Computer introduction for FYTA11

The purpose of this computer introduction is to give you a brief introduction to GNU/Linux, Emacs, and Java programming. You will gain some experience with a powerful computing environment used by professionals and enthusiasts alike. We give this introduction because a large part of the hand-in exercises for this course (and other courses given by Theoretical Physics) require you to do some independent work on a computer.

Overview

Linux

The operating systems distributed under names like Debian, Ubuntu, OpenSUSE or Fedora are based on the Linux kernel, which has drivers for many types of hardware and sets the stage for running programs in a modern Unix-like environment. GNU is a free software project which encompasses many of the programs needed to make a useful system, such as file system tools, text editors and compilers.

The text editor Emacs

Emacs is a GNU text editor which is usually distributed with Linux. It is incredibly powerful, which also means it can be a bit confusing to new users. Like with other text editors, its primary function is to let you edit the contents of text files. These files may e.g. be source code for programs, in which case Emacs may be able to assist you through syntax highlighting.

Gnuplot

Despite its name, gnuplot is not GNU software. It is, however, a program for plotting, both on the screen and to files of various formats.

A very brief introduction to programming in Java.

You will learn to write and compile a basic Java program that outputs some text.

LaTeX

Learn to write reports using LaTeX.

Linux

For many years we have used a text called Introduktion till Unix och X på EFD-datorerna for this computer introduction, but apart from the difficulty that that text is in Swedish, larger and larger portions of it have gradually become less and less relevant. I therefore prepared this new introduction to Linux a few years ago, and have been updating it regularly. I hope it still makes sense.

Note: I have marked literal strings like filenames and commands like this. Some computer terms that may need further explanation look like this.

Background

Unix is a family of operating systems derived from a common ancestor in the early 1970s. Linux and GNU are projects aimed at doing everything the Unices of old can do, except better and as free / open source software. Linux is the kernel, which knows about computer hardware, file systems and how to run programs and let them interact. The GNU project provides many programs, most in the form of small commands. The Unix way of doing things is to have one command for each simple task, and string commands together to perform more complex tasks. We will soon see some examples of this.

File system

Unlike on many other operating systems, everything resides in a single hierarchy of directories and files. For instance, the file /home/carl/public_html/index.html is my homepage.

/ (the root)
/home (home directories)
/home/carl (my files)
/home/carl/public_html (my web stuff)
/home/carl/public_html/index.html (my homepage)
/home/carl/src (my source code)
/bin (some programs)
/dev (devices: disks, partitions, ports, ...)

As you can see the root of this file system tree is called /, and / is also used to separate directory names in a path such as /usr/bin/man. (For someone with a background on other systems, it should perhaps be pointed out that there is no direct connection between the path to a file and where that file is stored physically. Different parts of the tree may reside on different storage media.) By convention, directories called bin (such as /bin, /usr/bin and /home/carl/bin) contain executable files or, in other words, programs. There are many similar conventions, but there is no point in learning them all now.

The shell

I will assume that you have already logged in and are currently sitting in front of a computer. Open a terminal window — this might be called something like shell, terminal or console depending on what system you’re using. With some luck you’re now looking at a mostly empty window with a prompt that looks something like this:

yourname@computer:~>

Good. What you’re seeing is a program — the shell — which wants you to make it do something. The window itself belongs to a program that emulates the computer terminals of old, in a nice demonstration of the principle of letting each program do one thing (managing the window versus being a shell). So what does a shell do?

A shell is, roughly speaking, a program that takes command names and runs the corresponding commands. Every process (instance of a running program), including the shell, has the notion of a current directory; where in the directory tree the process is at the moment. The current directory is usually shown in the shell’s prompt — in this case it is ~ (tilde), which is the shell’s way of saying your home directory. To see where that really is, try the pwd command:

carl@min:~> pwd
/home/carl

You can list the contents of the current directory (or any other directory!) with ls. To see a bit more than just the names of files and directories, try ls -l (that's dash ell). This long list format tells you whether things are files or directories, who owns them, how big they are, and a few more things.

To change the current directory, use the cd command, e.g., cd mydir. If you just type cd with no arguments, you will move to your home directory, just as if you typed cd ~. The parent of a directory is called .. so the command cd .. will move you one level up in the directory tree. And you may use more complicated paths; either absolute (beginning with /) as in cd /usr/bin or relative as in cd ../../usr/local.

One extremly useful feature of the shell is tab completion. If you start to type the name of a file or directory, press the tab key to have it automatically completed. If there is more than one possible match, press tab again to get a list of the matches (this is the normal behavior, but it can vary with your settings). Using the tab completion saves time and is a great way of ensuring that you don’t mistype filenames.

