Use of make-plots command
make-plots
The plots produced with rivet-mkhtml
are really rendered with the command make-plots
, which is called under the hood. The make-plots
command can also be used to create figures from the simple .dat text format produced by rivet-cmphistos
directly.
The make-plots
script is quite powerful, and includes several options to modify plotting style, adding curves or fits and more. For use with Rivet, the syntax documented in this document should be provided in the .plot file.
Internally, make-plots
the simple text format and converts them into PostScript or PDF files by creating a LaTeX file and running latex
, dvips
, and maybe ps2pdf
.
Usage
To run make-plots
call
make-plots [options] file.dat [file2.dat ...]
All available options can be listed by running
make-plots --help
Configuration files
make-plots
typically takes the plotting instructions and settings from the input ascii files as described in the “Input Format” chapter. It is also possible though to pass a global configuration file to make-plots
(cf. --help
) which allows to specify/overwrite settings for certain plots or histograms in a plot on top of what the input files specify. This could be useful if the ascii files are generated automatically (e.g. with rivet-mkhtml
or compare-histos
) and you still want to apply custom plotting options.
An example for this looks like:
# BEGIN PLOT figures/MC_WJETS/W_mass.dat
XMin=60.0
XMax=100.0
LegendXPos=0.65
# END PLOT
.*myLOrun.yoda/D0_2008_S7554427/d01-x01-y01::Scale=1.0
Here first the options in the PLOT
section of a specific ascii file are being amended/overwritten. The second part shows how to overwrite the Scale
property of one specific histogram line using the ID of the histogram.
Input Format
The ascii files which can be read by make-plots
are divided into sections. There are four types of sections which are called PLOT
, HISTOGRAM
, FUNCTION
, and SPECIAL
. Every file must contain exactly one PLOT
section and at least one section of the other three types. There may be multiple HISTOGRAM
, FUNCTION
, and SPECIAL
sections.
Empty lines and lines starting with #
are ignored, except for the section delimiters described below.
PLOT
The PLOT
section starts with
# BEGIN PLOT
and ends with
# END PLOT
Every file must have exactly one PLOT
section. In this section global parameters are specified, like the axis labels, the plot title, size, … An empty PLOT
section is perfectly legal, though.
In this section the following parameters can be set:
Titles, Labels
Title=<title>
The title of the plot.
XLabel=<label>
YLabel=<label>
ZLabel=<label>
Axis labels for the x-, y-, and z-axis.
XLabelSep=<distance>
YLabelSep=<distance>
ZLabelSep=<distance>
Distance between the axis label and the plot in units of \labelsep
.
XMajorTickMarks=<last_digit>
YMajorTickMarks=<last_digit>
ZMajorTickMarks=<last_digit>
XMinorTickMarks=<nticks>
YMinorTickMarks=<nticks>
ZMinorTickMarks=<nticks>
make-plots
tries to guess the distance between tickmarks automatically. If you are not satisfied with its result, you can override this by setting <last_digit>
to 1, 2, 5, or 10, and <nticks>
to the number of minor ticks you like. Note: These options are not available for logarithmic axes.
XTwosidedTicks=<0|1>
YTwosidedTicks=<0|1>
Draw tickmarks also on the upper and/or right side of the plot.
XCustomMajorTicks=<list>
YCustomMajorTicks=<list>
ZCustomMajorTicks=<list>
To specify major ticks at arbitrary positions and/or with arbitrary labels. <list>
is a whitespace-separated list of format value1 <spaces_or_tabs> label1 <spaces_or_tabs> value2 <spaces_or_tabs> label2 ...
.
[//]: # TODO: allow use of YAML-style list syntax to clarify delimiters?
XCustomMinorTicks=<list>
YCustomMinorTicks=<list>
ZCustomMinorTicks=<list>
To specify minor ticks at arbitrary positions. <list>
is a tab separated list of format value1 <tab> value2 <tab> value3 ...
.
PlotXTickLabels=<0|1>
RatioPlotTickLabels=<0|1>
Disable/enable plotting of the tick labels in the plot and ratio plot (useful if multiple plots are to be combined manually later).
Axes
LogX=<0|1>
LogY=<0|1>
LogZ=<0|1>
Use a logarithmic x-, y-, or z-axis. Default is linear.
