R. W. O'CONNELL
IDL EXERCISES I:
SCALARS, VECTORS, PLOTTING
NOTE: commands intended to be typed into your terminal during your IDL
session are given in CAPITAL letters below so they will stand out.
However, IDL itself is case-insensitive (except for file names), so
you needn't follow this convention.
0) From a UNIX shell, start X-windows if it is not already running.
From a shell in the X-window, "cd" into your "IDL" directory.
Normally that would be named "~/idl".
[On most UNIX systems, just type "idl".
This will start IDL running in the command-line mode using the
window from which you called it. In IDL V.5+ you can use
GUI-based "Development Environment" or "Workbench" versions of
IDL. These offer convenience features to the experienced user
but are correspondingly more complicated and not recommended for
learning the basics.]
Before starting the exercises, give the IDL command:
This will return control to the main program level if
an error occurs in a called routine.
1) PRACTICE WITH SCALARS
Define X and Y to be two different integers
Print their sum and difference to the terminal screen
Repeat, all on one IDL command line, using the "&" character
Repeat all of the above, using the recall buffer ("up" arrow)
Print the result of adding 3 times X to 2 times Y
Print the product of (X-Y) and (X+Y)
Print the common logarithm of the absolute value of X+Y (use
the ABS and ALOG10 functions)
Confirm this result by raising 10.0 to the power
you just obtained
What is the difference between X = 5/2 and Y = 5./2 ?
Between X = FLOAT(5/2) and Y = FLOAT(5)/2 ?
Set C equal to the numerical value of the speed of light in
Successively print powers of C until you locate the point
at which the result becomes "Inf". What power yields this
result? [The answer is non-integer.]
Now define C to be the speed of light in double precision.
Repeat the exercise above.
[Hint: you should accelerate all such exercises by making
liberal use of the recall buffer and line editing.]
String example: define a string variable for each word in the
sentence: "This is a concatenation"; then concatenate them
and print the result.
2) PRACTICE WITH VECTORS
Create X = INDGEN(100).
[NOTE on IDL syntax:
INDGEN is a function. In IDL the arguments of a function are
enclosed in parentheses. In IDL versions 1 through 4, the
indices of vector or array variables were also enclosed in
parentheses, as in X(10). Obviously, this introduced
possible ambiguities between function and array notation.
Therefore, starting with IDL version 5, array indices are
expected to be enclosed by brackets, as in X. You should
use this syntax when referring to array elements. This
notation for array elements is not compatible with IDL V.4
and earlier. However, IDL versions 5 and later do accept the
parenthesis syntax for array elements, so are compatible with
earlier IDL code in this regard.]
Use built-in IDL functions N_ELEMENTS, MIN, MAX, and TOTAL to
answer the following:
How many elements does X contain?
What is the minimum value in X?
What is the maximum value in X?
What is the sum of all the elements in X?
Use the IDL HyperHelp facility to obtain information on
the built-in TOTAL routine you just used: ?TOTAL.
Experiment on X or other vectors you create with some of the
special keywords available for TOTAL.
Print X to your terminal window
How are values in X related to the corresponding subscripts?
Is X a floating-point array? Compare the output you
just got with the result of PRINT,FINDGEN(100).
Print the fifteenth entry in X to your terminal
Then print the entry containing the number 15
Is X a row-vector or a column-vector?
The default configuration of the printed data on your screen
will tell you. To confirm this, try the following:
Try the following one-line command and inspect its output:
FOR I=0,99 DO PRINT,I,X[I]
Now try: FOR I=1,100 DO PRINT,I,X[I]
Using the WHERE function:
Define Q=2*X, then type FIND = WHERE(Q LE 40, COUNT)
FIND will be a vector. Examine the contents of FIND and
COUNT so that you understand how the WHERE function operates.
Predict and confirm the response if you type PRINT,Q(FIND)
Print to your terminal the values of the vector X/10
Then print the values of X/10.0 and compare the results.
Do the same for FLOAT(X)/10
Do the same for FIX(X/10.0)
What is the difference between Z = X*0.0 and Z = 0?
Print the 11 elements centered on X = 10
First do this using a FOR loop (on a single line)
Then do this using the standard IDL subscript range
notation, e.g. X[2:6]. No FOR loop is needed.
