gmane.comp.tex.pgfplots

Pgfplots version 1.8 released

Christian Feuersaenger — 2013-04-04T17:39:27

Hello users of pgfplots,

I am glad to announce the release of version 1.8!

Note that the version on CTAN is version 1.8, even if it still displays 
version number 1.7 (probably some missing database update on CTAN).

Attaches you find the release notes for version 1.8.

Kind regards

Christian


1.8:
- new feature: tight bounding box even if the axis is no box and bb 
excludes clip path
- new feature: mesh/color input=explicit
- new feature: shader=interp now has drivers for both dvipdfmx and xetex
- new feature: support for more color spaces in colormap definitions
- new feature: shader=interp and device-level gray colorspaces
- new feature: 'contour/contour dir=[xyz]' to draw contours in different 
directions
- new feature: statistics library with boxplot handler (both boxplot 
prepared and automatic computation)
- fixed bug: 3d centered axis lines and label placement (requires 
compat=1.8 or higher)
- fixed bug: axis lines and placement of labels, tick scale labels, and 
reversed axes
     (requires com

Re: Scatter plot: using a discrete/sampledcolormap

Christian Feuersaenger — 2013-03-29T16:17:48

Hi Mike,

Your observation is correct: colorbar sampled simply changes the way the 
colorbar is displayed; it does not control the way colors are mapped 
into the colormap.

This is actually unsupported and would need to become a feature request.

What you can use is to employ the low-level interface for scatter plots 
to write such stuff on your own.

A related application can be found in the pgfplots manual (search for 
"Low-Level scatter plot interface Example"). Please refer to that 
section to see if that is something for you. For your convencience, here 
is a copy of the mentioned example (the manual contains explanations):

\begin{tikzpicture}
% Low-Level scatter plot interface Example:
% use three different marker classes
% 0% - 30%    : first class
% 30% - 60% : second class
% 60% - 100% : third class
\begin{axis}[
scatter/< at >pre marker code/.code={%
    \ifdim\pgfplotspointmetatransformed pt<300pt
       \def\markopts{mark=square*,fill=blue}%
    \else
       \ifdim\pgfplotspointmetatransformed pt<60

Scatter plot: using a discrete/sampled colormap

Mike Hettich — 2013-03-29T11:58:24

Dear all,

I hevaily use pgfplots for preparing graphs. I encountered a problem I 
was not able to solve yet. While it is possible to use the key "sampled" 
for the colorbar the respective data points in the plot are still 
linearly mapped onto the provided colormap. I have not found an option 
yet to force a discrete colormapping for the scatter plot, i.e. one 
color for a range of values instead of color interpolation. I hope you 
can provide some advice on how to solve this issue.


Best regards
Mike

------------------------------------------------------------------------------
Own the Future-Intel(R) Level Up Game Demo Contest 2013
Rise to greatness in Intel's independent game demo contest. Compete 
for recognition, cash, and the chance to get your game on Steam. 
$5K grand prize plus 10 genre and skill prizes. Submit your demo 
by 6/6/13. http://altfarm.mediaplex.com/ad/ck/12124-176961-30367-2

Re: automatically selecting somenon-overlappingtick labels

Neal H. Walfield — 2013-03-21T13:47:48

FWIW, I've attached the R code that I'm using to choose ticks and tick
labels.  The algorithms are sound, but the lack of context means that
I have to make assumptions about the text size and available space.

Examples:

 [1]   0  10  20  30  40  50  60  70  80  90 100
 [1]   1.0   1.8   3.0   6.0  10.0  18.0  32.0  56.0 100.0
 [1] "0"   "10"  "20"  "30"  "40"  "50"  "60"  "70"  "80"  "90"  "100"
 [1] "0"   ""    "20"  ""    "40"  ""    "60"  ""    "80"  ""    "100"

Perhaps this code will be useful to someone.

Neal


make.multiple.of <- function(x, multiple,
                             towards.zero=FALSE,
                             away.from.zero=FALSE,
                             towards.positive.infinity=FALSE,
                             towards.negative.infinity=FALSE) {
  if (sum(sapply(list(towards.zero,
                      away.from.zero,
                      towards.positive.infinity,
                      towards.negative.infinity),
                 function (x) identical(x, FALSE)))

automatically selecting some non-overlappingtick labels

Neal H. Walfield — 2013-03-11T14:08:47

Hi,

I wrote a program to generate plots.  It specifies the tick labels
that pgfplots should use.  This works great except when I have too
many ticks.  In this case, the labels overlap.  Is it possible to have
pgfplots select, say, the maximum number of nicely spaced tick labels
between which there is at least 1em of space?  Right now, I have a
hacky solution that selects at most 8 tick labels, but sometimes even
this is too many.

