mpl-plotter 5.1.3a0

Publication-quality data representation library based on Matplotlib.

MPL Plotter

MPL Plotter is a Python plotting library built on top of Matplotlib with the goal of delivering publication-quality plots concisely. The full API documentation is available here. Read on to get started.

Table of Contents

1. Introduction

2. Install

3. Map of the library

4. Getting started

4.1 2D

4.2 3D

5. Comparisons and side by side plots

5.1 comparison

5.2 panes

6. Presets

6.1 Standard presets

6.2 Custom presets

7. Matplotlib

7.1 Retrieving axes, figures

7.2 Using Matplotlib's axis tiling

1. Introduction

Making plots for technical documents can be a time sink. MPL Plotter aims to reduce that overhead by allowing you to effortlessly and concisely

• Generate publication quality plots with a single call
• Plot curve comparisons
• Create figures with many plots

It is opinionated but built with flexibility in mind, which means that

• No default can't be changed
• Any and all further customization with Matplotlib is compatible. From ticks to legends to extra axes to whatever suits your needs

There's two ways to use MPL Plotter (plus any Matplotlib before or after):

• Calls to the 2D and 3D plotting classes
• Using presets, either those shipped with the library, or custom ones

It does the job for me and I expand it when it can't. Hope you find some use in it!

2. Install

pip install mpl_plotter

All dependencies will be checked for and installed automatically. They can be found in setup.py under install_requires.

Linux

PyQt5 may fail to install in Linux, prompting the following error:

FileNotFoundError: [Errno 2] No such file or directory: '/tmp/pip-build-4d8suz7p/PyQt5/setup.py'

To solve this, make sure pip is up to date and install PyQt5 5.14.0. Check this StackOverflow answer for further reference.

pip3 install --upgrade pip
pip3 install pyqt5==5.14.0

3. Map of the library

This is the map of the library for import reference.

module	class	function
module	class	function

• mpl_plotter
  • figure
  • get_available_fonts
  • markers
  • two_d
    • line
    • scatter
    • heatmap
    • quiver
    • streamline
    • fill_area
    • comparison
    • panes
  • three_d
    • line
    • scatter
    • surface
  • presets
    • publication
      • two_d
      • three_d
    • precision
      • two_d
      • three_d
    • preset
      • preset
      • two_d
      • three_d
  • color
    • schemes
      • colorscheme_one
    • functions
      • complementary
      • delta
    • maps
      • custom
      • mapstack

4. Getting started

In this section we'll go from the the most basic plot to a fairly customized version in 2 and 3 dimensions. The line demo scripts can be found in demo/scripts/line_demos/.

4.1 2D

For this example I'll use the 2D line class. Except for plot-specific arguments (line width etc. in this case), you can use the same inputs in this example with any of the other 2D plotting classes. Check the API reference for all general and specific arguments, or call help(<plotting class>) in your shell to access the docstrings.

As follows from the map above, the import to use the 2D line class is:

from mpl_plotter.two_d import line

And the following is the most basic MPL Plotter call, which will generate the image below (no input, and sin wave respectively).

line(show=True)	x = np.linspace(0, 2*np.pi, 100) y = np.sin(x) line(x=x, y=y, show=True)

Two important features are apparent:

1. MPL Plotter provides mock plots for every plotting class, so you can get straight into action and see what each does
2. MPL Plotter is somewhat "opinionated" and sets up quite a few parameters by default. This is based purely on my preference. You may not agree and you're more than welcome to play around with them!

---

Two more examples (results in the table below):

1. We can add some customization to make our line look a bit better:

    line(show=True, pad_demo=True, spines_removed=None)
   

   Our line has now some margins to breathe while the ticks are placed at the maximum and minimums of our curve, and no spines are removed.

