- A brief introduction, with links to help you get idl running on your display.
- Getting started
- A few basic hints that will help you get started with the command line interface.
- Programs and batch mode
- IDL can be run by typing commands interactively, by creating programs interactively, by reading programs in from files, or it can be run in batch mode.
- Variables and arithmetic
- Examples of setting scalar, vector, and array variables, and performing some basic operations.
- Examples of matrix operations.
- Plots of Y vs. X
- Plots of a function of one variable.
- Surface plots
- Examples of surface plots of functions of two variables.
- Example animation
- How to save your images to a postscript file.
- Why do my graphics get erased?
- When a window gets covered, then uncovered, someone has to keep a copy of the obscured part of the image.
IDL (Interactive Data Language) is a general purpose scientific computing package which provides a suite of mathematical functions, data analysis tools, as well as some scientific visualization and animation tools. IDL is a mature, robust, commercial package sold by Visual Information Systems.
IDL can be used interactively or in "batch" mode. There is an efficient command interpreter for interactive use, as well as the option of developing scripts which may be interpreted or compiled. There is also a graphical IDL development environment available using the command "idlde" which we will not discuss in this tutorial.
This tutorial is organized as a set of examples with explanations. This tutorial assumes that you are working in a directory named idl, that IDL programs are in a subdirectory named pro, and that data is in a subdirectory named dat. The examples used in this tutorial are available for download here Examples.
The most effective way for you to go through this tutorial is by running IDL in a separate window, and trying out the commands and programs as you read the tutorial. It should be kept in mind that this tutorial is only a glimpse at what you can do with IDL. Once you start using it, you will want to look at the online and/or hardcopy documentation, and explore.
Boston University SCF users, please go to the SCV IDL Help Page for information specific to the installation at this site. This will tell you on what machines IDL is running, how to set up your environment, how to set your display, and where the documentation is.
In this tutorial, we will be running IDL directly from the Unix prompt. You start IDL by typing the command "idl":
which will present you with a prompt:
Try typing the following four commands at the IDL prompt:
IDL> a = 5 IDL> print, a IDL> x = randomu(999) IDL> print, x
Observe the difference in syntax between the "print, a" statement and the "x = randomu(999) statement. IDL distinguishes between procedures (such as print) which do not return a value and do not have parentheses in their call syntax, and functions (such as randomu) which do use parentheses when called and always return a value.
Now try out these commands:
IDL> a = [55, 38, -17.333] IDL> b = indgen(3) IDL> print, "a = ", a IDL> print, "b = ", b IDL> print, "a*b = ", a*b
Observe that array variables look just like scalar variables. IDL is designed to work on multi-dimensional arrays, and many operators work on arrays in "the natural way." We will see more on this later in the tutorial.
To repeat a command, you can go up and down through previous commands using the arrow keys. When you reach the command you want to repeat, hit return.
Type a question mark
? for online help. This will pop up a separate GUI for accessing IDL user information. (Note that typing "help" at the IDL prompt does something quite different - it gives you a list of all current variables and functions).
Preceding a command with a dollar sign "
$" allows typing commands to the Unix shell, e.g., to print a listing of the current working directory:
A semicolon "
;" indicates that the rest of a line consists of comments. A dollar sign "
$" at the end of a line indicates line continuation, and an ampersand "
&" groups commands together. We will see more of this below in the section on running modes.
exit IDL, type:
There are some special characters which you should be aware of for execution management:
|Ctl-C||Stops running (runaway) program if possible, without killing IDL|
|Ctl-Z||Suspends IDL in usual Unix manner|
|Ctl-||Stops (aborts) running program, may leave variables and files in uncertain state|
|Ctl-D||Stops IDL, like exit|
IDL can be run by typing commands interactively, by creating programs interactively, by reading programs in from the command line, or it can be run in batch mode.
When you type commands on the command line, each line is executed immediately when you hit the return key. (It is possible to carry over to the next line using a dollar sign "
$" at the end of the line).
