lecture tuesday, 4/18

[note: sorry about the awful accoustics at the 11:15 lecture.  i'll
look into why the mic didn't work....]

intro inheritance -- a powerful tool for code reuse to reduce redundancy

example: class Face to draw a face -- outline, eyes, mouth.

now, suppose wish to make modifications: frown, colored eyes, etc.

goal: be able to make changes while reusing old, unchanged code

+ define new *subclass* that $extends$ old *superclass*
+ by default, *inherit* all of the old code unchanged from superclass
+ *override* old methods to change old behavior
+ add new methods and variables to add new functionality

application: 
+ make class Face "expose decisions" in the form of methods that can be overridden
+ define subclass Frown to override mouth method

note: complete code will go to online examples

class Face {
    void paint(Graphics g) {
        drawOutline(g);
	drawEyes(g);
	drawMouth(g);
    }
    void drawOutline(Graphics g) { ... }
    void drawEyes(Graphics g) { ... }
    void drawMouth(Graphics g) { ... }
}

class Frown extends Face {
    void drawMouth(Graphics g) { ... } // new implementation
}

class ColorFrown extends Face {
    Color c;
    // constructor ...

    void drawEyes(Graphics g) {
        Color save = g.getColor(); 
	g.setColor(c); 
	super.drawEyes(); // use original implementation
	g.setColor(save);
    }
}

partial critique: 
+ if have 3 choices of mouths, 4 choices of eyes, 2 choices of outline, 
  then this approach needs 3*4*2 = 24 classes, with *lots* of repeated code!
+ solution: Face = 1 Outline object + 2 Eye objects + 1 Mouth object;
  subclass Outline, Eye, Mouth -- therefore, have only about 3+4+2=7 classes.

next lecture: making inheritance more precise, 
e.g. box/scope diagrams to explain scope.
_______________________________________________________________________________

matlab

need password -- don't distribute, but *do* try it out and let us know
of any problems.

    username: cs100a  (not case-sensitive)
    password: ***censored*** (case-sensitive) <-- given out at prelim 3

exercise 7, due thursday 4/20: 
+ redo Project 1, Part 2 in matlab
+ hand in matlab session with both your code and resulting (interleaved) output

why matlab?

+ cs100 = "intro to programming, not intro to java " -- to see that you learned
  general, universal programming principles, need to see another language

+ very useful for matrix and processing scientific data,
  e.g. linear algebra and plotting/graphics

+ matlab is slightly different way of thinking, but not too different, 
  so not too hard to learn

+ matlab has pretty good online (built-in, not web) help

+ matlab can be much more concise than java, but not so much that it is
  unreadable (people joke that APL is "WORN", "write once, read never")

+ matlab is interactive: you can choose code as you see answers
_______________________________________________________________________________

matlab notes as we go through examples:

+ $%$ is like $//$ in java: comment out rest of the line

+ no declarations -- just initialize and use variables

+ basic type is an array, which *is* a value -- there are no references
  + matrix is 2-D array (careful: 1-D *is* "sometimes" special case of 2-D)
  + vector is 1-D array; row vector versus column vector
  + scalar (number) is 1-by-1 array

+ matlab tends to operate on entire arrays

+ matlab prints out results unless you end the statement with $;$

+ $rand$ returns random scalar; $rand(height,width)$ random matrix

+ WARNING: use $length$ for only vectors, not 2-D arrays

+ WARNING: matlab uses 1-relative indexing (java uses 0-relative)

+ $sum$: sum of row or col vector; column sums of matrix <-- argh, 1-d/2-d?!

+ since matlab uses $%$ for comments, it uses $rem$ for remainder

+ matlab uses 0 for false, 1 for true, so $sum$ counts number of $true$s

+ can easily extract out a whole sub-array, e.g. $a(1:n-1)$

+ matlab uses $~=$ instead of $!=$
_______________________________________________________________________________


% create 1-d array a with 50 integers from -100..100

    a = floor(rand(1,50) * 201) - 100

% plot a

    plot(a)

% T1.Q1: RMS of 1-d array a

    sqrt( sum(a .^ 2) / length(a) )

% project 2.2: count multiples of 5 in 1-d array a

    sum(rem(a,5) == 0)

% project 4.1: roll pair of 6-sided dice 1000 times; count 7s

    sum(7 == sum(1+floor(rand(2,1000) * 6)))

% T1.Q2: number of days in random $month$

    month = 1 + floor(rand * 12)
           % 1  2  3  4  5  6  7  8  9 10 11 31
    days = [31 28 31 30 31 30 31 31 30 31 30 12];
    days(month)

% T1.Q3: max pair sum in 1-d array a

    n = length(a); max(a(1:n-1) + a(2:n))

% project 2.3: count sign changes in 1-d array a

    n = length(a); % $;$ supresses display of the result
    signs = sign(a); % sign of each element of a
    sum(signs(1:n-1) ~= signs(2:n)) % note: matlab uses 1-relative indices

% project 2.4: find elevations with steepest gradient in 1-d array a

    [steepest,where] = max(abs(a(2:n) - a(1:n-1)));
    fprintf('steepest gradient %d between elevations %d and %d\n', ...
	steepest, a(where), a(where)+1)

    disp('steepest gradient'), disp(steepest) % etc.