Compare to the following: K=5 & PRINT,X[10-K:10+K]
Define Y to be the subarray consisting of those 11 elements, using
subscript range notation; print Y to your window as a
Using information utilities:
Verify lengths of X,Y using the N_ELEMENTS utility
Use the SIZE utility to find the sizes of X and Y; what
other info does it supply?
What information does the command HELP,X,Y provide?
Define Q = Y+3 -- Note the values that Q contains
Define Q = Y*3 -- Note the values that Q contains
Define Q = Y^3 -- Note the values that Q contains
Define Q = Y^4 -- Note the values that Q contains; why
are they not monotonic?
Define Q = FLOAT(Y)^4 -- Note the change.
Print the vector which results from Q = X*Y
What did IDL actually do to arrive at this result?
Using the built-in functions TOTAL and N_ELEMENTS:
Find and print the mean values of X and Y;
Find and print the variances of the two arrays
Do the same using the built-in functions MEAN and VARIANCE.
Do the same using the built-in function MOMENT
Determine the nature of the vector which results when you
write Q = X & Q = Y. What will happen if you
write Q = Y? Try it.
Predict, then verify, the outcome of the following operations:
Q = [Y,Y]
Q = [Q,Y]
Create a 16-element vector [1.01, 1.02, 1.03...] using
a simple one-line command employing FINDGEN. Verify.
Create a 16-element vector: [1,2,4,8,16....] using a simple
one-line command employing FINDGEN. Verify.
Create the 100-element vector Z = 10*X - 0.1*X^2, where
X contains the integers between 0 and 99.
Use MAX to determine the maximum value of Z.
Use WHERE to locate the X value for which this maximum
X was an integer vector but Z is not. Why?
Using the SHIFT function, shift the elements of Z three entries
to the left. Verify that the maximum is now in the expected
Create, using a simple one-line command employing FINDGEN, a
1001-element vector containing the base-10 logarithms
of the integers between 0 and 1000. Name this "ILOG".
Verify its contents.
Now create a five element vector, Y, containing the
integers 2, 100, 500, 20, and 999.
Explain the vector resulting when you type
NEW = ILOG[Y]. Why did we include an entry for 0
in defining ILOG?
Optional problem: ILOG in the previous exercise is called a
"lookup table." It can be used to accelerate computations
in problems where a large number of time-consuming
transformations such as logarithms are needed. You can
estimate the time savings for this example as follows:
Write a simple one-line IDL command script using "&" as a
link between individual commands. Use the SYSTIME(1)
function to determine the start time. Then, use a FOR
loop to compute the logarithm of an integer 40000 times
using the standard ALOG10 function. Then use SYSTIME(1)
again to determine the end time. Print the elapsed time.
Repeat, now using the "ILOG" lookup table instead
3) JOURNAL FILES AND SCRIPT FILES
Use the JOURNAL function to start a journal file & record your
session for posterity. Test starting & stopping a journal file.
Inspect the file to see what kinds of communications between you &
the IDL session are actually recorded. [Note that journal files
are actually written to disk by IDL only at long intervals or after
another JOURNAL command or EXIT.]
Practice cut-and-pasting commands from a saved journal file listing
into an active IDL window.
Practice generating a "script" file from edited parts of an IDL
journal file. Save it, rename it, and use it to re-create the
original session by using the "@" command.
Note that comments can be added at any time to the journal by
prefacing remarks with a semicolon.
4) BASIC PLOTTING
Type "WINDOW,0" to bring up a plotting window. Move the cursor
into the window.
Troubleshooting: If the other parts of your terminal screen blink
out or change color, then you are using an 8-bit (rather than
24-bit) color monitor for which the X-windows system is not
properly configured. To ameliorate the problem, try this:
As the very first two commands, type:
If this does not remedy the blinking, then other X-windows
applications have "reserved" too many display colors. You
may be able to reduce the color hogging using various
"Preference" or other settings on your applications. Exit
other applications (e.g. browsers) before starting IDL. If
this doesn't work, for the moment the best approach is just
to live with the blinking and try later to reconfigure.
For reference, you may want to call up the on-line help
documentation for the plotting keywords by typing: ?PLOT
Create the 100-element vector Z = 10*X - 0.1*X^2, where
X is the vector containing the integers between 0 and 99.
