Ugly output when combining 2 plots with GraphicsRow

Question

Here are the data that I want to plot:

meandatf1 = Mean /@ datf1
meandatf2 = Mean /@ datf2

{1.48908, 1.49641, 1.49354, 1.50385, 1.49835, 1.49617, 1.50569, \
1.50117, 1.50226, 1.50151, 1.50108, 1.50031, 1.49955, 1.49721, \
1.49898, 1.50266, 1.50177, 1.50227, 1.49886, 1.50002}

{-0.00127783, 0.000556012, 0.0000143709, -0.000602328, -0.000375952, \
-0.00125357, 0.0000228143, 0.000175103, 0.000868018, -0.0003298, \
0.000230178, 0.000222689, -0.0000624273, -0.0000760139, -0.000263302, \
0.0000826082, 0.000206463, 0.0000507921, -0.0000955452, 0.000184107}

And here is how I generate my individual plots:

genPlot[c_, dat_, ymin_, ymax_, legend_] :=
 ListPlot[{
   Table[c, {Length[dat]}], dat},
   AxesLabel -> {"N", "\[Mu]"}, PlotRange -> {Automatic, {ymin, ymax}},
   PlotLegends -> Placed[legend, Above],
   Joined -> True, Mesh -> None, InterpolationOrder -> 0, 
   ImageSize -> 400] (* I tried also without ImageSize, see below *)

plot1 = genPlot[3/2, meandatf1,  1.45,  1.55,  "\[ScriptCapitalD]1"]
plot2 = genPlot[0,   meandatf2, -0.004, 0.004, "\[ScriptCapitalD]2"]

The output of the individual plots is fine:

But when I combine the two plots with GraphicsRow[] I get this ugly output:

When I don't specify ImageSize in genPlot[] the combined plot is very small (but otherwise looks fine):

If I set ImageSize in the GraphicsRow[] (and not in genPlot[]) the container image is increased instead of the contained cells (which is reasonable, but was a desperate long shot by my side):

It must be something trivial that I am missing. Any ideas?

EDIT1: I am using Mathematica 9.0.1 under Mac OSX Mavericks.

EDIT2: Pretty much all of the proposed solutions worked for me. I wish I could accept >1 but, since this is not possible, I'll accept the one from @JasonB for its beauty of simplicity.

Answer 1

GraphicsRow, GraphicsGrid, or GraphicsColumn can be useful, when you need the output to be a Graphics object. But it's often easier to just use the corresponding non-Graphics version; I find the resizing and spacing options easier to use.

When I do

Grid[{{plot1, plot2}}]

I get

which I think is what you want.

Answer 2

You can often get tolerable results from the likes of GraphicsColumn by specifying explicit values for ImagePadding, making sure you allow enough padding for frame or axis labels.

Here's a stacked plot using GraphicsColumn:

commonopts = {Mesh -> None, InterpolationOrder -> 0, PlotRange -> All,
    PlotStyle -> Thick, Frame -> True, Axes -> False, ImageSize -> 400, AspectRatio -> 0.4, 
   BaseStyle -> {FontFamily -> "Calibri", 12}};

a = ListLinePlot[{meandatf1, 1.5 + 0 meandatf1}, 
   FrameLabel -> {"N", "\[Mu]", Style["GraphicsColumn - not a total disaster", Bold, 14]}, 
   ImagePadding -> {{70, 20}, {0, 40}}, 
   Epilog -> Text["D1", Scaled[{0.95, 0.9}]], commonopts];

b = ListLinePlot[{meandatf2, 0 meandatf2}, 
   FrameLabel -> {"N", "\[Mu]"}, ImagePadding -> {{70, 20}, {40, 0}}, 
   Epilog -> Text["D2", Scaled[{0.95, 0.9}]], commonopts];

GraphicsColumn[{a, b}, Center, 0]

Answer 3

@DavidPark's answer is high quality, extremely well thought out and typical of the care he takes putting together graphics which are designed to display the data in their best light.

An alternative to his Presentations package is LevelScheme. The code below is one graphic that might be produced using this free package.

