We estimate this tutorial will take 20–30 minutes. That includes time for a bit of playing around with the commands.
If you have installed MATLAB on your own computer you start it the way you would for any other application.
Athena is MIT’s UNIX-based computing environment. OCW does not provide access to this environment.
(The ampersand (&) after ‘matlab’ allows the xterm to be used for other things, it is not necessary.)
Wait for MATLAB to start. This can take a few minutes. When it’s ready you should see a MATLAB command prompt: >
If you have a question ask one of the people seated next to you. There’s a good chance they know the answer.
Starting the desktop: If you have a window with several sections then the desktop started automatically. If you have a simple command line window with the MATLAB command prompt you should start the desktop with the command
> desktop [return]
% This is an edited transcript of a MATLAB session.
% We've inserted comment lines, which begin with a '%'.
Command lines begin with a ‘>’. When copying a command you should enter everything after the ‘>’ and hit return. MATLAB’s response will be shown in the line or lines below the command.
Trick: you can use the up arrow to find previous commands. If you type the first letter of the command then the up arrow will only show previous commands that start with that letter.
% The basic operations are *, +, -, /, ^. Try the following:
> 2+3
ans = 5
> 2*3
ans = 6
> 2/3
ans = 0.66667
> 2^3
ans = 8
> 2*(3+1)^2
ans = 32
% You can store results in variables and use them in calculations. Try the following:
> x = 2+3
x = 5
> x
x = 5
> y = 1+2
y = 3
> x*y
ans = 15
> z = x^y
z = 125
% If you don't name an expression MATLAB names it _ans_ for you. This is useful if you forget to name something.
> 7+15
ans = 22
> ans
ans = 22
> x = ans
x = 22
% Of course MATLAB changes the meaning of _ans_ the next time you don't name an expression.
> 7+15
ans = 22
> ans
ans = 22
> 9 + 2
ans = 11
> x = ans
x = 11
MATLAB has all the functions you know and love.
% We'll start with functions on numbers.
> sin(1)
ans = 0.84147
> sin(1.4)
ans = 0.98545
> sin(3)
ans = 0.14112
% MATLAB knows about pi.
> pi
ans = 3.1416
> sin(pi/2)
ans = 1
% The exponential function is given by 'exp'.
> exp(0)
ans = 1
> exp(1)
ans = 2.7183
% An array is the same thing as a matrix, i.e. a table of numbers. (MATLAB is short for Matrix Laboratory.)
% You create arrays by using square brackets. You put commas between the entries in rows and semicolons between the rows.
% 1 x 4 array
> [1, 2, 3, 4]
ans =
1 2 3 4
% 2 x 3 array
> [1, 2, 3; 4, 5, 6]
ans =
1 2 3
4 5 6
% You can store arrays in variables.
> x = [1, 2, 3, 4]
x =
1 2 3 4
% You turn rows into columns and vice-versa with a single quote after the bracket.
> x = [1, 2, 3, 4]
x =
1
2
3
4
% Turning rows into columns is called transposing. In MATLAB we use a single quote. In mathematics we use a T, e.g. _A_T. You read this as 'A transpose'.
> y = [1, 2, 3; 4, 5, 6]
y =
1 2 3
4 5 6
> y = [1, 2, 3; 4, 5, 6]
y =
1 4
2 5
3 6
% If the entries are simple enough you can use spaces instead of commas between entries in a row. (You still need a semicolon between rows.)
> x = [1 2 3; 4 5 6]
x =
1 2 3
4 5 6
% Arrays are accessed with the notation A(row,col).
> A = [1, 2, 3; 4, 5, 6]
A =
1 2 3
4 5 6
> A(1,1)
ans = 1
> A(2,1)
ans = 4
> A(2,3)
ans = 6
% For one dimensional arrays you only need one index.
> x = [1, 2, 3, 4, 5]
x =
1 2 3 4 5 6
> x(1)
ans = 1
> x(4)
ans = 4
% You can add or subtract matrices (arrays) of _the same size._
> x = [1, 2, 3]
x =
1 2 3
> y = [4, 5, 6]
y =
4 5 6
> x + y
ans =
5 7 9
> x = [1 2; 3 4]
x =
1 2
3 4
> x + [6, 5; 1, 2]
ans =
7 7
4 6
% You can scale or divide matrices by a number.
> x = [1, 2, 3]
x =
1 2 3
> 2*x
ans =
2 4 6
> x/2
ans =
0.50000 1.00000 1.50000
% You can multiply compatibly sized matrices.
% 3x2 times 2x4 is 3x4.