XMin=<value>
XMax=<value>
YMin=<value>
YMax=<value>
ZMin=<value>
ZMax=<value>
FullRange=<0|1>
ShowZero=<0|1>
Specify the plot range. By default the range is chosen such that all data is visible in linear plots, and the zero is visible. ShowZero=0
suppresses plotting the zero in linear plots and thus zooms into the actual y-value range of the distribution. In logarithmic plots the automatic choice of YMin
is limited to be not smaller than 2e-4*YMax
, but manually you can specify any value. FullRange=1
also overrides the 2e-4*YMax
limit and plots the full range in y.
Normalization, Rebinning
NormalizeToIntegral=<1|0>
NormalizeToSum=<1|0>
Scale=<factor>
Normalize all histograms to their integral, to their sum of entries, or scale them by some arbitrary factor. Normalization and scale options in the PLOT
section override the corresponding option in the HISTOGRAM
section. The scale factor is applied after normalization.
Rebin=<nbins>
Rebin all histograms in this plot. Syntax and functionality is the same as for the Rebin option in the HISTOGRAM
section.
Sizes and Margins
PlotSize=<xsize,ysize>
Size in x and y direction of the plot. This can be specified in any unit LaTeX understands.
LeftMargin=<size>
RightMargin=<size>
TopMargin=<size>
BottomMargin=<size>
Distance between the plot and the paper edge.
FrameColor=<color>
Background color for the margin around the plot.
Legends
Legend=<0|1>
Display a legend in the plot.
CustomLegend=<text>
Custom text that is added to the legend.
LegendXPos=<pos>
LegendYPos=<pos>
Position of the legend within the plot. Anchor point is the top left corner of the legend, so units typically range between 0.0 and 1.0.
LegendAlign=<align>
Horizontal alignment of the legend: LegendAlign=l
is the default and will create a left-aligned legend, while LegendAlign=r
is right-aligned with the keys on the right hand side.
LegendOnly=<list>
Whitespace separated list of IDs. These can be histograms or functions. The legend is only shown for the listed objects. Without this option, all plotted objects which have a title enter the legend. The legend titles are plotted in the given order, so there are cases in which it makes sense to use LegendOnly
together with all histogram IDs. It is also possible to specify the legend order on an entry-by-entry basis using the LegendOrder=<int>
setting for each histogram or function.
Plotting Options
DrawOnly=<list>
Whitespace separated list of histogram IDs. Only the histograms in this list are plotted, even if there are more histograms defined in the file. The histograms are plotted in the given order, so there are cases in which it makes sense to use DrawOnly
together with all histogram IDs. This is especially useful for the Stack
option. It is also possible to specify the plotting order on a histogram-by-histogram basis using the PlotOrder=<int>
setting for each histogram.
Stack=<list>
Whitespace separated list of histogram IDs. The histograms will be added on top of each other. This is useful for example to compare data with background if the background has contributions from several histograms.
DrawSpecialFirst=<0|1>
DrawFunctionFirst=<0|1>
By default the SPECIAL
and FUNCTION
sections are plotted after the histograms. With these options you can override that behaviour.
ConnectGaps=<0|1>
If error bars are disabled and you want to bridge gaps in a histogram, you can set this parameter. By default it is off. Setting it in the PLOT
section affects all histograms, but you can also set it in the HISTOGRAM
section for individual histograms. The local setting overrides the global setting.
Comparison Plots
With the
RatioPlot=1
RatioPlotReference=<histogram_ID>
options you can create ratio plots for two or more histograms. Note that you must specify your reference data ID. This option is used by the compare-histos
script.
RatioPlotMode=<default|deviation|datamc>
By default, the ratio plot displays MC/Data. You can switch to (MC-data)/uncertainty (deviation
) or Data/MC (datamc
) with this option.
In ratio plots the following additional options are available and work in a similar way as their regular counterparts:
RatioPlotYLabel=<label>
RatioPlotYMin=<value>
RatioPlotYMax=<value>
RatioPlotYSize=<size>
RatioPlotErrorBandColor=<color>
By default, the reference data is plotted using a yellow error band around the central value of the ratio plot. If you would rather have it plotted in the same style as in the main plot (e.g. with black errorbars), you can specify:
RatioPlotSameStyle=1
If you only want the ratio plot without showing the actual data distribution, you can switch off the main plot. This option implies RatioPlot=1
:
MainPlot=0
Goodness of Fit
make-plots
can calculate the goodness of fit between histograms and display the result in the legend. It is also possible to change the color of the margin around the plot depending on the GoF. This is useful to provide a quick overview when looking at many plots.
