[ANN]: Gnuplot v1.1.0 - A Julia interface to gnuplot

[lyonsquark]

Thanks! That works great!

[tbeason]

Maybe this sort of goes against the whole idea but have you considered creating a second package, maybe GnuplotSimple.jl, that wraps the necessary functionality to make basic plotting easier? I am starting to get the hang of all of this, and I’m sure there are shortcuts I don’t yet know about, but it does seem a bit excessive in terms of LOC to make even the easiest of line plots.

Something like

function gnuplot(x,Y; title="", xlab="", ylab="", labels="", grid=true, xr=nothing, yr=nothing, save=false)
# all the @gp stuff
# save commands if asked
end

for the really basic functionality?

Anyways, this is likely my main plotting package for now.

[gcalderone]

One of the main driving idea of Gnuplot.jl is indeed to save typing as much as possible, and you can already pass x, y and many of the keywords you mentioned directly to @gp, e.g.:

@gp x y
@gp x y "w l"
@gp xlab="X" x y
@gp xr=[1,5] x y

I’m afraid that trying to be even more concise, may only result in a loss of generality…

In your case, no keyword is foreseen for terminal options and output file names, hence I guess you want to hardcode them in the function body to save typing. This is perfectly fine, but this would hardly be a general solution…

[tbeason]

I understand. So maybe you could provide me with a hint then. Suppose I am plotting 3 lines, each against the same x points, and I want to assign a label to each line. Let’s also say I want to plot from [0,1] only, and label each axis and the plot itself.

Do I need to write out

# setup
x = 0:0.05:2
y = x.^(1:3)'
# I won't count these lines against you :)

@gp "set grid"
@gp :- tit="The Example" xlab="My x" ylab="The y value" xr=[0,1]
@gp :- x y[:,1] "w l tit 'y1'"
@gp :- x y[:,2] "w l tit 'y2'"
@gp :- x y[:,3] "w l tit 'y3'"

or is there a shorter way?

Also, to be clear, my suggestion dealt with cases like this example. Certainly, you could not wrap all functionality in a plot function like that. But it is not often that you need to rapidly build plots that use all of the sophisticated features.

[gcalderone]

Yes, a shorter way (but hopefully still readable…) is:

@gp "set grid" ti="The Example" xla="My x" yla="The y value" xr=[0,1]
@gp :- x y[:,1] "w l t 'y1'" x y[:,2] "w l t 'y2'" x y[:,3] "w l t 'y3'"

Also, if you’re simply exploring data and don’t care about aesthetic details, you may simply type:

@gp x y[:,1] "w l" x y[:,2] "w l" x y[:,3] "w l"

[tbeason]

Forgive me if I don’t find that vastly better than what I wrote

I think I’ll probably just write it up myself

[gcalderone]

Yes, of course!

Just consider that I don’t think you could get anything shorter with other plotting packages…

[tbeason]

I started it here: GitHub - tbeason/GnuplotSimple.jl: A simple plotting recipe for using Gnuplot.jl

Have not done extensive testing, but it does work well enough for the stuff I’ve tried so far.

I stumbled upon your recipe example and thought it would work well if I built it off of that, so I did.

[gcalderone]

OK, now I got it! You want shortcut to plot the columns of the Y matrix against the x values.

Here’s how I would do it:

using Gnuplot

# Define an explicit recipe for this particular problem
function multiY(x::AbstractVector{Float64}, Y::AbstractMatrix{Float64}, labels=Vector{String})
    out = Vector{Gnuplot.PlotElement}()
    for i in 1:size(Y)[2]
        plotspec = "w l t '$(labels[i])'"
        push!(out, Gnuplot.PlotElement(data=Gnuplot.Dataset([x, Y[:, i]]), plot=plotspec))
    end
    return out
end

x = 0:0.05:2
y = x.^(1:3)'

@gp "set grid" tit="The Example" xlab="My x" ylab="The y value" xr=[0,1]
@gp :- multiY(x, y, ["y1", "y2", "y3"])

[tbeason]

Yea! That is pretty much what I did, but I just fancied it up a bit, so that I can include everything in that single call. Feel free to steal it and improve it lol

[lazarusA]

Hello,
I did some recipes and uploaded some examples at Plotting Market Data from AlphaVantage.jl with Gnuplot.jl - Nextjournal. Take a look.