Type PLOT,X,Z. Examine the plot, noting the abscissa, ordinate,
and default scaling adopted.
Do the same for PLOT,X,2*Z. Note the change.
Type OPLOT,X,Z and note what happens
Do the same for PLOT,X,Z^2
Do the same for PLOT,Z^2,X
Use the optional plotting parameters XRANGE and YRANGE with
PLOT,X,Z^2 to enlarge various parts of the plot of the
Using the XLOG or YLOG keywords, plot log_10(X) vs. Z and
X vs. log_10(Z). (Note: you'll have to limit the X,Y
scales using the RANGE parameters to avoid infinities
and plot compression. Keep the variable on the log
axis greater than or equal to 1.0.)
Label any of these plots, using the !P.TITLE, !X.TITLE,
and !Y.TITLE system variable strings.
All of the plots should have appeared in Window 0. Try alternating
successive plots between Window 0 and Window 1.
[NOTE: The intrinsic IDL routines for manipulating windows
are as follows: To create a new window, use WINDOW,N. To
expose an existing window, use WSHOW,N. To make a given
window "active"--- i.e. ready for I/O---use WSET,N.
The MOUSSE routine CHAN,N combines these three functions
and is more convenient.]
Create the 1000-element vector x, where x contains the
FLOATING-POINT conversion of integer values between -500
and +499. Create Z = 10*X - 0.1*X^2 .
Explore plotting features using PLOT,X,Z as in the previous
Type PLOT,Z. How does the resulting plot differ from that
Create the 1000-element vector X, where x contains the INTEGER
values between -500 and +499. Create Y = 10*X - 0.1*X^2.
Type PLOT,X,Y. Why does the Y function differ from Z?
Now define X to contain the integers in the range 0 to 100. Then
compute Z = SIN(X)/X. Is Z a floating point variable?
Print the value of Z at X = 0. If you weren't sure how IDL would
respond there, how would you manually insert a value Z = 1.0
at X = 0? Do so. Confirm that the Z vector is now defined
Using the PSYM and LINESTYLE keywords, plot Z vs. X for integer
values of X in the range 0 to 100:
with a solid line
then with plus signs
then with open triangles
then with a dashed line
[Hint: for help with the keywords, open the online IDL HyperHelp
system by typing "?" as the first element of the command line.
Then enter "GRAPHICS KEYWORDS" in the search box.]
Plot Z vs. X for the X range [0:10] with open triangles. Then, using
OPLOT, add a solid line overplot.
You can achieve the same result with a single PLOT command
for your choice of plotting symbol, K, by using PSYM=-K
rather than PSYM=K.
Calculate Z = SIN(X)/X at intervals of 0.01 for X in the range 0
to 10. Plot Z vs X.
Using the WHERE function, find all the locations where Z has an
absolute value smaller than 0.05. How many are there? Print
Z for all those locations (but only those locations) to your
terminal. Print X for all those locations (but only those
locations) to your terminal.
Plot Z vs X using a solid line. Now overplot open triangles at
those points you found where Z has an absolute value smaller
than 0.05. You can verify graphically that these fall in
the expected positions by plotting horizontal lines at
Z = 0.05 and -0.05 (hint: define and plot two new vector
functions of X).
5) EXPLORING PARAMETER SPACE
Copy the program bio.pro from the IDLexercises/data directory
to your local directory.
This routine computes and displays a well-known simple function
exhibiting "chaotic" behavior for certain choices of input
parameters. To see the code and the header, type .RUN -T BIO .
To see just the header section, type MAN,'BIO . [Note: MAN is a
MOUSSE routine.] Run the program as suggested in its header to
explore the behavior of this function as you change the two input
parameters. Note the effects of small changes in these around the
6) ITERATED PLOTTING
Graphically solve the transcendental equation
X + 6.0*EXP(-X/2) = 5.0
using iterated plots on the screen. Use successively finer
scale numerical grids to improve your estimate to 3
significant digits. Verify the solution numerically.
Optional: It can be faster to solve the problem using the
WHERE function. Do so.