GofType=chi2
The type of GoF. The default is chi2
and currently that’s the only option.
GofReference=<histogram_ID>
specifies the reference histogram to be used for the GoF calculation. If this option is omitted, the fallback is RatioPlotReference
.
The GoF calculation is activated by two options:
GofLegend=<0|1>
GofFrame=<histogram_ID>
GofLegend
calculates the GoF for all histograms and displays the results in the legend. With GofFrame
you can specify a single histogram for which the GoF result will be shown in the legend and used to assign a color to the plot margins. Note that FrameColor
overrides the color choice for the margin. You can use
GofFrameColor=<colorthresholds>
to specify the thresholds for the frame color. This option takes a list of <threshold>:<color>
pairs, separated by whitespace. The default is GofFrameColor=0:green 3:yellow 6:red!70
. Again, if you use FrameColor
, this option is disabled.
Color Palettes for 2-dim Plots
With the option ColorSeries
you can define a custom color palette for 2-dimensional plots. The syntax is the same as for the \definecolorseries
command in the xcolor
LaTeX package after the color series name, i.e. {core-model}{method}[begin-model]{begin-spec}[end-model]{end-spec}
. For more information you can consult the xcolor documentation. Here is an example:
ColorSeries={rgb}{last}[rgb]{1,0.97,0.94}[rgb]{0.6,0.0,0.05}
HISTOGRAM
The HISTOGRAM
section starts with
# BEGIN HISTOGRAM <ID>
and ends with
# END HISTOGRAM
There can be more than one HISTOGRAM
section in a file. Histograms are identified by <ID>
which can be any string not containing whitespace.
Data Format
Lines starting with a number (positive or negative) are interpreted as data. Each line specifies one bin. The fields in each line must be separated by tabs, not spaces (this needs to be fixes some day). For 1-dimensional histograms the format can be
<lowerbinedge> <upperbinedge> <value> <error>
<lowerbinedge> <upperbinedge> <value> <minuserror> <pluserror>
2-dimensional histograms are supported, too. They are plotted as colormap (errors are ignored) and specified as
<lowerxbinedge> <upperxbinedge> <lowerybinedge> <upperybinedge> <value> <error>
Titles
Title=<title>
Title of the histogram. This is used for the legend.
Linestyles
LineStyle=<style>
Any linestyle that is understood by the LaTeX pstricks package, e.g. solid
, dotted
, dashed
, none
, as well as a special dashdotted
(or dotdashed
) linestyle which does what you might expect.
LineColor=<color>
Color of the line. Default is black, but any color that pstricks understands can be used, including constructions like red!70!blue!20
(for mixing colors), {[rgb]{0.8,0,0.7}}
(for RGB-colors), {[wave]{580}}
(for wavelengths in nm), LineColor={[cmyk]{1,1,0,0}}
for CMYK-colors, or [hsb]{0.5,1,1}
for HSB-colors.
LineOpacity=<opacity>
Set the opacity of the line. Default is 1.0. This might not work for ps output.
LineWidth=<width>
Width of the line.
LineDash=<dashstyle>
If LineStyle
is set to dashed
, you can specify the dash style with this option. Anything that is understood by pstrick’s dash=...
option is valid. An example for a dash-dotted line is LineDash=3pt 3pt .8pt 3pt
. You can use LineStyle=dashdotted
or LineStyle=dotdashed
as an abbreviation for LineStyle=dashed
with LineDash=3pt 3pt .8pt 3pt
.
ConnectBins=<0|1>
Choose whether to connect adjacent bins’ horizontal lines together by a vertical line on the bin edge. This is enabled by default, but you may wish to disable it when plotting reference data with error bars and point markers.
ConnectGaps=<0|1>
If ConnectBins is enabled and you want to bridge gaps in a histogram, you can set this parameter. By default it is off. Setting it in the PLOT
section affects all histograms, but you can also set it in the HISTOGRAM
section for individual histograms. The local setting overrides the global setting.
SmoothLine=<0|1>
Draw a smooth curve rather than a histogram
Fillstyles
FillStyle=<style>
FillColor=<color>
To fill the area below a histogram, set FillStyle
and FillColor
to something pstricks understands. Examples for the style are solid
or vlines
. See LineColor
for examples of color definitions.