2. Lastly, an example using some of the parameters you can change:

    line(norm=True, line_width=4,
     
         title="Custom Line", title_font="Pump Triline", title_size=40, title_color="orange",
   
         label_x="x", label_y="$\Psi$",
         label_size_x=30, label_size_y=20,
         label_pad_x=-0.05, label_pad_y=10,
         label_rotation_y=0,
   
         aspect=1,
         pad_demo=True,
         workspace_color="darkred",
         grid=True, grid_color="grey",
         
         tick_color="darkgrey", tick_decimals=4,
         tick_number_x=12, tick_number_y=12,
         tick_rotation_x=35,
   
         color_bar=True, cb_tick_number=5, cb_pad=0.05,
   
         show=True)
   

1	2

4.2 3D

Same applies in 3D.

	Examples

5. Curve comparisons and multiple pane plots

from mpl_plotter.two_d import comparison, panes

5.1 comparison

Plot any number of curves in a single plot. Axis limits will be set to the maximum and minimum of all your curves. No data will be left out, among other niceties.

As to inputs: inputs must match (2 xs and 3 ys won't work), BUT the following inputs are all valid:

x	y	result	notes
array	array	1
array	[array, array]	2	Both ys share x
[array, array]	[array, array]	2	Each y has an x
[n*[array]]	[n*[array]]	n	Each y has an x

As to using different plotting functions for different curves:

• You can specify a plotting function for each curve in the plot, a custom one for all curves, or not specify any (defaulting to lines). How? Read below (or check the code block below that). This is nice as it allows to concisely combine lines, scatter plots, and any other of the MPL Plotter plotting classes in a single.

As to any and all other arguments:

• Singular arguments: the regular MPL Plotter plotting class arguments. Apply to all curves in the plot.
• Plural arguments: pass a list of arguments, one for each curve. The result is as you'd imagine.

from mpl_plotter.two_d import comparison, line, scatter
        
def f(x, y, **kwargs):
    line(x, y,
         line_width=2,
         **kwargs)
def g(x, y, **kwargs):
    scatter(x, y,
            marker="D",
            point_size=10,
            **kwargs)
def h(x, y, **kwargs):
    scatter(x, y,
            marker="s",
            point_size=5,
            **kwargs)

comparison([x, x, x],
           [u, v, w],
           [f, g, h],
           plot_labels=["sin", "cos", "tan"],
           zorders=[1, 2, 3],
           colors=['C1', 'C2', 'C3'],
           alphas=[0.5, 0.5, 1],
           x_custom_tick_labels=[0, r"$\frac{\pi}{8}$", r"$\frac{\pi}{4}$"],
           show=show, backend=backend
           )

5.2 panes

The panes function allows for the plotting of a series of graphs in side-by-side panes. As to data input, the table below applies. It uses the comparison, function under the hood so the same input guidelines apply for all other inputs.

x	y	result	notes
array	array	11
array	[array, array]	12	Both ys share x
[n*[array]]	[n*[array]]	1n	Each y has an x
array	[array, array]	21	Both ys share x
[array, array]	[array, array]	21	Each y has an x
array	[n*[array], n*[array]]	2n	All curves in all (2) panes share a single x
[array, array]	[n*[array], n*[array]]	2n	All curves in each pane share an x
[n*[array], n*[array]]	[n*[array], n*[array]]	2n	All curves in all (2) panes have their own x
[n*[array], ... up to m]	[n*[array], ... up to m]	mn	All curves in all panes have their own x

Code

The following plots one curve per pane (3 in total):

panes(x,                   # Horizontal vector
      [u, v, y],           # List of curves to be plotted
      ["u", "v", "y"],     # List of vertical axis labels
      ["a", "b", "c"]      # List of legend labels 
      )

And the following plots an arbitrary number of curves per pane. As you can see, you just need to input n lists of m curves (where m=2 in the example below), and you will get a plot with n panes, with m curves in each.

    panes(x,                               # Horizontal vector
          [[u, uu], [v, vv], [y, yy]],     # List of pairs of curves to be compared
          ["u", "v", "y"],                 # List of vertical axis labels
          ["a", "b"]                       # List of legend labels
          )

Demo

Preposterous demonstration to illustrate the n panes, m curves concept. The code for these is available in tests/test_panes.py.