> A = [1 2; 3 4; 5 6]
A =
1 2
3 4
5 6
> B = [1 2 3 4; 5 6 7 8]
B =
1 2 3 4
5 6 7 8
> A*B
ans =
11 14 17 20
23 30 37 44
35 46 57 68
% If the sizes aren't compatible you get an error message.
> A*A
Error using *
Inner matrix dimensions must agree.
% You can raise square matrices to powers.
% Try the following:
> A = [6, 5; 1, 2]
A =
6 5
1 2
> A^2
ans =
41 40
8 9
> A^3
ans =
286 285
57 58
% Given how much we love exponentials it's nice that MATLAB can compute the exponential of a square matrix.
> exp(A)
ans =
914.31 911.60
182.32 185.04
% See what happens if the matrix is not square.
> B = [1, 2, 3; 4, 5, 6]
B =
1 2 3
4 5 6
> B^2
% My version of MATLAB produces the following error message.
Error using ^
Inputs must be a scalar and a square matrix.
To compute elementwise POWER, use POWER (.^) instead.
Sometimes you want to raise each element of an array to a power. The notation is a bit odd: it requires a period before the caret.
% Notice the '.' before the '^'.
> x = [1 2 3; 4 5 6]
x =
1 2 3
4 5 6
> x.^2
ans =
1 4 9
16 25 36
> x.^0.5
ans =
1.0000 1.4142 1.7321
2.0000 2.2361 2.4495
% The same notation works to multiply each element of an array by the same element in another array.
> [1, 2, 3; 4, 5, 6] .* [2, 4, 6; 8, 10, 12]
> ans =
2 8 18
32 50 72
% Also to raise each element of an array to the same element in another array.
> [2, 2, 2, 2, 2] .^ [1, 2, 3, 4, 5]
ans =
2 4 8 16 32
% Example: find the sum of the squares of the integers from 1 to 1024.
> x = [1:1024];
> sum(x.^2)
ans = 358438400
% This can be done in one command.
> sum([1:1024].^2)
ans = 358438400
Here we will learn to find the determinant and inverse of a square array and the reduced row echelon form and rank of any array.
% Check each of the following computations.
> A = [6 5; 1 2]
A =
6 5
1 2
% Determinant
> det(A)
ans = 7
% Inverse
> inv(A)
ans =
0.2857 -0.7143
-0.1429 0.8571
% Another way to find inverses.
A^(-1)
ans =
0.2857 -0.7143
-0.1429 0.8571
% Solving systems. Example: Solve the system:
% 6x + 5y = 4
% x + 2y = 7
% We solve this using matrix methods:
% The coefficient matrix is our matrix A. Note the prime on the array in the next command.
> solution = A^(-1)*[4, 7]
ans =
-3.8571
5.4286
% Solving systems is so important MATLAB has another, better way of doing this.
> solution = A\[4 7]
ans =
-3.8571
5.4286
% Reduced row echelon form
> rref(A)
ans =
1 0
0 1
% That was too simple. Here's another example.
> B = [1 2 3; 4 5 6; 7 8 9]
B =
1 2 3
4 5 6
7 8 9
> rref(B)
ans =
1 0 -1
0 1 2
0 0 0
% Rank of a matrix
> rank(A)
ans = 2
> rank(B)
ans = 2
% [3:1:10] is the array from 3 to 10 counting by 1.
> x = [3:1:10]
x =
3 4 5 6 7 8 9 10
% Changing the 1 to 2 means count by 2's.
> x = [3:2:10]
x =
3 5 7 9
% You can count by any number.
> x = [3:1.8:10]
x =
3.0000 4.8000 6.6000 8.4000
% For counting by 1's, you can leave out the middle number.
> x = [3:10]
x =
3 4 5 6 7 8 9 10
% For large arrays you'll want to suppress output to the screen.
% You do this by putting a semicolon at the end of the line.
% We'll demonstrate on small arrays so you can see what happens.
% No semicolon (have output)
> x = [1:2:9]
x =
1 3 5 7 9
% With semicolon (no output)
> x = [1:2:9];
> y = 2*x;
> z = 2*x % no semicolon, yes output
z =
2 6 10 14 18
% If you forget the semicolon you can stop the screen output by typing a q followed by return. Try this with the following command:
> [1:1000]
Most functions can be used on arrays.
% We'll start with functions on numbers.
% Sin acts on arrays by taking the sin of each element.
> x = [1, 2, 3; 4, 5, 6]
x =
1 2 3
4 5 6
> sin(x)
ans =
0.84147 0.90930 0.14112
-0.75680 -0.95892 -0.27942
% exp also acts on arrays.
> exp(x)
ans =
2.7183 7.3891 20.0855
54.5982 148.4132 403.4288
We don’t need it just yet, but the sum function will be quite useful.
% For a single row or column _sum_ adds them up.
