~~NOTRANS~~
{{indexmenu_n>700}}

====== Plotting ======

It's quite difficult - practically impossible - to perceive data from rows and columns with numerical data. This holds true for complex (scientific) data; but also for much simpler data, e.g. temperature records over time. Graphical plots allow fast perception. That makes plots very important - also for debugging while development.

===== Plot Types =====

There are many plot types, you can select from. See https://datavizcatalogue.com/

===== Terminal / Command Line =====

[[http://www.gnuplot.info/|Gnuplot]] is a portable command-line tool for creating plots.

Usually, comma seperated value (CSV) files are used as input, see [[devop:csv|comma separated values]].
But it's also possible to plot from binary data files.

===== C++ =====

Following plotting libraries might be used for creating (debug) plot files.

First some meta results, which already give a good overview:

  * https://gasparri.org/2019/12/06/best-plot-and-charting-c-packages-for-data-visualization/
  * https://www.researchgate.net/post/What_are_the_best_plot_and_charting_C_package_which_can_be_used_for_data_visualization
  * https://stackoverflow.com/questions/4283731/plotting-package-for-c

Here some (additional) links:

  * sciplot
    * https://sciplot.github.io/
    * https://github.com/sciplot/sciplot
  * Matplot++
    * https://alandefreitas.github.io/matplotplusplus/
    * https://github.com/alandefreitas/matplotplusplus
  * VTK
    * https://vtk.org/
    * https://gitlab.kitware.com/vtk/vtk
    * https://kitware.github.io/vtk-examples/site/Cxx/Plotting/LinePlot/
  * PLplot
    * https://plplot.sourceforge.net/
  * https://github.com/ABRG-Models/morphologica
  * https://github.com/thclark/cpplot
  * Matplotlib-Cpp - requires python
    * https://github.com/lava/matplotlib-cpp
  * Immediate Mode Plotting (implot)
    * https://github.com/epezent/implot
    * Dear ImGui: https://github.com/ocornut/imgui

===== Python =====

A python script/program allowing interactive commands is better suited for experimenting/development.

However, C++ developers might use python when compiling their code as python module with bindings, e.g. with [[https://pybind11.readthedocs.io/|pybind11]]. Interface functions might utilize [[https://numpy.org/|numpy]] or [[https://eigen.tuxfamily.org/|Eigen]], see https://pybind11.readthedocs.io/en/stable/advanced/cast/eigen.html

In python, there are several plot libraries available:

  * Matplotlib
    * https://matplotlib.org/
    * https://github.com/matplotlib/matplotlib
  * PyQtGraph - builds on Qt
    * https://www.pyqtgraph.org/
    * https://github.com/pyqtgraph/pyqtgraph
  * seaborn
    * https://seaborn.pydata.org/
    * https://github.com/mwaskom/seaborn

i really like colored heatmaps:

  * https://stackoverflow.com/questions/33282368/plotting-a-2d-heatmap
  * https://github.com/DingWB/PyComplexHeatmap
  * https://matplotlib.org/stable/gallery/images_contours_and_fields/image_annotated_heatmap.html
  * https://seaborn.pydata.org/examples/spreadsheet_heatmap.html

===== Colormaps =====

Colormaps are used for visualizing or mapping mostly linear or grayscale data. IMHO, ''jet'' is very appealing for linear or logarithmic data. matplotlibs' documentation shows many colormaps: https://matplotlib.org/stable/tutorials/colors/colormaps.html

From technical point of view, ''turbo'' is an improved alternative for ''jet'', see https://ai.googleblog.com/2019/08/turbo-improved-rainbow-colormap-for.html

Other colormaps like ''viridis'', ''plasma'' or ''inferno'' provide perceptually uniform sequential lightness.

Here an interesting video on this topic: https://www.youtube.com/watch?v=xAoljeRJ3lU

In SDR and signal processing, such colormaps are used to visualize the power (or energy) in one single pixel to allow [[https://en.wikipedia.org/wiki/Spectrogram|spectrogram]] plots, which already use x and y axes for time and frequency. For sure, there are usages in many other fields.

==== Cyclic colormaps ====

Circular data, e.g. phase values, should be treated special - as -180° equals +180°. //Cyclic// colormaps like ''twilight'' or ''hsv'' are for that case.

These colormaps are ideally viewed on circle like here: https://stackoverflow.com/questions/62531754/how-to-draw-a-hsv-color-wheel-using-matplotlib - and not on a ruler.

==== Other links ====

  * https://blog.habrador.com/2023/04/colormaps-overview-code-implementations-rainbow-virids.html
  * https://stackoverflow.com/questions/7706339/grayscale-to-red-green-blue-matlab-jet-color-scale