Skip to content

Octave / MATLAB - lops, conditional processing, functions

Octave is an open source scientific computing package. It has a GUI, kernel and a programming language modeled after MATLAB. File extensions are .m. This page is a guide to the octave programming language.

General caveats

  • You end statements with a ;. If not, the kernel does not throw an error, instead will echo the output of each line.
  • If you do not catch the output of an operation in a variable, a default variable called ans is created and that holds the value. At any time, ans holds value of latest operation.
  • Matlab / octave functions can return multiple values.

Variables

Variables can be assigned without initializing.

a = 1;
b = [1,2,4]; % row vector
% Use % for comments
c = [5;6;7]; % col vector

2D arrays can be initialized as 

>> a2d = [1,2,3;4,5,6;7,8,9] % ';' is used to separate rows.
a2d =
   1   2   3
   4   5   6
   7   8   9

Standard matrix creation functions

>> eye(3)  % identity
ans =
Diagonal Matrix
   1   0   0
   0   1   0
   0   0   1

>> zeros(2,3)
ans =
   0   0   0
   0   0   0

>> rand(2,5)  % uniform random numbers between 1,1
ans =
   0.86289   0.18309   0.48379   0.36600   0.63669
   0.32459   0.89961   0.79682   0.39961   0.15351

>> randn(3,4)
ans =
   1.151159   0.022623  -1.030988   0.460505
  -0.386885  -0.188437   0.339421  -1.131755
   0.573182   0.451552  -0.307362  -1.717639

>> magic(3)  % creates matrix whose each row, col, diagonal sum to same value
ans =
   8   1   6
   3   5   7
   4   9   2

Automatic map operations

Octave is designed to work primarily with arrays and vectors. Thus when you pass an array or vector to a standard function or operator, it automatically maps the function on each element and runs it and returns either an array/vector or scalar, depending on the operation.

>> squares = [4,9,16,25,36,49];
>> sqrt(squares) % returns a vector output
ans = 2   3   4   5   6   7

>> min(squares)
ans =  4  % scalar output

When you run functions on a 2D matrix, they run on each column vector:

>> sum(magic(3)) % default is columnwise operation
ans =
   15   15   15

>> sum(magic(3), 2)  % does a row-wise operation
ans =

   15
   15
   15

>> prod(magic(3))  % multiply each element in the matrix. Here multiplies each in col vector
ans =
   96   45   84

>> max(magic(3))
ans =
   8   9   7

Functions in Octave can return multiple values. For instance:

>> [val, index] = min(squares)
val =  4
index =  1
>>

Automatic broadcasting

When you operate a scalar against a vector, matlab will automatically scale up that scalar into a vector and perform an element wise computation as shown below:

>> x=-2:0.7:2
x = -2.0000   -1.3000   -0.6000    0.1000    0.8000    1.5000
>> x+10
ans = 8.0000     8.7000     9.4000    10.1000    10.8000    11.5000
>>

Similarly when you multiply, subtract or divide a scalar from a vector, it is automatically broadcasted.

Looping through a matrix

The automatic map operations should have most things covered for you. Still, you might need to write a loop. Below is syntax of for loop.

for iterator=<range>,
   operation1;
   operation2;
   condition check
      operation3;
      break;
   condition check2
      operation4;
      continue;
end;

Note the , at end of the line with for.

>> indices=1:10;
>> for i=indices,
       disp(i);
   end;
 1
 2
 3
 .
 .
 .

While loop

The syntax for while loop is

while condition,
   operation;
end;

Once again, note the , at line 1.

Array multiplication

There are two types of multiplication - element wise - which is accomplished by prefixing the operator with . such as: .<operator> or .* - matrix multiplication - which will happen when you use *.

Thus:

>> x
x = -2.0000   -1.3000   -0.6000    0.1000    0.8000    1.5000

>> x2 = [1,2,3,4,5,6]
x2 = 1   2   3   4   5   6

>> x.*x2
ans = -2.00000  -2.60000  -1.80000   0.40000   4.00000   9.00000
>>

When you want to multiply two matrixes, you do:

>> x*transpose(x2)
ans =  7.0000

Transposing a matrix

You can call the built-in transpose() function (shown previously) or use the ' operator.

>> x2'
ans =
   1
   2
   3
   4
   5
   6

Reshaping matrixes

You can reshape using the reshape(<arr>, nrows, ncols) function. Note, it flows elements column wise, followed by rows as shown below:

>> vec = 1:16;
>> reshape(vec, 2,8)
ans =
    1    3    5    7    9   11   13   15
    2    4    6    8   10   12   14   16

>> reshape(vec, 4,4)  % note how elements flow along columns, not rows.
ans =
    1    5    9   13
    2    6   10   14
    3    7   11   15
    4    8   12   16

Range functions

You can generate uniformly spaced vectors, using 2 methods. If you want uniformly spaced vectors, use the : operator. If you want a specific number of elements within a range, then use the linspace function.

The : operator takes syntax start:spacing:end

first_10_even_numbers = 0:2:21
first_10_even_numbers = 0    2    4    6    8   10   12   14   16   18   20

The linspace function takes arguments linspace(start, end, numPoints) where start and end are included.