6. Presets

TL;DR: Take a parameter toml and forget about function inputs.

6.1 Standard presets

Standard presets are available to remove overhead. They're tailored for my use cases but you may find them useful anyway.

Publication

It is a common mistake to make a figure for a paper with unreadable labels. This preset tries to solve that, generating plots optimized to be printed on a small format, in side-by-side plots or embedded in a column of text.

from mpl_plotter.presets.precision import two_d
from mpl_plotter.color.schemes import one           # Custom colorscheme

x = np.linspace(0, 4, 1000)
y = np.exp(x)
z = abs(np.sin(x)*np.exp(x))

two_d.line(x, z, aspect=0.05, color=one()[-2], show=True)

Precision

Made to plot functions large on the screen, with equal x and y scales to avoid skewing the variables, and many ticks to visually inspect a signal.

from mpl_plotter.presets.precision import two_d

two_d.line(x, z, aspect=0.05, color=one()[-2], show=True)

6.2 Custom presets

Example workflow follows. For further reference check the preset tests.

1. Import the preset creation function

   from mpl_plotter.presets import preset
   

2. Create a preset, either from a plotter,

   from mpl_plotter.two_d import line
   
   _preset = preset(line)
   

   or from a dimension. In this case, the preset will contain all common parameters to all plots in 2 or 3 dimensions.

   _preset = preset(dim=2)
   

   The preset is a dictionary. You can edit its parameters as you would expect. However, it is more convenient to

3. Save your preset in a toml file. This will yield you a toml file containing all parameters for your plot or dimension, allowing you to easily inspect defaults and tailor settings to your liking. You may edit this file as you please, as long as you do not infringe on its syntax.

   _preset.save('tests/presets/test.toml')
   

4. Load the file in the same -or a different session.

   from mpl_plotter.presets import preset
   
   _preset = preset.load('tests/presets/test.toml')
   

5. Import an MPL Plotter preset plotter and load it with your preset

   from mpl_plotter.presets import two_d
   
   _two_d = two_d(preset=_preset)
   

6. Plot as you wish

   _two_d.line(show=True)
   _two_d.scatter(show=True)
   _two_d.<...>
   

7. Matplotlib

7.1 Retrieving axes, figures

The axis and figure on which each class draws are instance attributes. To retrieve them and continue modifications using standard Matplotlib:

from mpl_plotter.two_d import line

my_plot = line()
ax, fig = my_plot.ax, my_plot.fig

With the axis and figure, most Matplotlib functions out there can be used to further modify your plots.

7.2 Using Matplotlib's axis tiling

Matplotlib allows for subplot composition using subplot2grid. This can be used in combination with MPL Plotter:

Importantly:

• The auxiliary function figure (from mpl_plotter figure) sets up a figure in a chosen backend. This is convenient, as if the figure is created with plt.figure(), only the default non-interactive Matplotlib backend will be available, unless matplotlib.use(<backend>) is specified before importing pyplot.

from mpl_plotter import figure
from mpl_plotter.two_d import line, quiver, streamline, fill_area

backend = "Qt5Agg"  # None -> regular non-interactive matplotlib output

figure(figsize=(10, 10), backend=backend)

ax0 = plt.subplot2grid((2, 2), (0, 0), rowspan=1)
ax1 = plt.subplot2grid((2, 2), (1, 0), rowspan=1)
ax2 = plt.subplot2grid((2, 2), (0, 1), rowspan=1)
ax3 = plt.subplot2grid((2, 2), (1, 1), rowspan=1)

axes = [ax0, ax1, ax2, ax3]
plots = [line, quiver, streamline, fill_area]

for i in range(len(plots)):
    plots[i](ax=axes[i])

plt.show()

---

Back to top