When you are working with a lot of numbers and complicated functions it is often useful to be able to do some plotting of the data. GSL Shell offers a graphics module that allow to produce beatiful graphics with a very rich interface and almost illimited possibilities.
GSL Shell use the excellent Anti-Grain Geometry library, version 2.5, of Maxim Shemanarev.
GSL shell offer a graphics interface with few but powerful functions that, if appropriately used allows to the user to produce a wide range of plot types.
Let’s start with a simple example, let us suppose that we want to plot the function:
where α and ω are constants and t vary from 0 to t1. We can plot this function with GSL Shell with the following instructions:
function myplot(alpha, omega, t1)
-- create a new plot, it is not shown for the moment
local p = plot('f(x) = exp(-a t) sin(w t)')
-- we create a line that corresponds to our function
local ln = fxline(|t| exp(-alpha*t)*sin(omega*t), 0, t1)
-- we add the line to the plot and show it
p:addline(ln, 'red')
p:show()
return p
end
Then to plot something you have just to call the ‘myplot’ function. For example:
myplot(0.3, 5, 6)
The function fxline() takes three arguments, the function to plot and the initial anf final values of the variable. By default the function will be sampled with 256 points but if you want you can provide a fourth arguments to give the number of sample points.
In this example we have used the plot() function to create a plot, the fxline() function to create the line to draw and the method addline() to add the line to the plot (in red). These three operations can be done with a single function, fxplot(). It works like that:
p = fxplot(|x| sin(x), 0, 8*pi)
where the first arguments is the function to plot and the following arguments are the extrema of variation of the indipendent variable x.
To create many type of plots you don’t really need to use the graphics primitives because it would be annoying and tedious to give a lot of instructions just to produce some simple plots. Instead you can use the higher level plotting functions to make most of the most common type of plots.
Produces a plot of the function f(i) where ‘i’ is an integer variable going from a to b. In the second abbreviated form a take the default value of one. The function returns the plot itself.
Example:
-- plot the 'choose' function for some integer values
fiplot(|i| choose(12, i), 12)
-- plot of a geometric series
fiplot(|n| 0.9^n, 0, 36)
This function returns an graphical object of type path given by the points (x, f(x)) for x going from xi to xs with n sampling point.
Example:
-- we create a 'path' for the function f(x) = sin(x)*exp(-0.1*x)
-- and plot it
p = plot('Example')
line = fxline(|x| sin(x)*exp(-0.1*x), 0, 10*pi)
p:addline(line)
p:show()
You can produce very easily a nice illustration of the “area below a curve” with the following commands:
p = plot('Example')
line = fxline(|x| sin(x)*exp(-0.1*x), 0, 10*pi)
p:add(line, rgba(1,1,0,0.6))
p:addline(line)
p:show()
and it will produce the following output:
This function returns an graphical object of type path given by the points (i, f(i)) where ‘i’ is an integer variable going from a to b. It a is omitted values will be in the interval 1 .. b.
For its usage see the similar function fxline().
This function takes two column matrix of dimendions N as arguments and returns a graphical object of type path given by the points (x[i], y[i]) where i goes from 1 to N.
Example:
N = 256
sio2n = |w| sqrt(1.4923 + 0.61497*w^2/(w^2 - 0.115^2)-0.01059*w^2)
-- define a matrix with wavelength and refrective index in each column
d = new(N, 2, function(i,j)
local w = 0.4*(i-1)/(N-1)+0.4
return j == 1 and w or sio2n(w)
end)
p = plot('SiO2 dispersion curve in visible range')
line = xyline(d:col(1), d:col(2))
p:addline(line)
p:show()
This function takes an iterator function f and returns a path given by the points (x, y) returned by the iterator f. The variant ipathp() is able to treat the case when the function f fails and it does continue by calling the iterator again.
In the following example we shows how to create a circle by using an iterator to generates the points. We define first an iterator that use a counter to generate the points and then we pass the iterators to the function ipath(). In this way we obtain an object of type path and we can add into a plot.:
-- create a simple iterator that return n points uniformly spaced
-- in a circle centerd in (x0, y0) with radius R
circle = function(x0, y0, R, n)
local k = 0
return function()
if k <= n then
local th = 2*k*pi/n
local x, y = x0 + R*cos(th), y0 + R*sin(th)
k = k+1
return x, y
end
end
end
-- then we use ipath to create a path from the iterator
line = ipath(circle(1, 1, 2.5, 256))
p = plot('circle at (1,1) with R= 2.5')
p:add(line, rgba(1,1,0,0.6))
p:addline(line)
p:show()
This function takes an iterator function f and returns a path object that draws many adjacent rectangular boxs correspondings to the points (x, y) returned by the iterator f. This function simplify the creation of histograms.