Running in batch mode is similar, except the commands come from a file. You precede the file name with the at-sign "
@". The file batch_two_prints in the pro subdirectory contains the four lines
a = 5 print, a a = [2, 3] print, a
Type the following at the IDL prompt
This will execute the four commands exactly as if you had typed them. Also, this can also be done directly from your Unix shell, by typing
scc1% idl pro/batch_two_prints
When typing interactively and running in batch mode, each line is executed immediately. You can write programs which can be stored and run multiple times, and you can write functions which may be called from programs and other functions.
To create a program from the command line, use the executive command
.RUN, enter your commands, then type
END. At this time, your program will be compiled and executed:
IDL> .RUN - a = 25 - b = 3 - c = a * b - print, a, b, c - END % Compiled module: $MAIN$. 25 3 75
If you had the same set of commands in a file, you could execute the same program:
IDL> .RUN pro/simple_main.pro
If you don't want the program to be executed immediately, use
.COMPILE instead of
.RUN. The variables and procedures defined will then be available.
The examples above make it look like batch mode and running programs is very similar. Consider the following simple computation of the factorial function. As a batch file, pro/batch_loop looks like:
f = 1 for k=1,6 do begin $ f = k * f & $ print, f & $ end
The commands to be separated by ampersands "
&", and the lines must be continued. The entire loop must essentially be on one line, since each line is executed as soon as it is encountered. Imagine nested loops with long calculations inside.
The same computation as a program in a file does not require the &s and $s:
f = 1 for k=1,6 do begin f = k * f print, f endfor end
For an example of the use of functions:
IDL> .RUN pro/simple_routine IDL> simple
A program can be run from a batch file. Try:
This reads and runs simple_main.pro, then runs it again, then reads and runs simple_routine.pro.
For more information on creating and running programs, including other commands, see the IDL online help.
All of the examples shown in this section can be found in pro/arithmetic.pro. You can explicitly type variables, or not. See chapter 4 for information on type conversion, etc. The simplest thing to work with is scalars.
IDL> y = 2.5 IDL> z = x + y IDL> w = x^y + sin(z) IDL> print, x, y, z, w 3 2.50000 5.50000 14.8829
Square braces are used to define vectors (1-dimensional arrays):
IDL> v1 = [1, 2, 0] IDL> v2 = [1, 0, 0] IDL> print, "v1 = ", v1 v1 = 1 2 0 IDL> print, "v2 = ", v2 v2 = 1 0 0
Vectors can be componentwise added, multiplied, etc.:
IDL> v3 = v1 + v2 IDL> print, "v3 = v1 + v2 = ", v3 v3 = v1 + v2 = 2 2 0 IDL> print, "v1 * v2 = ", v1 * v2 v1 * v2 = 1 0 0 IDL> print, "v1 * sin(v3) = ", v1 * sin(v3) v1 * sin(v3) = 0.909297 1.81859 0.00000
There are other useful operators, such as
IDL> min1 = min(v1) IDL> max1 = max(v1) IDL> print, "min(v1), max(v1) = ", min1, max1 min(v1), max(v1) = 0 2
Scalars and arrays can be allocated with specific types. Scalar examples:
IDL> x = float(1.3) IDL> sx = fix(x) IDL> lx = long(x) IDL> bx = byte(x) IDL> dx = double(x) IDL> cx = complex(x) IDL> print, x, sx, lx, bx, dx, cx 1.30000 1 1 1 1.3000000 ( 1.30000, 0.00000)
IDL> a = fltarr(5) IDL> for i=0, 4 do $ IDL> a(i) = 2*i IDL> b = complex(a) IDL> print, "b = ", b b = ( 0.00000, 0.00000)( 2.00000, 0.00000) ( 4.00000, 0.00000)( 6.00000, 0.00000) ( 8.00000, 0.00000)
All of the examples shown in this section can be found in pro/matrix.pro. A matrix (which is a 2-dimensional array) may be defined algorithmically:
IDL> A = dblarr(2, 4) IDL> for i = 0, 1 do begin $ IDL> for j = 0, 3 do begin $ IDL> a(i, j) = 10 * i + j IDL> print, A 0.0000000 10.000000 1.0000000 11.000000 2.0000000 12.000000 3.0000000 13.000000
Note that as it is printed, the first index corresponds to the column, and the second index to the row. Another way to think of it is that the way the data is stored, the first index varies fastest, and the second varies the slowest. This agrees with the way the data is printed.
