GEOS 597e Spatiotemporal Data Analysis Workshop
Homework 0:  Get ready, get set

Get set up for computing in a Matlab/PC environment, from the CLUE Lab's PC workstations.

Last updated 8/23/06. To be completed prior to class session Weds., Sept. 30th.  

0.  Introduce yourself.  Send me email from the account you would me to use for course communications.  The email should have subject line  "STDA: [your name]" with the following information in the body of the message:

0.1.  Your name, and how you like to be called.
0.2.  Your department, degree sought, major, and minor.
0.3.  Your thesis topic.
0.4.  Your familiarity with Windows and Matlab, each on a scale of 1 (none) to 10 (fluent).
0.5.  A list of courses taken (and when) in relevant subjects such as climate dynamics, linear algebra, geostatistics, and computer programming.
0.6.  Why you are taking this course, and what you hope to get out of it.

For example your email might look like:
From: wilma@geo.arizona.edu
Subject: STDA: Wilma Wildcat

Wilma Wildcat ("Wil")
Geosciences, Ph.D., Geosciences, Global Change.
Thesis topic: Paleoclimatology of the Football Stadium region of southern Arizona.
Familiarity with Windows: 9; with Matlab: 2.
Courses taken: Math 215 (Linear Algebra), Spring 2003; Atmo 595b (Climate Change),  Fall 2003.
Why I'm taking this course: I want to interpret my own datasets using the techniques covered in this class.

1.  Get CLUE lab access.  Get a swipe card, door combination, userid and password to access computers in the CLUE Lab, Geosciences Dept., Gould-Simpson Bldg. Rm. 228, by contacting Eneida Lima, CLUE Lab Manager, Gould-Simpson Rm. 542.  Eneida will be in class Weds. Aug 23rd, 1:30pm, to have you sign forms, etc. to do all this if you haven't already.  Read the CLUE Lab rules.

2.  Meet Matlab on Windows.  Log into a PC workstation of your choice.  Start Matlab by clicking on the desktop icon labeled "MATLAB R2006a".  After a long while, you should get the Matlab startup screen, which looks something like this screenshot from my linux box:

ml desktop

There are five important sections to the matlab desktop:

On the right is the command window, into which you (not surprisingly) can enter matlab commands. but also you can pass commands to Windows by preceding them with an exclamation point.  For instance, execute the following command at the prompt (>>):

>> !cd \"Documents and Settings"
>> !mkdir stda

This will create the directory stda in the named directory.  Now click on the button labelled ... just above the command window.  You can then change directories to the stda directory you just created.  Files you subsequently save from matlab will save here (and importantly, be available to you here) when you log back in next time. 

The window on lower left of the matlab desktop is the command history.  Notice that the commands you typed have appeared here.  This is helpful for keeping track of where you've been and what you've already done.  An even more important feature: notice that pressing the up arrow in the command window gives you the previous command you entered.  This is immensely useful for avoiding typing mistakes, one of the biggest sources of coding bugs.  Also, notice that typing the first few letters of a command, then the tab key, gives you a menu of possible word completions,  from which you can select by double clicking on one.  It will appear on the command line.

The upper left window shows the working directory structure, something like "Explore" when you right click on the Windows Start menu.  This helps you organize your work.  By clicking on the "Workspace" at the bottom of this window you can alternately view the variables and their dimensions in memory.  Speaking of variables, let's create one by typing in the command window

y=randn(1000,1);

What does this do?  To find out how a matlab command or function works, try typing

>> help command

replacing command with the matlab command or function of interest.  For instance:

>> help more

gives the following result:

>> help more
 MORE   Control paged output in command window.
    MORE OFF disables paging of the output in the MATLAB command window.
    MORE ON enables paging of the output in the MATLAB command window.
    MORE(N) specifies the size of the page to be N lines.
 
    When MORE is enabled and output is being paged, advance to the next
    line of output by hitting the RETURN key;  get the next page of
    output by hitting the spacebar. Press the "q" key to exit out
    of displaying the current item.
 