Example:
-- draws the histogram of a simulated gaussian distribution
N = 800
r = rng()
f = |x| 1/sqrt(2*pi) * exp(-x^2/2)
p = plot('Simulated Gaussian Distribution')
b = ibars(sample(|x| rnd.poisson(r, f(x)*N) / N, -3, 3, 25))
p:add(b, 'darkgreen')
p:add(b, 'black', {{'stroke', width= 0.5}})
p:addline(fxline(f, -4, 4), 'red')
p:show()
We obtain the following plot:
In order to create a ‘path’ we can use the function path() which just creates an empty path. Then we add the the points by using the methods move_to() and line_to() methods. When you use move_to() a new curve is started at the given points and with line_to() you can draw your curve.
So to plot a triangle you can give the following instructions:
t = path() -- a new path is created
t:move_to(0, 0) -- starting point
t:line_to(10, 0) -- second vertex
t:line_to(5, 8.66) -- third vertex
t:close()
p = plot()
p:add(t, 'red')
p:show()
Please not that we have used the add() method instead of addline() to add the path.
Now let us suppose that we want to plot only the contour of the triangle with a line 10 pixel thick and with round edges. Then what you have to do is to supply to the add() method a third argument where you specify a stroke transformation:
p = plot()
p:add(t, 'red', {{'stroke', width=10, cap='round'}})
p:show()
As you can see we have used the ‘stroke’ trasnformation and we have provided two parameters: ‘width’ and ‘cap’ to specify of the stroke should be done.
Now, to continue our example, let us suppose that we want to plot a dashed line just one pixel thick. In this case we have to cascade two transformations, the ‘dash’ transformation and the ‘stroke’ transformation. It can be done as follows:
p = plot()
p:add(t, 'red', {{'stroke'}, {'dash', a= 10, b= 5}})
p:show()
the ‘a’ and ‘b’ parameters specifies the lengths of the dash and of the blank space between the dashes. This length is calculated in pixel as it does depends on the size of the windows. Let us suppose now that you want to give the length of the dashes in the user coordinate system. In this case you have to provide it as a fourth arguments to the add(). So you should do something like that:
p = plot()
p:add(t, 'red', {{'stroke'}}, {{'dash', a= 0.5, b= 0.25}})
p:show()
You can note that we have changed the size of the dashes because in the user coordinate system the value 10 is big as the edge of the triangle.
So, if general, the add() methods takes two optional arguments, the list of the post-transforms and the list of the pre-transforms. The post-transforms are made when we are already in the window coordinate system while the pre-transforms are made in the user coordinate system.
Some transformations are naturally expressed as post-transforms because they does operates naturally in term of pixel while other are usually expressed as pre-transforms because they operates in the user coordinates space.
We have seen in the previous paragraph that you can add more graphical elements in a plot by using the methods add() and addline(). The method addline() is just a shortcut to add elements with a ‘stroke’ post transform of unitary width.
If can add elements to a plot in any moments even when it is already shown. GSL Shell will automatically calculate the bounding box so that every elements is shown on the window.
A text object is used to display a text. This class of graphical object is special because it is not a shape that is resized rigidly with the window, the text is always displayed with the same size in pixel. Becasue of this difference a text object should be subject only to post-transforms.
A generical graphical transformation is expressed in the form of table with the following layout:
{'name',
property1 = value1,
property2 = value2,
...
}
For example, to express a ‘stroke’ transform you can write:
{'stroke', width= 5}
to mean a stroke transformation with a stroke width of 5.
Here a complete list of all the available transforms:
- stroke
A stroke create an outline of the given path. The properties are:
- width, the width of the stroke, default value is width=1
- cap, can be ‘round’, ‘butt’ or ‘square’. The default value is ‘butt’.
- join, can be ‘miter’, ‘miter.rev’, ‘miter.round’, ‘round’ and ‘bevel’
- dash
Transform the path to a sequence of dashes.
- a, the length of the dash
- b, the length of the gap between consecutives dashes
- curve
- This transformation make the ‘curve3’ and ‘curve4’ path elements became real curves.
- marker
Replace each vertex of the path with a circular mark
- size, the size of the marker
- translate
A translation along the x and y axis. This transformation can be used only in the user coordinate system.
- x, translation along the x axis
- y, translation along the y axis
- rotate
A rotation of a given angle with respect of the origin. This transformation can be used only in the user coordinate system.
- angle, the angle of the rotation, in radians.