FillOpacity=<opacity>
Set the opacity of the solid fillcolor. Default is 1.0. This might not work for ps output.
HatchColor=<color>
The color of a hatch pattern used for filling the area below a histogram. This is used for example when you use vlines
as style.
Data Points
ErrorBars=<0|1>
Turn on error bars.
ErrorBands=<0|1>
ErrorBandColor=<color>
Turn on error bands and set their color (see LineColor
for a description of color definitions).
ErrorBandOpacity=<opacity>
Set the opacity of the error band. Default is 1.0. This might not work for ps output.
PolyMarker=<dotstyle>
The marker style of the points. Any dot style which is understood by pstricks is valid, e.g. *
, o
, triangle
, diamond
, …
DotSize=<size>
DotScale=<factor>
The size of the markers. With DotSize
you can specify the absolute size, e.g. in units of pt
, while DotScale
is a relative measure with respect to the default size.
Normalization, Rebinning
NormalizeToIntegral=<1|0>
NormalizeToSum=<1|0>
Scale=<factor>
Normalize the histogram to the integral, to the sum of entries, or scale it by some arbitrary factor. If normalization and a scale factor are given, the scale factor is applied after normalization. This is useful for stacking histograms when the ratios are known.
Rebin=<nbins>
ErrorType=<stat|env>
Rebin the histogram. Starting with the lowest bin <nbins>
bins are combined into a new bin. If the number of bins in the histogram is not a multiple of <nbins>
, the remaining bins at the upper histogram end are silently ignored (i.e. if the original histogram has 10 bins and <nbins>
is 3, the plotted histogram shows three bins combining the bins 1—9 of the original histogram). The treatment of the errors is determined by the given ErrorType: stat
(default) assumes the errors are of statistical nature and combines them in quadrature sum, while env
allows to treat errors as envelope of various uncertainty runs which are combined linearly.
FUNCTION
make-plots
can draw arbitrary functions. These functions are defined as python code sniplets which are evaluated by make-plots
. The code sniplet must come after all other options in a FUNCTION
section and are preceded by Code=
on a single line. An example FUNCTION
section might look like this:
# BEGIN FUNCTION f_cc
LineColor=red
Code=
p0=16.4
p1=1.25
p2=0.9832
from scipy.special import erf
x-=0.5
if x<=0:
return 0
else:
return .5*p2*(1.+erf( (x-p0)/sqrt(x*p1) ))
# END FUNCTION
Common Options with HISTOGRAM
The following options have the same meaning as in the HISTOGRAM
section:
Title=<title>
LineStyle=<style>
LineColor=<color>
LineWidth=<width>
LineDash=<dashstyle>
FillStyle=<style>
FillColor=<color>
HatchColor=<color>
Function Range
You can limit the plot range of functions by specifying
XMin=<value>
XMax=<value>
SPECIAL
The SPECIAL
sections are used to include any custom pstricks code. This is useful for drawing arrows and lines, put text at any position into the plot, etc. The default coordinate system is defined to be (0,0)
at the lower left and (1,1)
at the upper right corner of the plot. By putting the \physicscoor
command in front of a coordinate pair, these coordinates are interpreted not in the pstricks coordinate system, but in the physics coordinate system of the plot, which is useful e.g. for marking cut values in a plot. Similar \physicsxcoor
and \physicsycoor
commands exist which will only treat the x or y coordinate respectively as being in physics units.
Hint: If you want to clip your SPECIAL
code to the plot area, you can use
\psclip{\psframe[linewidth=0, linestyle=none](0,0)(1,1)}
...
\endpsclip
An example of a SPECIAL
section might look like this:
# BEGIN SPECIAL
\psclip{\psframe[linewidth=0, linestyle=none](0,0)(1,1)}
\psline[linewidth=1.2pt,linecolor=red]{<-}\physicscoor(2.83,2)\physicscoor(2.83,18)
\uput{4pt}[180]{0}\physicscoor(2.83,12){observed}
\psline[linewidth=0.8pt,linecolor=red,linestyle=dashed]\physicscoor( 3.17,0)\physicscoor( 3.17,28.14)
\psline[linewidth=0.8pt,linecolor=red,linestyle=dashed]\physicscoor(-3.59,0)\physicscoor(-3.59,28.14)
\endpsclip
# END SPECIAL