7) MAKING PLOT HARDCOPIES
Make a hardcopy of any one of your plots using a PostScript
To send plots to a PostScript file instead of your terminal,
type SET_PLOT,'PS. Then repeat the same plotting commands you
gave to put the plot on your terminal (easiest to use the
recall buffer). Close the file using the command
DEVICE,/CLOSE. The name of the file will be "idl.ps". Print
the file from within IDL by putting $ as a preface to the
appropriate UNIX command as the first entry on the command
line. Return to using your terminal as your output device by
giving SET_PLOT,'X .
Repeat, checking the status of your output device by using the
8) USING HISTOGRAMS AND "SAVE" FILES
I recommend you use the IDL Astronomy Users Library procedure
PLOTHIST for these exercises.
You can accomplish the same thing with the IDL built-in
HISTOGRAM routine and separate plotting commands, but this is
awkward. PLOTHIST is simply an accelerator program which
combines several built-in commands for convenience.
To check that PLOTHIST is in your IDL path, simply type
PLOTHIST. If present, the procedure responds by printing the
syntax for the command to your terminal screen. If not present,
you'll get an error message. [Note that this technique for
checking syntax does not work for IDL built-in functions, but
you can use the HELP,/ROU procedure or the HyperHelp facility
(type ?) instead there.]
The latest version of PLOTHIST requires software from the
"Coyote" library of graphics programs, written by David
Fanning. If your IDL installation does not include those,
download them from this site. Put the downloaded tar file in a
directory in your IDL path, then unzip and untar the file.
The programs will be automatically compiled when you request
Assuming PLOTHIST is in your path, then type: MAN,'PLOTHIST to
see the header of the program; type .RUN -T PLOTHIST to see the whole
Copy the file "grades.sav" from the IDLexercises/data directory
to your local directory. This is a specially-formatted file that
contains data from an earlier IDL session preserved by a "SAVE" command.
Using the RESTORE command, "restore" the dataset to your IDL
The save file contained the variable "GRADELIST". This
is a set of actual final point scores for a UVa course.
How many students were in the class?
Determine the MEAN, MEDIAN, MAXIMUM, MINIMUM, and standard
deviation (using VARIANCE) for the set of scores.
Use PLOTHIST to produce a histogram of the scores
The default bin size in PLOTHIST is always 1.0, regardless of
the range of the variable being plotted. Using the BIN
keyword, experiment with different bin sizes in the range 1
Graphically determine the mode of the scores and note how
it changes with binsize. What is the mode for a binsize of
Plot the score histogram for a binsize of 5 with labels
for the axes and the plot.
Overplot a vertical line at the location of the median
value. (Hint: use the PLOTS command)
Make a hardcopy of the final plot via a PostScript file.
9) SORTING & PRINTING LISTS; WRITING AN ASCII FILE
Continue using the "GRADELIST" vector of part (8)
Using the SORT function, define a new list (call it "SCOREORDER")
of scores in order from lowest to highest value. Verify.
Now define and verify a new list (call it "RANKSCORE") of
scores in order from highest to lowest value. (Hint: use
Now define an auxiliary vector ("RANK") which gives the
rank of each student in the class (assigning 1 to the
student with the highest grade)
Using a (one-line) FOR loop, an appropriately defined format
string, and the PRINT command, print on your terminal the
rank and score of the students in order from highest to
Hint: the definition for an IDL format string might look like
the following. (IDL format statements are similar, but not
identical, to those in FORTRAN).
and the command to print would look like this:
FOR I=0,MAXI DO PRINT,FORMAT=FORMOUT,VAR1[I],VAR2[I]....
Then, use the AstUseLib utility FORPRINT to do the same thing
Use your rank-listing capability to determine the median of
the grade distribution. Does it agree with the value
returned by MEDIAN?
Now assign letter grades to the students.
Assume you are a tough grader and use a strict curve where
the top 10% of the class gets an "A", the next 10% gets a "B"
and so on. The lower 60% of the class will get an F in this
scheme (appropriate, e.g., for a pre-Med class).
Using STRARR, define a string array named "LETTER" to hold
the letter grades for the students. Fill it with blanks
Consult the ranked list of scores from your terminal
printouts and determine the letter grade break points.
Then assign values to the "LETTER" array using those break
points. [Hint: use the WHERE function.] "LETTER" should be
in the order of the original "GRADELIST"
Verify the assignments by using a FOR statement and a new format
definition to print on your terminal a list of rankings,
numerical scores, and letter grades in ranked order.
