Strategy question on producing publication plots using matplotlib

Main Forum Old-Forum - do not post here
1

Hello,

I’m producing my final publication plots at the moment like that:

  • Export a vtk file using

vtk = VTKOutput(ma = self.mesh,coefs=coefs,names=names, filename = directory + filename, subdivision=0, legacy=True) vtk.Do()

  • Import that .vtk file again using the vtk library

reader = vtk.vtkUnstructuredGridReader()
reader.SetFileName(filename)
reader.ReadAllVectorsOn()
reader.Update()

    data = reader.GetOutput()

    # a 2D matrix of size npoints x 3 
    coordinates = np.array(data.GetPoints().GetData())
    # print(coordinates.shape)
    # print(coordinates[0])

    connectivity = np.array(data.GetCells().GetData()).reshape(-1,4)
    # print(connectivity.shape)
    # print(connectivity[0])

    weights = data.GetPointData().GetArray("vec")
    npweights = np.array(weights)
  • Do the plotting using matplotlib

fig = plt.figure()
fig.set_size_inches(10.5, 10.5)
ax=fig.add_axes([0,0,1,1])
ax.set_title(label)

            tri = mtri.Triangulation(x,y,tri)

            ax.triplot(tri, alpha = 0.1)
            ax.scatter(x,y,s=0.5, color='black')
            
            tpc = ax.tripcolor(tri, w, alpha=0.5)
            fig.colorbar(tpc)
                        
            ax.set_aspect('equal')

This method works fine, but seems not super elegant. Are there any suggestions to improve it? Especially, how could I get the tri = mtri.Triangulation(x,y,tri) and the w directly from ngsolve fes and Coefficient- / GridFunction?

Thanks!