the code appropriately for
>> cases I'm not familiar with. My guess it that the if statement would need an
>> additional check for kdensity to call need_fill_border and plot_lines.
>
> As far as I understand: smooth_parameter = -1 by default. So possibly this could
> do the trick appropriately:
>
> --- /opt/compilation/gnuplot-5.0.3/src/graphics.c.orig  2016-07-05
> 09:49:40.886329514 +0200
> +++ /opt/compilation/gnuplot-5.0.3/src/graphics.c       2016-07-05 11:26:14.242687554
> +0200
> @@ -763,13 +763,14 @@
>                 plot_boxes(this_plot, Y_AXIS.term_zero);
>                 if (bar_layer == LAYER_FRONT)
>                     plot_bars(this_plot);
>                 break;
>
>             case FILLEDCURVES:
> -               if (this_plot->filledcurves_options.closeto == FILLEDCURVES_ATY1
> +               if (this_plot->smooth_parameter < 0
> +               &&  this_plot->filledcurves_options.closeto == FILLEDCURVES_ATY1
>                 ||  this_plot->filledcurves_options.closeto == FILLEDCURVES_ATY2
>                 ||  this_plot->filledcurves_options.closeto == FILLEDCURVES_ATR
>                 ||  this_plot->filledcurves_options.closeto == FILLEDCURVES_BETWEEN) {
>                     plot_betweencurves(this_plot);
>                 } else {
>                     plot_filledcurves(this_plot);
>
>
>
> --
> Dr. Roman Grothausmann
>
> Tomographie und Digitale Bildverarbeitung
> Tomography and Digital Image Analysis
>
> Institut für Funktionelle und Angewandte Anatomie, OE 4120
> Medizinische Hochschule Hannover
> Carl-Neuberg-Str. 1
> D-30625 Hannover
>
> Tel. +49 511 532-2900
>
> ------------------------------------------------------------------------------
> Attend Shape: An AT&T Tech Expo July 15-16. Meet us at AT&T Park in San
> Francisco, CA to explore cutting-edge tech and listen to tech luminaries
> present their vision of the future. This family event has something for
> everyone, including kids. Get more information and register today.
> http://sdm.link/attshape
> _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> Membership management via: https://lists.sourceforge.net/lists/listinfo/gnuplot-info

Re: [Gnuplot-info] kdensity plot not filled with filledcurves y=0

[Grothausmann, R. Dr. <gro...@mh...>]

On 05/07/16 10:03, Grothausmann, Roman Dr. wrote:
> On 04/07/16 20:17, Ethan Merritt wrote:
> Understanding, that the work-around is not so suitable for SVGs as it changes
> the way of plotting I couldn't resist taking a look into the code to find out
> what is wrong in the first place. I came as far as getting the expected
> behaviour if y= is specified when applying the changes given by the patch below.
> However I'm not sufficiently confident in adjusting the code appropriately for
> cases I'm not familiar with. My guess it that the if statement would need an
> additional check for kdensity to call need_fill_border and plot_lines.

As far as I understand: smooth_parameter = -1 by default. So possibly this could 
do the trick appropriately:

--- /opt/compilation/gnuplot-5.0.3/src/graphics.c.orig	2016-07-05 
09:49:40.886329514 +0200
+++ /opt/compilation/gnuplot-5.0.3/src/graphics.c	2016-07-05 11:26:14.242687554 
+0200
@@ -763,13 +763,14 @@
  		plot_boxes(this_plot, Y_AXIS.term_zero);
  		if (bar_layer == LAYER_FRONT)
  		    plot_bars(this_plot);
  		break;

  	    case FILLEDCURVES:
-		if (this_plot->filledcurves_options.closeto == FILLEDCURVES_ATY1
+		if (this_plot->smooth_parameter < 0
+		&&  this_plot->filledcurves_options.closeto == FILLEDCURVES_ATY1
  		||  this_plot->filledcurves_options.closeto == FILLEDCURVES_ATY2
  		||  this_plot->filledcurves_options.closeto == FILLEDCURVES_ATR
  		||  this_plot->filledcurves_options.closeto == FILLEDCURVES_BETWEEN) {
  		    plot_betweencurves(this_plot);
  		} else {
  		    plot_filledcurves(this_plot);



-- 
Dr. Roman Grothausmann

Tomographie und Digitale Bildverarbeitung
Tomography and Digital Image Analysis

Institut für Funktionelle und Angewandte Anatomie, OE 4120
Medizinische Hochschule Hannover
Carl-Neuberg-Str. 1
D-30625 Hannover