Another thing should be mentioned regarding the shell: there is a subtle difference between commands that are built into the shell, such as cd, and those that are actual programs, such as pwd and ls. If cd were a program, it would be started by the shell, then this new process would change its current directory, but it would then terminate and leave the shell just like it was before. Because processes cannot manipulate each other, commands like cd are part of the shell itself.

A few commands

You have seen pwd, cd and ls. You can have more fun with these if you use mkdir to create new directories, e.g., mkdir mydir or mkdir mydir/mysubdir. You can also remove files and directories with rm. For instance, rm foo removes the file named foo. However, to remove the entire directory mydir (and all its contents) you need to use rm -r mydir. You can also use rmdir mydir but only if mydir is already empty. Be very careful with rm -r, or you might end up removing too much. We’ve all been there, and there is no easy way to recover deleted files.

The command man brings up manual pages, so if you ever wonder how, say, the command ls works, just type man ls and you will learn. There is also the newer info command which is slowly replacing man and which in some cases gives you more information. For those commands built into the shell, there is the help command. (Assuming that you’re using the bash shell. Other shells exist, but let’s ignore them.) Of these three, man is the one you will use the most. Get used to looking up things in the manual — it’s much easier than trying to memorize all commands and their options.

>, < and |

Try typing echo Hello. You see what echo does. Now try echo >myfile Hello. Where did it go? Does ls give you a hint? Try less myfile. To quit less, press q. (If you press something else and get stuck in a strange state, try pressing Esc a few times and type :q.)

So adding >filename to a command makes the output of the command go to that file. Similarly, adding <filename makes the program read input from the specified file. You can try it with wc <myfile to see how many lines, words and characters there are in your file.

What really happens here is that the shell opens a file for writing (>) or reading (<) and passes it to the command’s process. Many commands will read from this standard input only if not given any filenames as arguments. Thus wc myfile does the same thing as the example above, albeit in a subtly different way.

More special is |, the pipe sign, which makes the output from one program the input to another. This allows arbitrary commands to be connected in useful ways. An instructive but rather silly example is echo Hello world | wc.

Even more commands

You have already seen less, which is used to view the contents of text files. There is also an older program called more which does the same thing, but is less capable. A simpler and somewhat different command is cat, which merely outputs the entire contents of a file at once. It can be useful for viewing small files but also for merging or duplicating files, as in this example: cat myfile myfile myfile >triplefile.

The mv command is used to rename a file or directory, e.g., mv oldname newname. To copy a file use cp, as in cp fromfile tofile. If you try to copy an entire directory you will find that cp does not want to do this without the -r option: cp -r mydir mynewdir. Note also that if the destination (here: mynewdir) is an existing directory, cp creates a copy in that directory (here it would be mynewdir/mydir).

At this point a note about options is due. Most, if not all, commands accept a multitude of options in a short (e.g., -r) or long (e.g., --recursive) form. Most of these options affect details of how the commands operate, and you can read all about the options on the manual pages for the different commands. The --help option is accepted by most commands, and will print a short help text that is often more useful than the full manual page.

Traditionally the options had to be given before any additional arguments, as in cp -r foo bar, but the modern GNU commands accept options anywhere in the command line, as in cp foo bar -r. Because options may appear anywhere, it could be difficult to deal with files whose name start with a -, and the -- option therefore can be used to turn off option parsing. So rm -- -z is the way to remove a file called -z.

Searching in files

Create a file with a few lines of text in it. One way to do this is to type cat >myfile, and then enter a few lines of text, followed by ctrl-d (which signifies end-of-file). Now myfile should contain the text you entered. Try grep word myfile. Replace word with some word that is in the text you entered. What you should see is that grep is a tool for searching for text in files. It can be used to search for complicated patterns, as in the example grep 'h[ae]llo' myfile (a regular expression that matches both hallo and hello). grep -i word myfile performs a case-insensitive search in myfile, so both word and WoRd will match. grep word * searches for word in all files in the current directory. Similarly, grep -r word . will search for word in the current directory (.) and recurse into all subdirectories and files.

grep may default to case-insensitive searches, depending on the current language setting. One way to force a case-sensitive search is this: LANG=C grep word myfile.

It is really the shell itself that expands * into the names of all files. So echo * will print the names of all files in the current directory, ls -l ../*.txt will list all files in the parent directory whose names end in .txt, and so forth. That the shell performs this expansion (and more!) also means that arguments with funny characters in them need to be quoted in "" or ''. For example, echo '*' prints a single *.

Searching for files

There is a command called locate, which can be used to locate a file if you know its name. As a demonstration, locate resolv.conf should find /etc/resolv.conf and a few other files, assuming that locate has been installed (it might not be). The database which locate uses to find files quickly is updated regularly, but very new files will not be found. It is then better to use the more powerful find. find . -name 'myfile*' finds all files whose names start with myfile in the current directory.