Needs["LevelScheme`LevelScheme`"]
Block[
 {},
 md1 = {1.48908, 1.49641, 1.49354, 1.50385, 1.49835, 1.49617, 1.50569,
    1.50117, 1.50226, 1.50151, 1.50108, 1.50031, 1.49955, 1.49721, 
   1.49898, 1.50266, 1.50177, 1.50227, 1.49886, 1.50002};

 md2 = {-0.00127783, 0.000556012, 
   0.0000143709, -0.000602328, -0.000375952, -0.00125357, 
   0.0000228143, 0.000175103, 0.000868018, -0.0003298, 0.000230178, 
   0.000222689, -0.0000624273, -0.0000760139, -0.000263302, 
   0.0000826082, 0.000206463, 0.0000507921, -0.0000955452, 
   0.000184107};
 {mnx, mxx} = {1, Length@md1};
 {ny1, xy1} = {Min@md1, Max@md1};
 {ny2, xy2} = {Min@md2, Max@md2};
 Figure[{
   SetOptions[SchemeObject, FontFamily -> "Helvetica", FontSize -> 10],
   ScaledLabel[{.5, .97}, "Comparison", FontSize -> 12, 
    FontWeight -> Bold, Offset -> {0, 0}],
   Multipanel[{2, 1},
    Margin -> 50,
    XPlotRanges -> {mnx, mxx},
    YPlotRanges -> {{1.4, 1.6}, {-.002, .002}},
    XFrameLabels -> {"N"}, BufferB -> 5,
    YFrameLabels -> {"\[Mu]", "\[Mu]"}, BufferL -> 10,
    XFrameTicks -> {LinTicks[1, 20, 2, 2]},
    YFrameTicks -> {
      LinTicks[1.4, 1.6, .05, 1],
      LinTicks[-.002, .002, .002, 1]},
    YGapSizes -> .1, XGapSizes -> .05,
    YPanelSizes -> {1, 1},
    XPanelSizes -> {1},
    First -> "A",
    Order -> Vertical
    ],
   FigurePanel[{1, 1}],
   RawGraphics@
    ListPlot[md1, InterpolationOrder -> 0, PlotStyle -> Black],
   ScaledLabel[{0.085, .86}, "\[ScriptCapitalD]1"],
   FigurePanel[{2, 1}],
   RawGraphics@
    ListPlot[md2, InterpolationOrder -> 0, PlotStyle -> Black],
   ScaledLabel[{0.085, .85}, "\[ScriptCapitalD]2"]
   },
  ImageSize -> 500
  ]
 ]

This produces the image

Answer 4

I am rather aghast at how poor those solutions are! Are the two sets of data suppose to be related? They have the same horizontal range so why not put them on top of each other and have only one x axis? Then it is really poor technique to use an Axis plot when the data is crossing the axis. The ticks and labels clobber the data! Look in a journal such as Science and see how many such Axes plots you can find! Then the "mean" line probably should not have been drawn with ListPlot. It is better to draw it separately and use Style to turn off Antialiasing to obtain a sharp line.

Also, I would not too quickly get into writing routines such as genPlot. In a case like this it is better to produce a graphic first by working with the individual elements and doing a lot of perfecting and then, if you are going to produce many such plots, write a routine.

In any case, here is a graphic I produced using the Presentations Application, which I sell for $50. It is an application designed to aid in the production of custom graphics, tables, dynamic presentations, and generally literate notebooks. It has a number of essays on using Mathematica and numerous examples. It is 38 MB in size and has 511 files.

Some of the Presentation routines used here are:

XTickLine and YTickLine for producing free standing scales.

YAxisBreak for producing a broken y axis.

Aliasing as a convenient methods for producing sharp horizontal and vertical lines.

TranslateOp and ScaleOp as convenient post-op commands for graphics transformations.

The ability to freely mix curves and graphics primitives, just drawing one thing after another.

The graphic might be further improved but I don't know enough about what the data represents.

In this approach I am simply drawing on a piece of paper and using scaling and translation to position the curves.