Tel. +49 511 532-2900

Re: [Gnuplot-info] kdensity plot not filled with filledcurves y=0

[Grothausmann, R. Dr. <gro...@mh...>]

On 04/07/16 20:17, Ethan Merritt wrote:
> On Monday, 04 July 2016 11:08:52 AM Ethan A Merritt wrote:
>> On Monday, 04 July 2016 04:11:44 PM Grothausmann, Roman Dr. wrote:
>>> Any idea how to avoid this raggedness?
>>
>> Sorry, that is in part a consequence of the work-around.
>>
>> I think the issue is that you are using a partially transparent fill.
>> The intended behaviour (that isn't working) is to draw the entire curve
>> as one large semi-transparent area.  The workaround causes it to be
>> drawn instead as a series of adjacent semi-transparent quadrilaterals.
>> Whenever there is an imperfect join between two adjacent quadrilaterals
>> due to coordinate round-off, there can be a glitch in rendering.
>
> By the way:  the glitch is in the rendering program, not the svg file per se.
> So changing the scale slightly or using a different viewing program may
> reduce or exacerbate the problem.

Thanks Ethan for the feedback.
Understanding, that the work-around is not so suitable for SVGs as it changes 
the way of plotting I couldn't resist taking a look into the code to find out 
what is wrong in the first place. I came as far as getting the expected 
behaviour if y= is specified when applying the changes given by the patch below. 
However I'm not sufficiently confident in adjusting the code appropriately for 
cases I'm not familiar with. My guess it that the if statement would need an 
additional check for kdensity to call need_fill_border and plot_lines.
What do You think?

--- /opt/compilation/gnuplot-5.0.3/src/graphics.c.orig	2016-07-05 
09:49:40.886329514 +0200
+++ /opt/compilation/gnuplot-5.0.3/src/graphics.c	2016-07-05 09:50:22.897553791 
+0200
@@ -763,16 +763,13 @@
  		plot_boxes(this_plot, Y_AXIS.term_zero);
  		if (bar_layer == LAYER_FRONT)
  		    plot_bars(this_plot);
  		break;

  	    case FILLEDCURVES:
-		if (this_plot->filledcurves_options.closeto == FILLEDCURVES_ATY1
-		||  this_plot->filledcurves_options.closeto == FILLEDCURVES_ATY2
-		||  this_plot->filledcurves_options.closeto == FILLEDCURVES_ATR
-		||  this_plot->filledcurves_options.closeto == FILLEDCURVES_BETWEEN) {
+		if (0) {
  		    plot_betweencurves(this_plot);
  		} else {
  		    plot_filledcurves(this_plot);
  		    if (need_fill_border(&this_plot->fill_properties))
  			plot_lines(this_plot);
  		}

Re: [Gnuplot-info] kdensity plot not filled with filledcurves y=0

[Ethan M. <merritt@u.washington.edu>]

On Monday, 04 July 2016 11:08:52 AM Ethan A Merritt wrote:
> On Monday, 04 July 2016 04:11:44 PM Grothausmann, Roman Dr. wrote:
> > On 01/07/16 19:10, Ethan Merritt wrote:
> > > I do not know why the filledcurves style is failing in conjunction with
> > > smoothing, but a work-around is to plot  first to a datablock and then plot
> > > the smoothed curve in a separate step:
> > >
> > > set table $SMOOTHED
> > > plot $original_data using col:(weight) smooth kdensity bandwidth sigma
> > > unset table
> > > plot $SMOOTHED using 1:2 with filledcurves y=0
> > 
> > Many thanks Ethan for the quick reply and the work-around. I got it working 
> > (https://github.com/romangrothausmann/gnuplot/blob/5f513bf1cf81c1709ff450e2830c8c59b76d2c80/hist-ar%2Bkd.gp) 
> > 
> > but the resulting plot of the "filled table" is kind of "ragged" when saved to 
> > SVG. See attached compressed SVG for the test data created with:
> > gnuplot -e "datafile='test.dat'; outfile='test.svg'; bin=10; sigma=3.5; xmin=0; 
> > xmax=180; col=1; sep=','; xlabel='test'" hist-ar+kd.gp
> > 
> > Any idea how to avoid this raggedness?
> 
> Sorry, that is in part a consequence of the work-around.
> 
> I think the issue is that you are using a partially transparent fill.
> The intended behaviour (that isn't working) is to draw the entire curve
> as one large semi-transparent area.  The workaround causes it to be
> drawn instead as a series of adjacent semi-transparent quadrilaterals.
> Whenever there is an imperfect join between two adjacent quadrilaterals
> due to coordinate round-off, there can be a glitch in rendering.

