PlotLikelihoodScan

Plot a likelihood scan. This action supports plotting

• several 1D scan curves on the same canvas
• several 2D scan contours on the same canvas

Usage:

The following is an example for a single 1D curve.

+PlotLikelihoodScan {
  +HWWRun2GGF {
    <arrangement="$TQPATH/../SFramework/share/templates/PlotLikelihoodScanSignificance.txt">
    <outputFile="workspaces/run2-ggf-$(fitLabel)/likelihood.pdf">
    +asmGGF {
      <source = "LikelihoodScans/ScanLikelihood.asmGGF/Scan">
      <style.lineColor=kRed, style.title="#mu_{GGF}">
    }
    <style.title.xAxis = "#mu_{ggF} (profiled)">
    <style.showLegend = true>
  }
}

The following is an example for two 2D contours (asimov and observed) plotted together.

+PlotLikelihoodScan {
  +Run2GGF2j {
    <outputFile="./workspaces/run2-comb-$(fitLabel)/likelihood.pdf">
    +2Dasm {
      <source = "/HWWRun2/LikelihoodScans/ScanLikelihood.2Dasm/Scan">
      <style.lineColor=kBlue>
      <style.markerColor=kBlue>
      <style.markerStyle=34>
      <style.showPoints = true>
    }
    +2Dobs {
      <source = "/HWWRun2/LikelihoodScans/ScanLikelihood.2Dobs/Scan">
      <style.lineColor=kRed>
      <style.markerColor=kRed>
      <style.markerStyle=47>
      <style.showPoints = true>
    }
    <style.title.xAxis = "#mu_{ggH}">
    <style.title.yAxis = "#mu_{VBF}">
    <style.showLegend = false>
  }
}

The name of the output file can be determined with the outputFile tag globally. The curves are added as subfolders to the action. The source of each scan is identified with the source tag for each curve separately. The source can either point to a result of a ScanLikelihood action, or to the result of a ScanContour action. Some global options exist to control the styling:

• style.title.xAxis, style.title.yAxis: choose titles for the axes.
• style.showLegend, enable/disable showing the legend. Additionally, style options can be set for each curve separately:
• style.lineColor, style.lineStyle Control the color and style of the lines.
• style.showPoints Control whether points will be drawn to indicate the anchors used for interpolating the smooth curve.
• style.markerColor, style.markerStyle Control the color and style of the markers drawn with style.showPoints=true.
• style.title Set the title of the curve to be displayed in the legend (if drawn).
• showMinimum Draw a marker to indicate the minimum.
• lineOption Choose the draw option used for the line ("C" or "L")
• yTag instead of using the defalt y tag of the scan result, use some other tag to extract the result. The result can be scaled using the yScale option. One typical use case is to plot the non-sanitized, not minimum-subtracted ("raw") curve with the combination yTag="minNll", yScale=.5. Some specific options exist for 1D plots:
• outlierThreshold control at which threshold (above the minimum) outliers should be removed from the plot. Can be helpful to get rid of points from failed fits.
• plotWalk=x If given, in addition to the likelihood curve, an additional curve will be shown, indicating how far the individual best-fit-values of the points in the likelihood curve are away from each other -- showing the walking distance from one point to the next. As the total value of this is meaningless, the curve can be scaled freely, for which the value of this tag can be used. Some specific options exist for 2D contour plots:
• style.shading select whether contours should be progressively faded out (such that, e.g., the 2-sigma contour has a less strong color than the 1-sigma contour).
• contours.x=y Define a contour x at value y. If not given, the default values for the 1,2,3,4,5 sigma contours in 2D will be used.
• colormap Choose a color in which to overlay the area of the curve
• interpolate.x,interpolate.y By default, the graph triangulation will be used, which can sometimes lead to sub-optimal plotting results (wiggly lines). With this option, the scan will first be converted into a 2D histogram with a defined, regular grid before the contours are computed. This tag can be used to decide the number of bins in x and y direction to choose.
• binPrecision By default, the graph triangulation will be used, which can sometimes lead to sub-optimal plotting results (wiggly lines). With this option, the scan will first be converted into a 2D histogram with a defined, regular grid before the contours are computed. This tag is an integer defining the number of significant digits used to identify the grid lines (a useful choice for scans with less than 100 points is 2)
• When using the binPrecision tag, the resulting histogram can be auto-sanitized by filling empty bins (i.e., failed fits) from their neighbors. The tag binSmoothing defines the radius of the smoothing to be applied in number of bins (i.e., how large the neighborhood is from which values can be taken). Additionally, binSmoothingSources can be used to identify how many bins in a neighborhood need to be filled (at least) in order to proceed with the smoothing in that neighborhood.