Concluding remarks

You have probably already received more information than you can digest in one go. If you’re going to remember something, remember that manual pages are your friend. info coreutils brings up a list of many of the most common commands that you may need.

Exercises

• Create a directory called myfiles.
• In myfiles, create two files called foo and bar. The file foo should contain the text string This is a test and bar should contain This is not a TEST. Now, make sure you’re in the directory myfiles, so that a directory listing shows the two files.
• Where in the directory tree are you now?
• Find all files that contain the word test.
• Find all files that contain the word test regardless of whether it’s in lowercase or UPPERCASE.
• How big are the files?
• What does the command wc do?
• How many words do foo and bar contain?
• How much free disk space is there (hint: man df)?
• Rename bar to baz and remove foo.
• Go up one level (that is, leave myfiles) and then remove myfiles and its contents.
• How much free disk space is there now?

Useful things to know

• Some commands that you should definitely know: cd, pwd, ls, mkdir, rm, less, mv, cp, man.
• Some commands that are very useful: grep, du, df, cat, echo, info.
• Some commands that are useful too: locate, find, sort, cut, head, tail.
• Some networking tools: ssh, scp, wget, unison
• Some file viewers: gv (for postscript, .ps / .eps), acroread / okular / kpdf / evince (for PDFs), less (for text files), xv / ee / display / eog (for images)

Emacs

Emacs is a text editor that some people like for its flexibility. It is exteremly powerful and customizable, and can in fact be used for just about any purpose, including editing text.

Start Emacs from the shell by typing

emacs

or

emacs files_to_edit

possibly followed by an & (to have Emacs run in the background). Emacs will open a window of its own, and you can go to work. Basic editing such as removing and inserting text or moving the cursor around should need no further instructions. Operations such as copying blocks of text or searching and replacing can be accessed through the menus, where keyboard shortcuts for those operations are also listed. To some extent the mouse can be used for editing too. Note that the way to copy and paste text is not the same as on certain other systems, and in particular ctrl-c does not copy text. Saving, loading and quitting are other operations that can be accessed through the menus, or you could learn their (somewhat cumbersome) keyboard shortcuts: ctrl-x ctrl-s to save, ctrl-x ctrl-f to load, and ctrl-x ctrl-c to quit. Another extremely useful shortcut is esc esc esc (or ctrl-g) to escape without harm if you have accidentally started to enter a command.

One very powerful aspect of emacs is its knowledge of the syntax of different programming languages. When you edit a Java file, you will notice that emacs does syntax highlighting, possibly making it easier to read the code. Emacs will also help you indent the code properly – press tab to have the current line indented according to emacs's rules.

Gnuplot

gnuplot is a useful and popular program for generating plots. In this section we will go over some basic gnuplot commands. Feel free to test more advanced stuff on your own afterwards. Type gnuplot in a terminal window, and you should see a copyright notice followed by a gnuplot prompt, like this one:

gnuplot>

Plot

At the prompt, type

plot x * 5 + 1
plot sin(x)

Try the following:

show xrange
show yrange

As you can see, xrange and yrange control the area to plot. These values can easily be changed. For instance,

set xrange [-2:6]

The plot is not automatically updated, so you’ll have to issue another plot command:

plot sin(x)

A range of * tells gnuplot to try to figure out a reasonable value for every plot command. To restore the x range to this behavior, you can type

set xrange [*:*]

It’s also possible (and usually more convenient) to set the range directly in the plot command:

plot [-5:15][-2:1.5] sin(x)

There is a replot command, which will let you add more plots in an existing window:

replot cos(x)

Feel free to play around with different plot commands. You could for instance try these:

plot sin(x) * 5*sin(20*x), 5*sin(x), -5*sin(x)
plot [0:10] x**1.5 title "up", x**-1.5 title "down"
plot exp(-x*x)

3D Plots

It’s possible to plot in three dimensions, using the splot command:

splot [-10:10][-10:10][:] x*x + y*y

Plotting data from a file

It is of course also possible to plot data from a text file. Use Emacs to create a file called data.dat and enter the following table into it.

1 1.21 2.22
2 1.10 2.50
3 1.05 2.75
4 1.30 2.66
5 1.26 2.88


To plot the data from this file, you can type

plot "data.dat" using 1:3 with linespoints

This will create a plot where x coordinates are taken from column 1 and y coordinates from column 3. If you don’t specify the columns (i.e., you remove using 1:3), the first columns will be used.

linespoints means that the data points will be plotted as points connected with lines. Some other useful styles are lines, points, dots and yerrorbars. With yerrorbars you need to have a column with error values, and the using clause should specify three columns (e.g. using 1:3:5). To set the plotting style for all subsequent plot commands, use set style data and/or set style function, e.g.:

set style data linespoints

In older versions of gnuplot, set style data was instead called set data style, and so on.