By the way:  the glitch is in the rendering program, not the svg file per se.
So changing the scale slightly or using a different viewing program may
reduce or exacerbate the problem.

	Ethan


> 
> This glitch is more evident when the fill area is transparent, so one
> thing you could try is to make the fill style "solid" rather than 
> "transparent solid".
> 	
> Ethen

-- 
mail:   Biomolecular Structure Center,  K-428 Health Sciences Bldg
        MS 357742,   University of Washington, Seattle 98195-7742

Re: [Gnuplot-info] kdensity plot not filled with filledcurves y=0

[Ethan A M. <EAM...@gm...>]

On Monday, 04 July 2016 04:11:44 PM Grothausmann, Roman Dr. wrote:
> On 01/07/16 19:10, Ethan Merritt wrote:
> > I do not know why the filledcurves style is failing in conjunction with
> > smoothing, but a work-around is to plot  first to a datablock and then plot
> > the smoothed curve in a separate step:
> >
> > set table $SMOOTHED
> > plot $original_data using col:(weight) smooth kdensity bandwidth sigma
> > unset table
> > plot $SMOOTHED using 1:2 with filledcurves y=0
> 
> Many thanks Ethan for the quick reply and the work-around. I got it working 
> (https://github.com/romangrothausmann/gnuplot/blob/5f513bf1cf81c1709ff450e2830c8c59b76d2c80/hist-ar%2Bkd.gp) 
> 
> but the resulting plot of the "filled table" is kind of "ragged" when saved to 
> SVG. See attached compressed SVG for the test data created with:
> gnuplot -e "datafile='test.dat'; outfile='test.svg'; bin=10; sigma=3.5; xmin=0; 
> xmax=180; col=1; sep=','; xlabel='test'" hist-ar+kd.gp
> 
> Any idea how to avoid this raggedness?

Sorry, that is in part a consequence of the work-around.

I think the issue is that you are using a partially transparent fill.
The intended behaviour (that isn't working) is to draw the entire curve
as one large semi-transparent area.  The workaround causes it to be
drawn instead as a series of adjacent semi-transparent quadrilaterals.
Whenever there is an imperfect join between two adjacent quadrilaterals
due to coordinate round-off, there can be a glitch in rendering.

This glitch is more evident when the fill area is transparent, so one
thing you could try is to make the fill style "solid" rather than 
"transparent solid".
	
Ethen

Re: [Gnuplot-info] kdensity plot not filled with filledcurves y=0

[Grothausmann, R. Dr. <gro...@mh...>]

On 01/07/16 19:10, Ethan Merritt wrote:
> I do not know why the filledcurves style is failing in conjunction with
> smoothing, but a work-around is to plot  first to a datablock and then plot
> the smoothed curve in a separate step:
>
> set table $SMOOTHED
> plot $original_data using col:(weight) smooth kdensity bandwidth sigma
> unset table
> plot $SMOOTHED using 1:2 with filledcurves y=0

Many thanks Ethan for the quick reply and the work-around. I got it working 
(https://github.com/romangrothausmann/gnuplot/blob/5f513bf1cf81c1709ff450e2830c8c59b76d2c80/hist-ar%2Bkd.gp) 

but the resulting plot of the "filled table" is kind of "ragged" when saved to 
SVG. See attached compressed SVG for the test data created with:
gnuplot -e "datafile='test.dat'; outfile='test.svg'; bin=10; sigma=3.5; xmin=0; 
xmax=180; col=1; sep=','; xlabel='test'" hist-ar+kd.gp

Any idea how to avoid this raggedness?

> ------------------------------------------------------------------------------
> Attend Shape: An AT&T Tech Expo July 15-16. Meet us at AT&T Park in San
> Francisco, CA to explore cutting-edge tech and listen to tech luminaries
> present their vision of the future. This family event has something for
> everyone, including kids. Get more information and register today.
> http://sdm.link/attshape
> _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> Membership management via: https://lists.sourceforge.net/lists/listinfo/gnuplot-info
>

-- 
Dr. Roman Grothausmann

Tomographie und Digitale Bildverarbeitung
Tomography and Digital Image Analysis

Institut für Funktionelle und Angewandte Anatomie, OE 4120
Medizinische Hochschule Hannover
Carl-Neuberg-Str. 1
D-30625 Hannover

Tel. +49 511 532-2900