Thanks,

Neal

------------------------------------------------------------------------------
Symantec Endpoint Protection 12 positioned as A LEADER in The Forrester  
Wave(TM): Endpoint Security, Q1 2013 and "remains a good choice" in the  
endpoint security space. For insight on selecting the right partner to 
tackle endpoint security challenges, access the full report. 
http://p.sf.net/sfu/symantec-dev2dev

Re: x coord inv trafo/.code & logarithmic axes

Neal H. Walfield — 2013-01-19T13:03:24

Hi Christian,

Thank you for your help.

At Sat, 19 Jan 2013 12:05:50 +0100,
Christian Feuersaenger wrote:

So there is no way for the 'x coord inv trafo/.code' code to determine
that the argument is the log of the value and not the value?  That's
too bad.  Perhaps this is something that could be added to a future
version of pgfplots.


It does!  Thanks!

:) Neal

------------------------------------------------------------------------------
Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS,
MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current
with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft
MVPs and experts. SALE $99.99 this month only -- learn more at:
http://p.sf.net/sfu/learnmore_122912

Re: Unusual warning messages

Denis Bitouzé — 2013-01-19T11:22:43

Le samedi 19/01/13 à 12h10,
Christian Feuersaenger <cfeuersaenger-gM/Ye1E23mwN+BqQ9rBEUg< at >public.gmane.org> a écrit :


You're welcome! That's little work compared to pgfplots ;)

Re: Unusual warning messages

Christian Feuersaenger — 2013-01-19T11:10:08

Hi Denis,

I have applied your patch proposal. Thanks!

Best regards

Christian

Am 16.01.2013 15:17, schrieb Denis Bitouzé:


------------------------------------------------------------------------------
Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS,
MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current
with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft
MVPs and experts. SALE $99.99 this month only -- learn more at:
http://p.sf.net/sfu/learnmore_122912

Re: x coord inv trafo/.code & logarithmic axes

Christian Feuersaenger — 2013-01-19T11:05:50

Hi Neal,

The inverse coord transformations accept log values for logarithmic 
axes, that is correct.

In order to convert the number back to the correct number, you should use

       x coord inv trafo/.code={%
           \pgfplotscoordmath{x}{exp}{#1}%
         \show\pgfmathresult
         \expandafter\secondspp\expandafter{\pgfmathresult}}

the "coordmath" stuff ensures a suitable precision and respects the 
correct log basis. It assigns the result to \pgfmathresult.

Your \secondspp seems to overwrite \pgfmathresult, that's why I used the 
two \expandafter constructs to pass by value instead of by reference.


PGFplots assumes that user-provided transformations fit into the axis, 
i.e. that the transformation "knows" that it is to be applied in a log 
context. Consequently, you'd need two transformations, one tailored for 
linear axes and one tailored for log axes. This could be done by means 
of styles.

I hope this helps so far.

Best regards

Christian


Am 16.01.2013 14:10, schrieb Neal H. Walfield:

Unusual warning messages

Denis Bitouzé — 2013-01-16T14:17:14

Hi,

AFAIK, usual warning messages provided by packages start by:

Package <package name> Warning:

Those provided by pgfplots start with an extra exclamation mark
(followed by a space):

! Package pgfplots Warning:

Some TeX editors, for instance TeXstudio, parse the log file looking
for, I guess:

<beginning of the line>Package <something> Warning:

in order to filter warnings.

Hence, instead of (cf. line 33 of pgfplotscore.code.tex):

\def\pgfplots< at >warning#1{\pgfplots< at >message{! Package pgfplots Warning:
#1}}%

what about a regular:

\def\pgfplots< at >warning#1{\PackageWarning{pgfplots}{#1}{}}%

as that's already the case for pgfplots errors:

\def\pgfplots< at >error#1{\PackageError{pgfplots}{#1}{}}%

Best regards.

x coord inv trafo/.code & logarithmic axes

Neal H. Walfield — 2013-01-16T13:10:09

Hi,

I'm using x coord inv trafo/.code to transform seconds into more
(human) meaningful labels.  For instance, my macro turns 7000 seconds
into '2h'.  This works correctly for normal axes, however, if I use a
logarithmic axes, my macro is passed the exponent rather than the
value.  How do I programmatically determine what I am being passed?

