CODEWITHSUNDEEP
julia
Moana Lioni
Moana Lioni

Reputation: 45

Julia - Wanting to make subplot animations faster

I saw in Animating subplots using Plots.jl efficiently that there were solutions to make the animation go faster, using the same methods as function 3 and 4 of the answer

...
p[1][1][:z] = data
p[2][1][:z] = data
...

but as here I do not want the shared z-axis, I do not know how to adapt the code. I searched for solutions in the documentation of subplots, but found nothing that worked.

The code that works but which is very slow is

p1 = plot3d(1, xlim = (-5, 5), ylim = (-5, 5), zlim = (-5, 5), markersize = 2, seriestype = :scatter, legend = false, aspect_ratio = 1)
p2 = histogram(randn(1), bins = 50, xlims = (0,7), ylims = (0, 20), legend = false)
p = plot(p1, p2, layout = (1,2), title = "Cluster Simulation")

anim = @animate for iter=1:Ncalc
    step!(sim)

    scatter!(p1, sim.pos[:,1], sim.pos[:,2], sim.pos[:,3], xticks = false, yticks = false, zticks = false)

    histogram!(p2, sim.rad[:], xticks = false, yticks = false)

    next!(prog)
end every Int(Ncalc/Nframes)

Here is the code which does not work:

p1 = plot3d(1, xlim = (-5, 5), ylim = (-5, 5), zlim = (-5, 5), markersize = 2, seriestype = :scatter, legend = false, aspect_ratio = 1)
p2 = histogram(randn(1), bins = 50, xlims = (0,7), ylims = (0, 20), legend = false)
p = plot(p1, p2, layout = (1,2), title = "Cluster Simulation")

anim = @animate for iter=1:Ncalc
    step!(sim)

    p[1] = sim.pos[:,1], sim.pos[:,2], sim.pos[:,3]

    p[2] = sim.rad[:]

    next!(prog)
end every Int(Ncalc/Nframes)

The error I get is the following:

ERROR: MethodError: no method matching setindex!(::Plots.Plot{Plots.GRBackend}, ::Array{Float64,1}, ::Int64)

Where did I miss something?

Thanks in advance

Upvotes: 2

Views: 244

Answers (1)

Benoit Pasquier
Benoit Pasquier

Reputation: 3015

The error is telling you that you can't set the value of p[1] to a vector of Float64s. That's because p[1] is a plot object, not just the (x,y,z) data tuple. You need to assign the data to the :x, :y, and :z fields of the first (only) data series. I can't test this without your data, but I think you want to do

p[1][1][:x] = sim.pos[:,1] # p[1][1] == p1[1] is the series
p[1][1][:y] = sim.pos[:,2] # so p[1][1][:x] is the x data, and so on...
p[1][1][:z] = sim.pos[:,3]

for p1.

For p2, you'd have to do more work to get the histogram's data and update the bars "manually" (via StatsBase.jl I think). Not sure how to best do that, but maybe the other part speeds up your animation creation enough already?

Upvotes: 1

Related Questions