Axis

Creating an Axis

The Axis is a 2D axis that works well with automatic layouts. Here's how you create one

using CairoMakie

f = Figure()

ax = Axis(f[1, 1], xlabel = "x label", ylabel = "y label",
    title = "Title")

f

Plotting into an Axis

You can use all the normal mutating 2D plotting functions with an Axis. These functions return the created plot object. Omitting the ax argument plots into the current_axis(), which is usually the axis that was last created.

lineobject = lines!(ax, 0..10, sin, color = :red)
scatobject = scatter!(0:0.5:10, cos, color = :orange)

f

Deleting plots

You can delete a plot object directly via delete!(ax, plotobj). You can also remove all plots with empty!(ax).

using CairoMakie

f = Figure()

axs = [Axis(f[1, i]) for i in 1:3]

scatters = map(axs) do ax
    [scatter!(ax, 0:0.1:10, x -> sin(x) + i) for i in 1:3]
end

delete!(axs[2], scatters[2][2])
empty!(axs[3])

f

Setting Axis limits and reversing axes

You can set axis limits with the functions xlims!, ylims! or limits!. The numbers are meant in the order left right for xlims!, and bottom top for ylims!. Therefore, if the second number is smaller than the first, the respective axis will reverse. You can manually reverse an axis by setting ax.xreversed = true or ax.yreversed = true.

Note that if you enforce an aspect ratio between x-axis and y-axis using autolimitaspect, the values you set with these functions will probably not be exactly what you get, but they will be changed to fit the chosen ratio.

using CairoMakie

f = Figure()

axes = [Axis(f[i, j]) for j in 1:3, i in 1:2]

for (i, ax) in enumerate(axes)
    ax.title = "Axis $i"
    poly!(ax, Point2f0[(9, 9), (3, 1), (1, 3)],
        color = cgrad(:inferno, 6, categorical = true)[i])
end

xlims!(axes[1], [0, 10]) # as vector
xlims!(axes[2], 10, 0) # separate, reversed
ylims!(axes[3], 0, 10) # separate
ylims!(axes[4], (10, 0)) # as tuple, reversed
limits!(axes[5], 0, 10, 0, 10) # x1, x2, y1, y2
limits!(axes[6], BBox(0, 10, 0, 10)) # as rectangle

f

Setting half-automatic limits

You can set half limits by either giving one argument as nothing or by using the keyword syntax where only low or high is given.

using CairoMakie

f = Figure()

data = rand(100, 2) .* 0.7 .+ 0.15

Axis(f[1, 1], title = "xlims!(nothing, 1)")
scatter!(data)
xlims!(nothing, 1)

Axis(f[1, 2], title = "xlims!(low = 0)")
scatter!(data)
xlims!(low = 0)

Axis(f[2, 1], title = "ylims!(0, nothing)")
scatter!(data)
ylims!(0, nothing)

Axis(f[2, 2], title = "ylims!(high = 1)")
scatter!(data)
ylims!(high = 1)

f

This also works when specifying limits directly, such as Axis(..., limits = (nothing, 1, 2, nothing)).

Auto-reset behavior

When you create a new plot in an axis, reset_limits!(ax) is called, which adjusts the limits to the new bounds. If you have previously set limits with limits!, xlims! or ylims!, these limits are not overridden by the new plot. If you want to override the manually set limits, call autolimits!(ax) to compute completely new limits from the axis content.

The user-defined limits are stored in ax.limits. This can either be a tuple with two entries, where each entry can be either nothing or a tuple with numbers (low, high).It can also be a tuple with four numbers (xlow, xhigh, ylow, yhigh). You can pass this directly when creating a new axis. The same observable limits is also set using limits!, xlims! and ylims!, or reset to (nothing, nothing) using autolimits!.

using CairoMakie

f = Figure()

lines(f[1, 1], 0..10, sin)
lines(f[1, 2], 0..10, sin, axis = (limits = (0, 10, -1, 1),))

f

Modifying ticks

To control ticks, you can set the axis attributes xticks/yticks and xtickformat/ytickformat.

You can overload one or more of these three functions to implement custom ticks:

tickvalues, ticklabels = MakieLayout.get_ticks(ticks, scale, formatter, vmin, vmax)
tickvalues = MakieLayout.get_tickvalues(ticks, vmin, vmax)
ticklabels = MakieLayout.get_ticklabels(formatter, tickvalues)

If you overload get_ticks, you have to compute both tickvalues and ticklabels directly as a vector of floats and strings, respectively. Otherwise the result of get_tickvalues is passed to get_ticklabels by default. The limits of the respective axis are passed as vmin and vmax.

A couple of behaviors are implemented by default. You can specify static ticks by passing an iterable of numbers. You can also pass a tuple with tick values and tick labels directly, bypassing the formatting step.

As a third option you can pass a function taking minimum and maximum axis value as arguments and returning either a vector of tickvalues which are then passed to the current formatter, or a tuple with tickvalues and ticklabels which are then used directly.