A minimal example follows.

Thanks for your help,

Neal


\documentclass{article}

\usepackage{tikz}
\usepackage{pgfplots}

% number
% threshold
% divisor
% suffix
% else
\def\secondspphelper#1#2#3#4#5{%
  \pgfmathfloatparsenumber{#1}%
  \edef\num{\pgfmathresult}%
  \pgfmathfloatparsenumber{#2}%
  \edef\threshold{\pgfmathresult}%
  \pgfmathfloatparsenumber{#3}%
  \edef\divisor{\pgfmathresult}%
  %
  \pgfmathfloatlessthan{\num}{\threshold}%
  \ifdim\pgfmathresult pt = 1 pt%
    \pgfmathfloatdivide{\num}{\divisor}%
    \pgfmathfloattofixed{\pgfmathresult}%
    \pgfmathprintnumberto[/pgf/number format/precision=0]{\pgfmathresult}{\pgfmathresult}%
    \edef\pgfmathresult{\p

Re: pgfplots manual

Christian Feuersänger — 2013-01-13T17:06:17

Dear Dr. Zolili Ndlela,

this is a bug, probably in the viewers, see

http://tex.stackexchange.com/questions/79931/does-the-pgfplots-manual-crash-your-preview-and-or-skim-app-on-osx

I am unsure how to proceed here as I the success of acroread and all
linux viewers in displaying that stuff indicates that it (the
pgfplots.pdf) is correct.

And I cannot really experiment since I do not have the viewers at hand (?).

The solution for you would be to use acrobat reader to open the manual.

I would have an idea how to change some shading drivers such that the
pdf becomes simpler. Perhaps I can generate such a version and send it
so you, if you are willing to experiment.

Best regards

Christian

2013/1/11 Ndlela, Zolili <zndlela-AGnLgv6Ev6xGmzGZ86OUrw< at >public.gmane.org>:

------------------------------------------------------------------------------
Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS,
MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current
with LearnDevNow - 3,200 s

Re: pgfplots Frage zu Legendeneinträgen

Christian Feuersaenger — 2012-12-25T16:50:38

Dear Nils Schuhardt,

thanks for your request! Since it might be of interest to a larger 
audience, I cc'ed the pgfplots mailing list.

The question was "How can I use a different marker size for the legend, 
assuming that my plot has a 'mark size=0.01mm' set?" .

The answer is to rely on the style 'every legend image post' (or its 
equivalent short-cut key 'legend image post style').

Here is a suitable minimal example:

\documentclass{article}
\usepackage{pgfplots}

\begin{document}
\begin{tikzpicture}
     \begin{axis}
         \addplot+[
             mark=+,
             blue,
             only marks,
             mark size={0.1mm},
             legend image post style={mark size=1mm},
         ]
         {x};
         \addlegendentry{$x$}
     \end{axis}
\end{tikzpicture}
\end{document}

I wish a merry christmas and a happy new year!

Kind regards

Christian


Am 25.12.2012 16:27, schrieb Nils Schuhardt:


------------------------------------------------------------------------------
LogMeIn Rescue: Any

Re: lualatex and plotting file

Christian Feuersaenger — 2012-12-01T08:28:59

Hi Aurélien,

good that it works.

You can try out how much impact the sampling density has.

Best regards

Christian

Am 30.11.2012 23:56, schrieb aurelien coillet:


------------------------------------------------------------------------------
Keep yourself connected to Go Parallel: 
INSIGHTS What's next for parallel hardware, programming and related areas?
Interviews and blogs by thought leaders keep you ahead of the curve.
http://goparallel.sourceforge.net

Re: lualatex and plotting file

aurelien coillet — 2012-11-30T22:56:20

Hi Christian,

Thanks, that works indeed! I should have tried a little bit harder, I
suppose.

Regarding this huge number of points, it is actually needed: my data
consists in thousands of lines of various amplitude, and I want to be able
to distinguish one of these lines from its neighbour as well as comprehend
the global variation in amplitude. However, you're right that it may be a
pain for the user, and I will consider using an image instead. Actually,
that was my workaround, but I was not satisfied with this.

So thanks for your help and comments.

Best regards,

Aurélien


2012/11/30 Christian Feuersaenger <cfeuersaenger-gM/Ye1E23mwN+BqQ9rBEUg< at >public.gmane.org>

Re: lualatex and plotting file

Christian Feuersaenger — 2012-11-30T22:22:57

Hi Aurélien,

that is very strange, indeed. I can reproduce the problem with a file 
named "Sinus" (without extension): lualatex fails whereas pdflatex works.