> x = [1,2,3];
x =
1 2 3
> sum(x)
ans = 6
> y = x
y =
1
2
3
> sum(y)
ans = 6
% Find the sum of the first 1000 integers.
> sum([1:1000])
ans = 500500
% Find the sum of the squares of the first 1000 integers.
> sum([1:1000].^2)
ans = 333833500
% For a two dimensional array _sum_ produces the sum of each of the columns.
> x = [1, 2, 3, 4, 5; 7, 9, 11, 13, 15]
x =
1 2 3 4 5
7 9 11 13 15
> sum(x)
ans =
8 11 14 17 20
% Note the result of summing a 2x5 array is a 1x5 array.
% The basic plot command is plot(x,y). Try the following:
>> x = [0:.1;20];
>> plot(x,sin(x))
% MATLAB use _properties_ to style and decorate plots. Two of the most useful properties are Color and LineWidth. Here are some examples you can try.
>> x = [0:.1:6*pi];
>> plot(x,sin(x), 'Color', 'red', 'LineWidth', 3)
>> plot(x,cos(x), 'Color', 'blue', 'LineWidth', 3)
% Often we want to include more than one plot on the same axes. This is set up using the command hold on. 'Hold on' holds the plots on the graph when the next one is drawn. 'Hold off' turns this off and the next plot command will erase all the graphs before drawing the next one. Run these commands one at a time:
>> x = [0:.1:6*pi];
>> plot(x,sin(x), 'Color', 'red', 'LineWidth', 3)
>> hold on
>> plot(x,cos(x), 'Color', 'blue', 'LineWidth', 3)
>> plot(x,(x/20).^2, 'Color', 'magenta', 'LineWidth', 3)
>> hold off
% With hold now set to off the next plot will erase all the previous ones before drawing.
>> plot(x,exp(-(x/20).^2), 'Color', 'black', 'LineWidth', 3)
% Colors can also be specified as an array of 3 numbers between 0 and 1. The triple is a red, green, blue triple. For example
>> hold off
% Blue
>> plot(x,sin(x), 'Color', [0.0, 0.0, 1.0], 'LineWidth', 3)
>> hold on
% Orange
>> plot(x,cos(x), 'Color', [1.0, 0.625, 0.0], 'LineWidth', 3)
>> hold off
% To plot _z = f(x,y)_ you must specify the grid _(xi, yj)_ of lattice points to evaluate the function over. We do this by giving the _x_ vector and _y_ vector and using the MATLAB command meshgrid.
% Here is an example.
% First make vectors with spacing 0.1, over the interval [-2.2] for both axes.
>> x = [-2: 0.1: 2];
>> y = [-2: 0.1: 2];
% Then use meshgrid to make a grid of points _(xi, yj)_.
>> [x, y] = meshgrid(x, y);
% Next we plot the function _f(x,y) = y2 - 2x2_ over the rectangle [-2,2] x [-2,2].
(Note the dots for array operations)
>> z = y.^2 - 2*x.^2;
% We can plot a mesh of lines.
>> mesh(x,y,z)
% We can plot a filled in surface.
>> surf(x,y,z)
% In most installations you can manipulate the plot with the mouse. This can also be done with MATLAB commands.
% A 2D plot with 20 level curves.
>> contour(x,y,z,20)
% Or a 3D plot with 20 contour curves.
>> contour3(x,y,z,20)
This is just the tip of the plotting iceberg. MATLAB allows for fine-grained control over most aspects of a plot, including axis labels, text, font size and style, axis …
We estimate this tutorial will take 10 minutes. That includes time for a bit of playing around with the commands.
% This is an edited transcript of a MATLAB session.
% We've inserted comment lines, which begin with a '%'.
Command lines begin with a ‘>’. When copying a command you should enter everything after the ‘>’ and hit return. MATLAB’s response will be shown in the line or lines below the command.
Trick: you can use the up arrow to find previous commands. If you type the first letter of the command then the up arrow will only show previous commands that start with that letter.
Given a square matrix the eig function returns eigenstuff.
% Our favorite 2 x 2 matrix
> A = [6 5; 1 2]
A =
6 5
1 2
% The eig function returns the eigenvalues:
> eig(A)
ans =
7
1
% To get the eigenvectors and eigenvalues use the following syntax:
> [S,D] = eig(A)
S =
0.98058 -0.70711
0.19612 0.70711
D =
Diagonal Matrix
7 0
0 1
% Matlab put the eigenvectors as the columns of S and the eigenvalues in the diagonal matrix D. Notice that S is not guaranteed to have nice (to humans) columns like [5 1]' or [1 -1]'
% You can check the diagonalization formula:
> S*D*S^(-1)
ans =
6 5
1 2