>> linspace(1,10,7)
ans = 1.0000    2.5000    4.0000    5.5000    7.0000    8.5000   10.0000

Conditional processing

Logical operators in octave are <, >, <=, >=, ~, ==, ~=, &, |. Matlab returns 1 for True and 0 for False. Thus

>> heat = 74
heat =  74
>> heat < 100
ans = 1
>> heat > 100
ans = 0
>> heat == 74
ans = 1

Conditional operators also map on all elements of the vector. Thus to find elements in a vector greater than a threshold, do:

>> v1 = linspace(1,10,7);
>> ind = v1 > 5  % returns truth vector
ind = 0  0  0  1  1  1  1 
>> vals = v1(ind) % use truth vector as index
vals = 5.5000    7.0000    8.5000   10.0000

To find values that fall between a range:

>> v1(v1<8 & v1>3)
ans = 4.0000   5.5000   7.0000

if-elseif-else blocks

Jump to defining functions if you want to read that first.

function charge = parkingRates(hours)
  %Calculates parking charge based on number of hours%
  if hours <=1
    charge = 2;
  elseif hours<=8 && hours>1
    charge = hours*0.75;
  else
    charge = 9;
  end
end

The && operator is a performance opt that Matlab editor suggested. Else I would use only one of it.

Styling print outputs

Use disp to display. You can style print outputs with sprintf and sending the string composed to disp().

>> vec1 = [2,3,4,5];
>> disp(vec1)
   2   3   4   5

>> disp(sprintf("Mean of the array is %0.3f", mean(vec1)))
Mean of the array is 3.500

Use C style statements within sprintf.

Array indexing, slicing, dicing

Consider this 4x4 matrix:

>> readings=linspace(1,10,16);
>> readings = reshape(readings, 4,4)
readings =

    1.0000    3.4000    5.8000    8.2000
    1.6000    4.0000    6.4000    8.8000
    2.2000    4.6000    7.0000    9.4000
    2.8000    5.2000    7.6000   10.0000
>>

To get the first column vector, use :,n where : means all rows here and replace n with column number, starting with 1 instead of 0.:

>> readings(:,1)  % Note: matlab is 1 indexed, not 0 indexed
ans =
   1.0000
   1.6000
   2.2000
   2.8000
>>

Also note, I am using () to index arrays, not [] as in Python and other languages.

To dice the array by pulling 2nd, 3rd rows, 2nd, 3rd columns, do the following:

>> readings([2,3], [2,3])
ans =
   4.0000   6.4000
   4.6000   7.0000

Array concatenation

You can concatenate along rows or columns using the , or ; operators:

>> [eye(3), randn(3)]  % use , to concatenate along columns
ans =
   1.00000   0.00000   0.00000   0.68098  -1.07370  -1.99950
   0.00000   1.00000   0.00000   0.19421   0.06390   0.29864
   0.00000   0.00000   1.00000  -0.03696  -0.24735  -0.18180

>> [eye(3); randn(3)]  % use ; to concatenate along rows
ans =
   1.00000   0.00000   0.00000
   0.00000   1.00000   0.00000
   0.00000   0.00000   1.00000
   0.78756  -0.22399   0.59963
  -0.93718   0.34779   0.27082
  -2.00364  -0.45362  -1.47926
>>

Statistical operations on arrays

Let us start with an array as below:

>> vec45 = reshape(linspace(1,10,20), 4,5)
vec45 =
    1.0000    2.8947    4.7895    6.6842    8.5789
    1.4737    3.3684    5.2632    7.1579    9.0526
    1.9474    3.8421    5.7368    7.6316    9.5263
    2.4211    4.3158    6.2105    8.1053   10.0000
>>

Find mean of each column

>> mean(vec45)
ans =   1.7105   3.6053   5.5000   7.3947   9.2895

Find mean of each row

>> mean(vec45,2)  % pass the dimension arg. Set it to 2 for mean along rows
ans =
   4.7895
   5.2632
   5.7368
   6.2105

Another way of doing this is to transpose the matrix and take regular mean.

Find mean of entire matrix Turn the matrix into a column vector and pass that to the mean.

>> mean(vec45(:))
ans =  5.5000

Writing functions

You call functions as 

y = func_name(arg1, arg2);

The syntax to write functions is:

function [out1, out2...] = functionName(in1, in2, ...)
    %Doc string%
    statement1;
    statement2;
    out1 = statement;
    out2 = statement;
    ...
end

There is no return statement. All variables you define in the out parameter array (or scalar) will get returned. Matlab allows you to return more than one variable.

function result = hmean(vec1)
  %calculates harmonic mean of the given vector%
  vec1 = vec1(:)'; %convert to vectors
  reciprocals = 1./vec1;  %convert to reciprocals
  sum_reciprocals = sum(reciprocals); % sum of reciprocals - the denominator

  result = size(vec1)(2) / sum_reciprocals;
endfunction

>> hmean([2,3,4,5])
ans =  3.1169
>>

Reusing functions

A quick and easy way to reuse functions in other scripts is to put just the function (not anything else) in a separate .m file with the same name as the function. Matlab will automatically search for it and import it.

To hack the search path, use addpath('full_path') to add a folder to Octave path. This way, you can load a function or module from a different directory.