<< Presentations`
Draw2D[
 {YTickLine[{0, 4, -0.01}, {-0.0015, 0.0010}, {-0.001, 0.0010, 0.001},
    5],
  YAxisBreak[5, 2, 0.2, 0.2],
  YTickLine[{6, 10, -0.01}, {1.48, 1.51}, {1.48, 1.51, 0.01}, 5],
  XTickLine[{0, 20, -0.2}, {0, 20}, {0, 20, 5}, 5],
  (* Draw the second set *)
  {Aliasing@Draw[0, {x, 0, 20}],
     ListDraw[meandatf2,
      Joined -> True, Mesh -> None, PlotStyle -> Thick,
      InterpolationOrder -> 0]} // ScaleOp[{1, 1600}, {0, 0}] // 
   TranslateOp[{0, 2.4}],
  (* Draw the first set *)
  {Aliasing@Draw[3/2, {x, 0, 20}],
     ListDraw[meandatf1,
      Joined -> True, Mesh -> None, PlotStyle -> Thick,
      InterpolationOrder -> 0]} // ScaleOp[{1, 400/3}, {0, 0}] // 
   TranslateOp[{0, -191.33}],
  (* Add Labels *)
  Text["\[ScriptCapitalD]1", {13, 10}],
  Text["\[ScriptCapitalD]2", {15, 3.5}],
  Text["Sample Points", {10, -2}],
  Text[Style["Graphical Presentation Matters", 14, Bold], {10, 12}]},
 AspectRatio -> 1/2,
 Frame -> False, Axes -> False,
 PlotRange -> {{-3, 21}, {-3, 13}},
 BaseStyle -> {FontSize -> 12},
 ImageSize -> 400]

Answer 5

Another solution that worked for me (besides using Grid[] that @JasonB suggested) is to set ImageSize->Full in genPlot[]. According to the docs this means "fill out the enclosing region". Then, use ImageSize->600 or whatever value to GraphicsRow. As the container increases in size, the contained cells follow because they try to fill in all the region.

Answer 6

Aside from the variety of good formatting alternatives presented in the answers here, I suspect that the problem comes from the PlotLegends option, which causes the output to have the head Legended instead of Graphics.

Head /@ {plot1, plot2}
(* {Legended, Legended} *)

Further Legended has its own DisplayFunction that I suspect interferes with its display in GraphicsRow. GraphicsRow uses Inset to align graphics inside another Graphics. I have occasionally but rarely found that Inset does not play nice with some objects, but I've never been able to fully understand why.

If you use PlotLabel instead of PlotLegends, the plots can be shown in GraphicsRow. If you need PlotLegends, then adapt one of the other answers.

genPlot[c_, dat_, ymin_, ymax_, legend_] := 
 ListPlot[{Table[c, {Length[dat]}], dat}, AxesLabel -> {"N", "\[Mu]"},
   PlotRange -> {Automatic, {ymin, ymax}}, PlotLabel -> legend, Joined -> True, 
   Mesh -> None, InterpolationOrder -> 0, ImageSize -> 400]

plot1 = genPlot[3/2, meandatf1, 1.45, 1.55, "\[ScriptCapitalD]1"];
plot2 = genPlot[0, meandatf2, -0.004, 0.004, "\[ScriptCapitalD]2"];

GraphicsRow[{plot1, plot2}]

Answer 7

I believe the following is the most effective abd elegant solution to "ugly" or rasterized text in graphics. This also works for Plot3D ListPointPlot3D etc.

Note: Also this improves the output of MaTeX[] available here

ggrough = Grid[{{plot1, plot2}}];
Export[NotebookDirectory[] <> "rough.pdf", ggrough]
Export[NotebookDirectory[] <> "rough.svg", ggrough]

ggsmooth=Graphics[Inset[gg], ImageSize->{UpTo[900], UpTo[1100]}, AspectRatio -> Full]
Export[NotebookDirectory[] <> "smooth.pdf", ggsmooth]
Export[NotebookDirectory[] <> "smooth.svg", ggsmooth]

While there is a discernible difference in the MMA Notebook display, enlarging makes the distinction clearer. Note the change of font:

• Rough PDF and SVG: and
• Smooth PDF and SVG: and