WRITE AND VERIFY AN ASCII FILE
Now create an output ASCII file on your disk and output to it, in
the original (not ranked) order: the numerical grade, the
class rank, and the final letter grade for each student.
[Hint: You already have defined an appropriate format
string. The other commands you will need are GET_LUN, OPENW,
PRINTF, and CLOSE.]
Verify your results (e.g. use $MORE).
Now reverse the process and read the disk file back into your IDL
session, using variables with different names. [Hint: You
will need GET_LUN, OPENR, READF, and CLOSE.] Verify the
results by comparing the variables.
10) ASTROPHYSICS PLOTTING PROBLEM #1
Make a plot of the energy distribution functions B_nu, B_lam, and
N_lam, which are the flux per unit frequency, the flux per unit
wavelength, and the number of photons per unit wavelength,
respectively, for a black body (Planck law) at a temperature of
All three functions should be displayed on a convenient scale on
the same plot; each function should be normalized to its
maximum. Plot against wavelength in Angstroms, covering the
range 1000 to 10000 A. Label the plot properly. Make
a hardcopy of the plot and save the resulting PostScript file.
If you want to see a sample solution (J. Oishi, Fall 2001;
PostScript file), click here.
Optional: write an IDL "procedure" which will compute any of the
three versions of the Planck function listed above for a given
wavelength vector and temperature. Put it in your IDL directory
(and give it a unique name, like "plot_a_planck.pro"). Use a
"keyword" input parameter to select between the three functions.
To verify performance and documentation, let someone else try to
use your procedure.
You might want to compare your program to "planck.pro" in the
Astronomy User's Library.
11) ASTROPHYSICS PLOTTING PROBLEM #2
Explore the sensitivity of optical band "colors" to the
temperature of a black body as follows. Compute the Planck
function B_lam for selected temperatures in the range 1000 K to
100000 K for wavelengths of 1500, 3600, 4400, 5500, 10000, and
22000 Angstroms. (It's up to you to choose appropriate intervals
for the T grid.) Compute "colors" with respect to 5500 A
in the form
ALOG10( B_LAM[I] / B_LAM )
for each T. Plot these against T or LOG T on the same plot.
Label the axes. Label each curve (see the XYOUTS procedure).
Make a hardcopy of the plot and save the resulting PostScript
file. Consider the usefulness of the various combinations. Which
is most sensitive over the T range 3000-30000K appropriate for
IDL supports numerical integration of analytic or tabular
functions in one, two and three dimensions. Intrinsic IDL
routines performing integration of analytic functions of a single
variable include QROMB and QSIMP. The latter uses Simpson's Rule.
These routines all use the same, somewhat awkward, method for
defining the integrand. The user must place the name (i.e. an
IDL string) of an IDL function defining the integrand in the
calling sequence of the integration routine.
In this example, we illustrate integration of a simple power
law. We use an updated version of the intrinsic QSIMP routine
that is included in the Astronomy User's Library. Although its
Library name is also QSIMP, to avoid confusion we have renamed it
NEWSIMP and placed a copy in the IDL Exercises directory.
NEWSIMP uses the same basic algorithm as QSIMP, but it is faster
than intrinsic QSIMP because it takes a vector, rather than
scalar, argument. Also, it takes advantage of "keyword
inheritance" to provide additional arguments for the integrand
function. This eliminates the need to make up a new integrand
file for each change in the functional form of the integrand.
Copy the following programs to your local directory:
To see the coding and instructions for use of NEWSIMP, type:
.run -t newsimp
To see the integrand function for a power law, f(x) = x^p,
.run -t xpwr
To perform an integration with power law index 3 between the
bounds 0 and 10, type:
newsimp,'xpwr',0,10,answer,pwr=3 & print,answer
Experiment with other values for the range and power law index,
and verify that the integrations are correct.
[Note that since you are dealing with filenames here, you must
give them exactly as they appear in the UNIX directories.]
END OF IDL EXERCISES PART I
Part II covers 2D arrays and image displays.
Part III covers image processing.
IDL Tutorial Home Page
Translations of this page:
Copyright © 2000-2015 Robert W. O'Connell. All rights reserved.
Last modified by RWO, January 2015