plot_surface and facecolors

[mardukbp]

Apparently, the cmap keyword in plot_surface colors the surface in the Z direction. I needed to color the surface of a sphere with a scalar function. This is achieved with the facecolors keyword. In Matplotlib this is the solution:

norm = matplotlib.colors.Normalize()
ax.plot_surface(x, y, z,  rstride=1, cstride=1, facecolors=cm.jet(norm(colorfunction)))

However, in PyPlot.jl the ColorMap type does not have a method that takes an array of floats and returns the associated colors in the colormap. The workaround is

map = get_cmap("Blues")
map[:__call__](linspace(0,1,10))

Something similar happens with matplotlib.colors.Normalize. As a result, in PyPlot.jl the facecolors keyword must have the value

facecolors=map[:__call__](matplotlib.colors.Normalize()[:__call__](colorfunction))

I am new to Julia and don't have the skills (yet) to submit a pull request fixing this. Hopefully it won't be difficult to implement.

[stevengj]

I agree that using :__call__ is annoying. But this will go away in Julia 0.4, when call overloading should be available (JuliaLang/julia#8008). At that point, you will be able to call Python objects in Julia naturally, without resorting to :__call__ or pycall.

[mardukbp]

Good to know. Thank you.

[gozzilli]

Finding this saved my life. I spent hours trying to get the colormap to work, and couldn't figure out how to use it. Could this be added to the documentation, or to this recordnotfound page, which has the most amount of detail on colormaps, or to the README of this project? Thanks!

[stevengj]

@gozzilli, in Julia 0.4 the __call__ should no longer be needed, because call overloading works.

[IlyaOrson]

Hello,
I was trying to emulate this examples with julia 0.4.5 and matplotlib 1.5.1 and I couldn't avoid the use of __call__ to get them working.

For example, I used this:

number_of_lines = 5
viri = get_cmap("viridis")
cm_subsection = linspace(0.0, 1.0, number_of_lines)
colors = [viri[:__call__](x) for x in cm_subsection]
for col in colors
       plot(rand(10), color = col)
end

because a direct call throws an error:

colors = [viri(x) for x in cm_subsection]

ERROR: MethodError: call has no method matching call(::PyPlot.ColorMap, ::Float64)
Closest candidates are:
BoundsError()
BoundsError(::Any...)
DivideError()
...
in anonymous at no file

I get a similar error for the suggestion made here.
Am I missing something?
Thanks

[stevengj]

I can't reproduce:

julia> using PyPlot; c = get_cmap("RdBu")
ColorMap "RdBu"

julia> c(0.5)
(0.9657054999295402,0.9672433698878569,0.968089198364931,1.0)

julia> [c(x) for x in linspace(0,1,10)]
10-element Array{Any,1}:
 (0.40392157435417175,0.0,0.12156862765550613,1.0)               
 (0.711880063309389,0.12179931034060085,0.1816993519371631,1.0)  
 (0.8622837452327504,0.4295271150037354,0.3427143558567646,1.0)  
 (0.9686274528503418,0.7176470756530762,0.6000000238418579,1.0)  
 (0.9820069214876961,0.9061899325426888,0.861591703751508,1.0)   
 (0.8838908181470984,0.928489046938279,0.9530180727734285,1.0)   
 (0.6549019813537598,0.814379096031189,0.8941176533699036,1.0)   
 (0.3234909813778073,0.6149173624375287,0.7854671408148373,1.0)  
 (0.14248366800009035,0.4173010459133223,0.6833525755826164,1.0) 
 (0.019607843831181526,0.1882352977991104,0.3803921639919281,1.0)

[stevengj]

Maybe you need a Pkg.update()?

[IlyaOrson]

Pkg.checkout("PyPlot") did the trick. Thanks!