Support for complex numbers?

[VarLad]

Makie.jl doesn't support complex numbers for data, yet?
Is there a way to plot a 100 x 2 data of complex numbers?

[auberjine]

I'm not sure if plotting complex numbers actually makes sense, what one usually do is to plot either the real/imaginary parts, or the modulus/argument separately
What would you like to see when plotting complex numbers ?

[VarLad]

If I give a single vector of complex numbers as argument, I expect it to be plot on the complex plane(real part vs imaginary part) (with the axes getting displayed properly)
If I give more vectors/matrices as argument, I expect it to plot the modulus(behave like normally giving numbers)

THANKS FOR RESPONDING！

[auberjine]

I think that's why complex number are not supported directly : depending on the application people expect different results ^^

In your first case you can do
scatter(real(complex_array), imag(complex_array))

And in your second
heatmap(abs.(complex_matrix))

[VarLad]

Figured that out
Although I think its better if people could do that with scatter(z) or scatter(z1,z2).
More appealing for the package🙂
Thanks anyway！

[SimonDanisch]

Well, there isn't a problem, as long as the behavior depends on the type!

[VarLad]

why not use function overloading？
function plot(a::Complex)

end
somewhere in the Makie src or something like that
Works right？

[SimonDanisch]

Yeah, that's an option:
http://makie.juliaplots.org/dev/recipes.html

[Uroc327]

Is there more detailed documentation for convert_arguments available?
I had hoped I could implement this somehow using

convert_arguments(::Type{<:Scatter}, x::Complex) = (real(x), imag(x),)

but I wasn't able to get a call like scatter([1+5im, 2+6im, 3+7im, 4+8im]) to work.

[SimonDanisch]

Well, if you want to plot a vector of complex number, you need to overload the conversion for a vector of complex numbers.
This should work:

function convert_arguments(::Type{<:Scatter}, numbers::AbstractVector{<: Complex})
     return (map(x-> Point2f0(real(x), imag(x)), numbers),)
end

[Uroc327]

Thanks so much! I assumed this was recursive.

[Uroc327]

For further reference, I used

convert_arguments(::Type{<:Scatter}, coords::AbstractVector{<:Complex}) = (map(x -> Point2(float(real(x)), float(imag(x))), coords),)
convert_arguments(::Type{<:Scatter}, coords::AbstractVector{<:Complex}, heights::AbstractVector{<:Real}) = (map((x, z) -> Point3(float(real(x)), float(imag(x)), float(z)), coords, heights),)

[lambertbrady]

I second the request for built-in support for plotting complex numbers. I'm new to Julia, and decided to try it out specifically for its typing system, but have found it extremely difficult to use so far in practice. Ideally, I would be able to pass in an array of complex numbers and have them plotted in the complex plane. For example, I would like to use something like lines(zeta.(vals)), where zeta is the complex-valued Riemann Zeta function from the SpecialFunctions package. Instead, I've had to deconstruct the complex values every time: lines(real.(zeta.(vals)), imag.(zeta.(vals))). Essentially, a list of complex numbers should be interpreted the same way as lines(positions) shown in the documentation.

[SimonDanisch]

Maybe someone can turn those converts into a PR? :)

[VarLad]

@Uroc327 Can you make the PR？

[jonathanlilly]

I'd also like to second the vote for this. I use complex numbers often because they make bivariate data (e.g. horizontal velocity or displacement) quite simple. For complicated plots writing real(z),imag(z) many times becomes tedious. Naturally when you write a complex number you mean z = x + im y, so it's natural to expect that behavior.

... And if someone has the patience to explain to a Julia newbie what I need to do to take advantages of the changes Uroc327 made, I would greatly appreciate it.

[berjine]

Personally when I want to easily do a special plot I define a function to do it, this way it is easy afterwards to add custom ticks, labels, scalings, colors, etc

could be simply :

function zplot(data)
    lines(real.(data), imag.(data))
end

or with more formatting to prepare for a publication for example :

function zplot(data, fname::String)
    f = Figure(size=(800,600))
    ax = CairoMakie.Axis(f[1,1], title="Title", xlabel="Horizontal velocity", ylabel="Vertical velocity",
    xticks = LinearTicks(10), yticks = LinearTicks(10), xgridvisible = false, ygridvisible = false)
    lines!(ax, real.(data), imag.(data))
    save(fname*".svg", f)
    f
end