Seems as if they have different path resolution algorithms.

However, it worked as soon as I introduced a file suffix: both lualatex 
and pdflatex find "Sinus.dat" .

My suggestion is to go along that path.

However, I would like to point out that 78000 points might be 
essentially overkill - and I am not talking about the time that pgfplots 
takes (be it lualatex or pdflatex). If you have 78000 points, you blow 
up the size of your .pdf file. In addition, displaying it will take some 
time. And: the user might not benefit at all unless you really have high 
frequencies! Are you sure that you need such a sampling density? Often, 
it is completely sufficient to have less samples for a .pdf document. 
Note that if you generate some .png graphics containing JUST the plot, 
you could use pgfplots to overlay an axis by means of its \addplot 
graphics feat

lualatex and plotting file

aurelien coillet — 2012-11-30T18:19:34

Hi,

I'd like to plot a large file (78000 points) with pgfplots, and it
obviously fails with pdflatex. I tried to use lualatex instead, but it does
not seem to find my file...
So I tried a very simple example :

\documentclass{article}
\usepackage[utf8]{luainputenc}
\usepackage{pgfplots}
\pgfplotsset{compat=1.7}

\begin{document}

\begin{tikzpicture}
  \begin{axis}
    \addplot file {Sinus};
  \end{axis}
\end{tikzpicture}

\end{document}

with the Sinus file (a small one, 100 points) starting with:

 0.00000000e+00 0.00000000e+00
 2.02020202e-02 2.02006461e-02
 4.04040404e-02 4.03930481e-02
 6.06060606e-02 6.05689655e-02
 8.08080808e-02 8.07201641e-02
 1.01010101e-01 1.00838420e-01

and it fails with:

! Package pgfplots Error: sorry, plot file{Sinus} could not be opened.

Note that compiling this document with pdflatex works as intended. What am
I doing wrong? I searched on google, but couldn't find anything...

Thanks for your help! Regards,

Re: Layout of bar plots

Christoph Bier — 2012-11-28T18:45:04

Christoph Bier schrieb am 25.11.2012 20:35:


And another helpful answer by Christian be found here:
http://tex.stackexchange.com/a/84520/10682

Re: Dates as input coordinates with seconds

Christian Feuersänger — 2012-11-28T13:09:12

Hi Denis,

before you are frustrated, you should take a look at the hint given by
Juernjakob - it seems to address exactly the requested use-case.

Best regards

Christian

2012/11/28 Denis Bitouzé <dbitouze-39ZsbGIQGT5GWvitb5QawA< at >public.gmane.org>:

------------------------------------------------------------------------------
Keep yourself connected to Go Parallel: 
INSIGHTS What's next for parallel hardware, programming and related areas?
Interviews and blogs by thought leaders keep you ahead of the curve.
http://goparallel.sourceforge.net

Re: Dates as input coordinates with seconds

Denis Bitouzé — 2012-11-28T08:25:03

Le mardi 27/11/12 à 21h48,
Christian Feuersaenger <cfeuersaenger-gM/Ye1E23mwN+BqQ9rBEUg< at >public.gmane.org> a écrit :


Hi Christian,


You're welcome :)


Too bad!


At least in educational area, that's the case.


Yes, could be nice.


OK.


Sigh...

BTW, wouldn't be possible to delegate date time data handling to
gnuplot? AFAICS, it is not so bad for this task: as said p. 19 of
http://www.gnuplot.info/docs_4.6/gnuplot.pdf, "Gnuplot now
tracks time to millisecond precision" ;)

Best regards.

Re: Dates as input coordinates with seconds

Christian Feuersaenger — 2012-11-27T20:48:35

Hi Denis,

thanks for your request.

The key question when we talk about plotting is: how large is the data 
range? The dateplot lib maps a complete date to some integer number. 
Unfortunately, the available precision is (appears to be) insufficient 
for seconds.

However, if your data range is one or two days, one could think about a 
different data transformation which "throws away" the month and year 
information and uses the full precision for the mapping.

This, however, would need to be implemented as coordinate transformation.

In other words: the answer to your question is: "Yes, if (a) your data 
range is small enough and (b) if someone implements it."

Best regards

Christian

Am 26.11.2012 15:46, schrieb Denis Bitouzé:


------------------------------------------------------------------------------
Keep yourself connected to Go Parallel: 
DESIGN Expert tips on starting your parallel project right.
http://goparallel.sourceforge.net