    Reference page in Help browser
       doc more
>>

Try it out: type more on, then ask for help on randn.  The purpose of the help command is to show you how a function is called (the syntax); what arguments it takes (the values or parameters in the parentheses on the right hand side of the command structure, and the output it returns (the variables in the left hand side of the function syntax. 

Now bring up the editor window:

>> edit test.m

You'll get a new window called test.m.  Let's make test.m into a script: a set of instructions for doing some work for us. In our script type the following commands:

clear
y=randn(100,1);
x=1901:2000;

You now know how to find out what each of these commands does, so go ahead and investigate these functions.  While you're at it, in the command window type

>> help /

to find out about how to add, subtract, multiply, divide, and compare relational values (relops).   (The helpdeskis a much fancier way to explore matlab's functions, including demonstrations, tutorials, and examples.  Check it out, especially if this is your first experience with matlab, but also to see what new features may have been added since you started working in the environment.)  For more about online help, including the very useful  lookfor command, see next week's prework reading, section 3.4.

You'll see the variables x and y appear in thw workspace as a vector with 100 elements, organized as a row vector (something we'll learn about next week in more detail).  In the command window you can type

>> x

to see the values in each  element of the row vector x; similarly, for y.   What happens when you type

>> x;

?  Use the semicolon in your script commands  to avoid typing the result of every last command out to the command window, which may slow your script down by a factor of ten or more.

To plot values we call up a figure window by adding to our script

plot(x,y,'o-')

and to save a copy of the plot we can type

print -deps f1.eps

or

print -djpeg100 f1.jpg

You can save a copy of the work your script has made either by adding

save testwork.mat

and/or

save -ascii xy.dat x y

to your script test.m, which should now read:

clear
y=randn(100,1);
x=1901:2000;
plot(x,y,'o-')
print -deps f1.eps
print -djpeg100 f1.jpg
save testwork.mat
save -ascii xy.dat x y

Save your script in the editor window.  Now in the command window type

>> test

to run your script.  This is essentially how you will develop your codes for EOF analysis and singular spectrum analysis over the semester, and to graphically display the results.

3. Comment my code.  Copy and paste the following script into an editor file; call it synt.m.   Your task is to familiarize yourself with the commands called in this script, and to comment the script to explain what each line does.  What does the code do overall?  Put this comment at the very top of the script.  Remember, to paraphrase a librarian friend of mine: good commenting is like sex: when it is good, it is very very good; and when it is bad, it is still better than nothing at all.  You can comment a line (or the remainder of a line) by starting it with the percent sign, %.  In your classwork scripts you won't need to comment every last basic matlab function call, but you will want to comment sections of your code to explain to yourself and to me what you intended to do.


% script synt.m: here is a line to replace with a real comment
clear  % this command clears all variables from memory when the script is called

nt=1000
r1=0.4;
t=1:nt;

e(1)=randn(1,1);
for i=2:10*nt
e(i)=r1.*e(i-1)+randn(1,1);
end
e=e(length(e)-999:length(e));
v1=var(e); v2=e*e'./(nt-1);
[v1 v2]

y1=randn(1,nt);

y2=sin(2*pi*t./20);

y3=t./1000;

y4=(t-mean(t)).^2/1000^2; y4=y4-mean(y4);

y5=ones(1,1000).*4;

y=y1+y2+y3+y4+e+y5;

subplot(211)
plot(t,y)
set(gca,'Fontname','Times','Fontsize',16)
     title('plot of Y vs. T')
     xlabel('time(years)')
     ylabel('amplitude (no units)')
     grid
axis([0 1000 -2 12])
     subplot(212)
plot(t,y)
set(gca,'Fontname','Times','Fontsize',16)
     title('plot of Y vs. T')
     xlabel('time(years)')
     ylabel('amplitude (no units)')
     grid
axis([0 100 -2 12])
print -deps synt.eps
print -djpeg90 synt.jpg
save syntwork.mat

4.  Deliverables.  Send me the email requested in part 0, and print out and turn in your commented script from part 3 of this classwork.
Back to Schedule/Syllabus.