Skip to content

j-fu/GridVisualize.jl

Repository files navigation

Build status

GridVisualize

Plotting companion module for ExtendableGrids.jl Provides plotting of grids, scalar piecewise linear functions and vectors (2D only) for various plotting backends on simplicial grids in one, two or three space dimensions. The main supported backends are CairoMakie, GLMakie, PyPlot, and PlutoVista. Plots is only partially supported (1D and 2D rectangular grids).

Disclaimer

The code is rather complex and many codepaths are hard to test. Please be prepared to help fixing things.

Sample usage:

Plotting a grid, a function or a vector field

gridplot(grid, Plotter=GLMakie)
scalarplot(grid, function,Plotter=GLMakie)
vectorplot(grid, vectorfunction,Plotter=GLMakie)
streamplot(grid, vectorfunction,Plotter=GLMakie)

This works for 1/2/3D grids and either a function represented by its values on the nodes of the grid, or a scalar function of 1, 2 or 3 variables, respectively. Vector and stream plots are currenly available for 2D only.

The grid argument should be an ExtendableGrid as defined by the ExtendableGrids.jl package. Instead of grid, the following arguments can be passed in order to describe a grid (which then is internally created on-the-fly):

  • An AbstractVector X specifying an 1D grid
  • AbstractVectors X,Y specifying a 2D grid
  • AbstractVectors X,Y,Z specifying a 3D grid
  • coord, cellnodes, where coord is a dim x nn matrix of point coordinates and cellnodes is as dim+1 x nc connectivity matrix of simplex node indices, describing a dim-dimensional simplex grid with nn nodes and nc simplices.

Plot appearance can be tweaked by a number of keyword arguments.

Plotters

Plotters can be e.g. Plots, PyPlot, GLMakie, CairoMakie PlutoVista - we pass the module exported by the package. Different plotters can be used at the same time.

Multiple plots in one plotting window

vis=GridVisualizer(Plotter=GLMakie, layout=(1,2))
gridplot!(vis[1,1],grid)
scalarplot!(vis[1,2],grid,function)
reveal(vis)

Transient plots

This using fast updating via observables for GLMakie and persistent divs with PlutoVista.

vis=GridVisualizer(Plotter=GLMakie)
for i=1:N
   function=calculate(i)
   scalarplot!(vis,grid,function)
   reveal(vis)
end

Movies

Currently, these can be recorded with GLMakie, CairoMakie and Plots backends both from the REPL and Pluto notebooks. MP4 files and gifs can be created. PyPlot possibly will follow.

This shows the animated graphic in the REPL (essentially the same as above) and creates an embedded video in a Pluto notebook:

vis=GridVisualizer(Plotter=Plots)
movie(vis) do vis
  for i=1:N
     function=calculate(i)
     scalarplot!(vis,grid,function)
     reveal(vis)
  end
end

To save to a file instead of showing in a notebooks or from the REPL , use:

vis=GridVisualizer(Plotter=CairoMakie)
movie(vis, file="video.mp4") do vis
  for i=1:N
     function=calculate(i)
     scalarplot!(vis,grid,function)
     reveal(vis)
  end
end

Setting a default plotter

Instead of specifying a Plotter in calls to GridVisualizer, gridplot or scalarplot, a default plotter can be set:

default_plotter!(PyPlot)
gridplot(grid)
scalarplot(grid, function)

or

default_plotter!(GLMakie)
vis=GridVisualizer(layout=(1,2))
gridplot!(vis[1,1],grid)
scalarplot!(vis[1,2],grid,function)

Switching off plotting

Just pass Plotter=nothing in the respective places, or set default_plotter!(nothing) and all plotting functions will do nothing.

Available plotting backends and functionality.

  • 'y': available
  • 'i': some level of interactive control
  • '(y)': availability only on rectangular grids
  • 'p': planned (no schedule though)
  • 'n': probably not, also in the future
PyPlot Makie PlutoVista Plots VTKView
scalarplot, 1D y y y,i y y
vectorplot, 1D y y y y y
gridplot, 1D y y y y
scalarplot, 2D y y y (y) y,i
vectorplot, 2D y y y y
streamplot, 2D y y p n
gridplot, 2D y y,i y (y) y,i
scalarplot, 3D y y,i y,i n y,i
gridplot, 3D y y,i y,i n y,i
vectorplot, 3D p p p n
streamplot, 3D p p n
movie n y n y

Sample output

vscode

Plotting into the plot pane of Visual Studio Code is working. Here, you can use CairoMakie or WGLMakie as backend. This works only with the mutating functions, i.e. you should use something like

vis=GridVisualizer(Plotter=WGLMakie)
gridplot!(vis,grid,clear=true,show=true)

Notebooks

Pluto

Plotting in Pluto notebooks for CairoMakie, PyPlot, Plots, GLMakie is working, WGLMakie probably works in combination with JSServe.

Plotting in Pluto notebooks can use PlutoVista.jl based on plotly.js (1D) and vtk.js (2/3D). See the example notebook: pluto, html.

Jupyter

Chances are that things work with Jupyter as well. Please ping me if you would like to volunteer with testing and possibly fixing this.