Plot figure wrong color in white in python

I/O
1

nx = ny = 20
mesh = UnitSquareMesh(nx, ny)

F = FunctionSpace(mesh, ‘P’, 1)
k = Function(F)

k_init = Expression(‘abs(x[0]-0.5)<l1 && x[1]<l2 ?1:0’, degree=2, l1=0.2,l2=0.4)
k.interpolate(k_init)
plt.colorbar(plot(k))

This snippet produce the following image:
download

a white region is shown, at which the value of function being plotted should be 1.
However, I also evaluate the value in that region, which is 1. Besides, if I save it as xdmf and read it by paraview, everything is fine.

The following is my import if that matters,

from future import print_function
from fenics import *
from dolfin import *
from mshr import *
import numpy as np
from tqdm import tqdm
import shutil
import matplotlib.pyplot as plt
%matplotlib inline

Besides, I am using docker installation of fenics,

DOLFIN_VERSION=2018.1.0.post1
PYPI_FENICS_VERSION=>=2018.1.0,<2018.2.0

Can you reproduce this result?
Any hint? What’s the suggested way in Fenics to do some simple visualization?

2

Hi, I can reproduce this error and suspect this is a bug. For workaround consider

from dolfin import *
import matplotlib.pyplot as plt
import matplotlib.tri as mtri

nx = ny = 20
mesh = UnitSquareMesh(nx, ny)

F = FunctionSpace(mesh, 'P', 1)
k = Function(F)

k_init = Expression('abs(x[0]-0.5)<l1 && x[1]<l2 ?1:0', degree=2, l1=0.2,l2=0.4)
k.interpolate(k_init)

x, y = mesh.coordinates().T
triangles = mesh.cells()
triang = mtri.Triangulation(x, y, triangles)

z = k.compute_vertex_values()

plt.tricontourf(triang, z)
1 Like
3

You can use vtkplotter:

from dolfin import *

nx = ny = 20
mesh = UnitSquareMesh(nx, ny)

F = FunctionSpace(mesh, 'P', 1)
k = Function(F)

k_init = Expression('abs(x[0]-0.5)<l1 && x[1]<l2 ?1:0', degree=2, l1=0.2,l2=0.4)
k.interpolate(k_init)

from vtkplotter.dolfin import plot
plot(k, style=1, lc='w')

image
image.png602×589 46 KB

1 Like
4

This seems great, but can it plot in jupyter notebook? I try, but it didn’tshow figure.

5

…I just tried and it seems to work for me:

you need to install k3d:
pip install k3d

image
image.png1122×810 113 KB

1 Like
6

Oh, this is great. I was using jupyter lab, which only show:

Plot(antialias=3, axes=[‘x’, ‘y’, ‘z’], background_color=16777215, camera=[0.5141421356237309, 0.5141421356237…

Thanks! this works in jupyter notebook.

1 Like
7

The original problem can easily be solved by changing the plot command to

plt.colorbar(plot(k, extend='max'))

These white regions frequently occur where functions are close to discontinuous. I assume that the large gradients will cause some kind of overshoots within the interpolation scheme. Interpolated function values that exceed the largest vertex values within a plot are not filled by default and thus remain white. The above keyword argument changes this behaviour.

8

Thanks, your code works. But the white region inside is continuous, discontinuity occurs when crossing the boundary.