[Alois]

Dear gcalderone,
I am struggling with gnuplot.jl and dates. Copying your example code throws:
ERROR: LoadError: AssertionError: Invalid argument at position 1.
Are there any changes I am not aware of? (I am on windows, if this matters at all).
Thank you in advance

[lazarusA]

which example? Could you copy paste here? or a link…

[oheil]

He means the example of @gcalderone which he replied to:

using Gnuplot, Dates

dates = [ Date(2020, 4, floor.(Int, x .* 30 .+ 1)) for x in rand(10) ]
val=rand(10)

@gp "set xdata time" "set timefmt '\"%Y-%m-%d\"'" :-
@gp :- "set format x '%Y-%m-%d'" "set xtics rotate by -45" :-
@gp :- "set rmargin 7" :-
@gp :- string.(dates) val "u 1:2 w p notitl"

I can confirm, i get the same error.

The @gp makro here doesn’t accept an array of strings. This is the code, which leads to the error:

       d = args[iarg]

        ok = false
        if typeof(d) <: Number
            ok = true
        elseif typeof(d) <: AbstractArray
            if typeof(d[1]) <: Number
                ok = true
            end
            if typeof(d[1]) <: ColorTypes.RGB
                ok = true
            end
        end
        if ndims(d) > maxDim
            maxDim = ndims(d)
        end

        @assert ok "Invalid argument at position $iarg"

in src\Gnuplot.jl line 294.

[Alois]

Yes, this example.
It would be a smart idea to add a working example involving dates to your list of examples here, https://lazarusa.github.io/gnuplot-examples/,
which is a perfect reference for gp.

[oheil]

In the mean time I was trying to produce a plot (Windows 10, J v1.5.3, GP v1.0.0) but it doesn’t work.

I stop now because after reading
https://gcalderone.github.io/Gnuplot.jl/v1.3.0/options/#Display-options-1

and finding that Gnuplot.options doesn’t exist, I give up.

There is either something trivial missing, which one needs to know, or there is a major issue with v1.0.0 and Windows. Let’s see if @gcalderone can give us a hint.

[PetrKryslUCSD]

Works with

using Gnuplot, Dates

dates = [ Date(2020, 4, floor.(Int, x .* 30 .+ 1)) for x in rand(10) ]
val=rand(10)

@gp "set xdata time" :-
@gp "set timefmt '\"%Y-%m-%d\"'" :-
@gp :- "set format x '%Y-%m-%d'" "set xtics rotate by -45" :-
@gp :- "set rmargin 7" :-
@gp :- string.(dates) val "u 1:2 w p notitl"

Julia 1.6 on Windows 10.

Edit: better esthetics.

using Gnuplot, Dates

dates = [ Date(2020, 4, floor.(Int, x .* 30 .+ 1)) for x in rand(10) ]
val=rand(10)

@gp "set xdata time" :-
@gp :- "set timefmt '\"%Y-%m-%d\"'" :-
@gp :- "set format x '%Y-%m-%d'" "set xtics rotate by -45" :-
@gp :- "set rmargin 7" :-
@gp :- "set style line 1 lc rgb 'blue' lt 1 lw 8 pt 7 ps 2.0" :-
@gp :- string.(dates) val "u 1:2 ls 1 notitl"

[oheil]

ok, It’s not easy to give up, so I tried further.

My conclusion is, install Gnuplot.jl v1.3.0:

(@v1.6) pkg> add Gnuplot@1.3.0

I had issues with Julia 1.5.3, so I tried with 1.6 and voila, it works now as in the example.

(Ninja’d by @PetrKryslUCSD )