For formatting, you can pass a function which takes a vector of numbers and outputs a vector of strings. You can also pass a format string which is passed to Formatting.format from Formatting.jl, where you can mix the formatted numbers with other text like in "{:.2f}ms".

Predefined ticks

The default tick type is LinearTicks(n), where n is the target number of ticks which the algorithm tries to return.

using CairoMakie

fig = Figure()
for (i, n) in enumerate([2, 5, 9])
    lines(fig[i, 1], 0..20, sin, axis = (xticks = LinearTicks(n),))
end
fig

There's also WilkinsonTicks which uses the alternative Wilkinson algorithm.

MultiplesTicks can be used when an axis should be marked at multiples of a certain number. A common scenario is plotting a trigonometric function which should be marked at pi intervals.

using CairoMakie

lines(0..20, sin, axis = (xticks = MultiplesTicks(4, pi, "π"),))

Here are a couple of examples that show off different settings for ticks and formats.

using CairoMakie

scene, layout = layoutscene(resolution = (1200, 900))

axes = layout[] = [Axis(scene) for i in 1:2, j in 1:2]

xs = LinRange(0, 2pi, 50)
for (i, ax) in enumerate(axes)
    ax.title = "Axis $i"
    lines!(ax, xs, sin.(xs))
end

axes[1].xticks = 0:6

axes[2].xticks = 0:pi:2pi
axes[2].xtickformat = xs -> ["$(x/pi)π" for x in xs]

axes[3].xticks = (0:pi:2pi, ["start", "middle", "end"])

axes[4].xticks = 0:pi:2pi
axes[4].xtickformat = "{:.2f}ms"
axes[4].xlabel = "Time"


scene

Minor ticks and grids

You can show minor ticks and grids by setting x/yminorticksvisible = true and x/yminorgridvisible = true which are off by default. You can set size, color, width, align etc. like for the normal ticks, but there are no labels. The x/yminorticks attributes control how minor ticks are computed given major ticks and axis limits. For that purpose you can create your own minortick type and overload MakieLayout.get_minor_tickvalues(minorticks, tickvalues, vmin, vmax).

The default minor tick type is IntervalsBetween(n, mirror = true) where n gives the number of intervals each gap between major ticks is divided into with minor ticks, and mirror decides if outside of the major ticks there are more minor ticks with the same intervals as the adjacent gaps.

using CairoMakie

theme = Attributes(
    Axis = (
        xminorticksvisible = true,
        yminorticksvisible = true,
        xminorgridvisible = true,
        yminorgridvisible = true,
    )
)

fig = with_theme(theme) do
    fig = Figure()
    axs = [Axis(fig[fldmod1(n, 2)...],
        title = "IntervalsBetween($(n+1))",
        xminorticks = IntervalsBetween(n+1),
        yminorticks = IntervalsBetween(n+1)) for n in 1:4]
    fig
end

fig

Hiding Axis spines and decorations

You can hide all axis elements manually, by setting their specific visibility attributes to false, like xticklabelsvisible, but that can be tedious. There are a couple of convenience functions for this.

To hide spines, you can use hidespines!.

using CairoMakie

f = Figure()

ax1 = Axis(f[1, 1], title = "Axis 1")
ax2 = Axis(f[1, 2], title = "Axis 2")

hidespines!(ax1)
hidespines!(ax2, :t, :r) # only top and right

f

To hide decorations, you can use hidedecorations!, or the specific hidexdecorations! and hideydecorations!. When hiding, you can set label = false, ticklabels = false, ticks = false, grid = false, minorgrid = false or minorticks = false as keyword arguments if you want to keep those elements. It's common, e.g., to hide everything but the grid lines in facet plots.

using CairoMakie

f = Figure()

ax1 = Axis(f[1, 1], title = "Axis 1")
ax2 = Axis(f[1, 2], title = "Axis 2")
ax3 = Axis(f[1, 3], title = "Axis 3")

hidedecorations!(ax1)
hidexdecorations!(ax2, grid = false)
hideydecorations!(ax3, ticks = false)

f

Log scales and other axis scales

The two attributes xscale and yscale, which by default are set to identity, can be used to project the data in a nonlinear way, in addition to the linear zoom that the limits provide.

Take care that the axis limits always stay inside the limits appropriate for the chosen scaling function, for example, log functions fail for values x <= 0, sqrt for x < 0, etc.

using CairoMakie

data = LinRange(0.01, 0.99, 200)

f = Figure(resolution = (800, 800))

for (i, scale) in enumerate([identity, log10, log2, log, sqrt, Makie.logit])

    row, col = fldmod1(i, 2)
    Axis(f[row, col], yscale = scale, title = string(scale),
        yminorticksvisible = true, yminorgridvisible = true,
        yminorticks = IntervalsBetween(8))

    lines!(data, color = :blue)
end

f