Title and labels

To give the diagram a title and set labels on the axes, you can type:

set title "Title goes here"
set xlabel "My x label"
set ylabel "My y label"

You can also give the functions/data titles:

plot sin(x) title "Sine", cos(x) title "Cosine"


Most commands and options can be shortened to just one or a few letters, as in plot "data.dat" u 1:3 w lp ti "data".

Log scale

Once in a while you may need to plot something with a logarithmic scale on either axis. For instance, if you want a logarithmic y axis:

set logscale y

And to set the scale back to linear:

unset logscale y

Or, with older versions of gnuplot:

set nologscale y

If you don’t specify an axis (e.g. set logscale), both axes will be affected.

Exercise: Try to fit an exponential function (ae^bx, written in gnuplot as a*exp(b*x)) to the data below, using the plot command.

0.75 1.85
1.25 2.98
1.75 4.23
2.25 6.57
2.75 10.3
3.25 16.5


Printing

To print a plot or include it in a report or presentation, you may want to save it to a file. The other option is to take a screenshot, but the low screen resolution makes that a bad option for plots that are to be printed. It is better then to generate a file in postscript (ps/eps) or PDF. This is done by telling gnuplot to output in postscript or PDF, rather than to the screen. Commands needed for this are:

set terminal postscript monochrome
set output "yourfilename.ps"

You will then have to re-run the plot command (or simply replot). The postscript plot will now be in the file yourfilename.ps. To print or view this file, open another terminal window. There, you can start the postscript viewer gv:

gv yourfilename.ps

It’s possible to print from gv, or you can do it directly from the shell using lpr yourfilename.ps. Depending on what computer you are using and what printer you want to use, a different command (or a -P argument to lpr) may be needed.

After generating a postscript file you’ll probably want to go back to plotting to the screen:

set terminal X11

To see other available output formats in your version of gnuplot, use help set term.

The contents of an output file are not necessarily written until the next set output or set terminal command. Also note that some formats, such as PNG images, only work properly with a single plot command per output file.

Getting help

gnuplot has a help command which lists all of the available commands and their options, and provides some useful examples. It may be a bit cumbersome to use, but once you get used to it it’s not that bad. To get a list of all help topics, type

help

If you want a list of the available commands:

help commands

If you want help with a specific command:

help plot
help set key
help set term post

There is also a meaty gnuplot manual available here.

Another exercise: Play around with gnuplot!

Java programming exercises

Here is a simple exercise to get you started on writing, compiling and running Java programs.

Hello World

Write a program that writes Hello World on the screen. Save the following lines into a file called HelloWorld.java

public class HelloWorld 
{
    public static void main(String[] args) 
    {
        System.out.println("Hello world!");
    }
}

Java programs are run by a program called the java virtual machine (JVM). The JVM can’t use the source code directly; first you need to compile the code with the Java compiler javac. Do this by typing the command

javac HelloWorld.java

The compiled program is now in a file called HelloWorld.class. You can now run the program with the command

java HelloWorld

which starts the JVM and tells it to run the compiled program found in HelloWorld.class

To further explore what you can do in Java, see the FYTA11 lecture notes.

LaTeX

LaTeX is a document preparation system that is useful when writing text with lots of equations. It is widely used by mathematicians and physicists for writing scientific text. This introduction will only show you how to get started. See the text Att skriva rapporter med LaTeX for many more the details.

The file Report-template.tex is a template LaTeX document that will serve as an example here. Download and save this file in a directory and open it with an editor, like emacs.

The first few lines of the document are headers specifying different settings. After that there are a few lines where you can enter the title of the document, etc. The document itself consists of the text between the tags \begin{Document} and \end{Document}. Each section begins by the line \section{SectionTitle}.

Enter some text into this document, change title, author and date, and type something in one of the sections. Then compile the document by typing the command

pdflatex Report-template.tex

in the directory where you saved the template. This creates a PDF document called Report-template.pdf which you can view with a PDF viewer such as kpdf (like this: kpdf Report-template.pdf).

You could also create a postscript file and then a PDF file through the following sequence of commands:

latex Report-template.tex
dvips Report-template.dvi
ps2pdf Report-template.ps

We recommend that you use this template when writing reports for the different exercises in this course. All reports should be handed in by sending a PDF file, also if you decide to write them using some other program (like MS Word or OpenOffice). The MiKTeX package is a Windows implementation of LaTeX that can be downloaded for free.

Links to manuals, tutorials and other stuff

gnuplot

• gnuplot demos and documentation

Linux

• Linux Knowledge Base and Tutorial
• The Linux Documentation Project Homepage
• LowFat Linux
• The Linux Journal
• Linux Online

Java

• The Sun Java Tutorials
• Java Platform, Standard Edition 6, API Specification

LaTeX

• Att skriva rapporter med LaTeX.