A matrix may be constructed explicitly from vectors:
IDL> v1 = [1, 2, 0] IDL> v2 = [1, 0, 0] IDL> v3 = [4, 5, 6] IDL> A = [[v1], [v2], [v3]] IDL> print, A 1 2 0 1 0 0 4 5 6
Create the transpose:
IDL> Atrans = transpose(A) IDL> print, Atrans 1 1 4 2 0 5 0 0 6
Take the determinant:
IDL> d = determ(float(A)) % Compiled module: DETERM. IDL> print, d -12.0000
IDL> Ainv = invert(A) IDL> print, Ainv 0.00000 1.00000 0.00000 0.500000 -0.500000 0.00000 -0.416667 -0.250000 0.166667
Multiply vectors by matrices:
IDL> v = [1, 2, 3] IDL> print, A 1 2 0 1 0 0 4 5 6 IDL> print, v 1 2 3 IDL> print, A ## v 5 1 32 IDL> print, v ## A 15 17 18
You can solve a linear system Ax = b for x by Cramer's rule (the cramer function expects float or double inputs, requiring an explicit type conversion):
IDL> b = float([1, 2, 16]) IDL> A = float(A) IDL> x = cramer(A, b) IDL> print, x 2.00000 -0.500000 1.75000
Two examples of plots of a function of one variable.
The program pro/fit_y.pro reads in a set of 10 y values. It creates 10 x values from 0 to 9 using the "
findgen" command. It then fits a cubic to the data, and plots the original data as points, and the fitted function as a curve.
The program pro/fit_xy.pro reads in a set of 10 (x,y) pairs, into a 2 x 10 array. It separates them into two 1-dimensional arrays, and fits a cubic to it, and plots this as in the preceding example.
IDL provides an interactive viewer for surface plots, called xsurface. An example of its use is in pro/xsurface.pro
An example of drawing a surface plot directly, rendered as a wire mesh, is in pro/mesh_surf.pro.
An example of drawing a surface plot directly, rendered as a shaded surface, is in pro/shade_surf.pro.
There are two examples of producing surface plots from scattered data. This is data in which the (x,y) locations for which we have the function evaluated do not lie on a regular gird. To make a surface plot, IDL needs to have the function evaluated on a regular rectangular grid. There are two steps involved. The first is to form a triangulation using the input (x,y) points to use for interpolation, and the second is to produce a mesh from that interpolation. The first example pro/triangulate.pro shows this process and renders the result as a wire mesh surface plot. The second example pro/tri_shade does the same thing, and in addition draws a shaded surface plot, using two different kinds of shading. (To see this example, first load the program:
IDL> .run pro/tri_shade, then execute the compiled procedure:
One method of producing animation is to create a sequence of images and then display them in order. This is quite easy using IDL. An example is in pro/making_waves. (To see this example, first load the program:
IDL> .run pro/making_waves, then execute the compiled procedure:
You can save your plots and other images by rendering them to a postscript file instead of to an X window. An example is given in pro/tri_save_ps.pro. Basically, you set your plotting area to be a file, then change the graphics device to be a postscript device, and close that device when you are done:
IDL> set_plot,'PS' IDL> device, filename='your_filename.ps' IDL> ... IDL> ... plot some stuff ... IDL> ... IDL> device, /close
When a window gets covered, then uncovered, someone has to keep a copy of the obscured part of the image. You may or may not want to have all the images saved when they are obscured, by reasons of speed and memory.
In any case, this topic is called "backing store". It can be done by IDL, done by the windowing system, or not done. By default, the X window system does not have backing store turned on.
In IDL, there is a keyword RETAIN, for specifying which kind of backing store to use.
- RETAIN = 0 implies no backing store,
- RETAIN = 1 IDL asks the window system to do it,
- RETAIN = 2 IDL does it
This may be done on a per window basis, e.g.,
Backing store will now be maintained for this window by IDL.