Short introduction
to
Octave
Lecture Notes
by Piotr Kowalczyk
Computational Finance – p. 1
Features
high level programming language and interactive
environment
intended especially for numerical calculations
natively supports many mathematical concepts
many function libraries available
free and compatible with MATLAB
www.octave.org
Computational Finance – p. 2
Basic use
simple calculator:
>> 2+2
ans = 4
>> 3^(2+1)
ans = 27
>> 0^0
ans = 1
>> (-4)^(.5)
ans = 1.2246e-016 + 2.0000e+000i
built-in support for complex arithmetic
special kinds of “numbers”: Inf, NaN
Computational Finance – p. 3
Basic use
built-in functions — all of the usual mathematical functions:
>> exp(1)-e
ans = 0
>> sqrt(-4)
ans = 0 + 2i
>> 1.5*log(2+sin(3+pi))
ans = 0.92996
named variables:
>> deg = pi/180
deg = 0.017453
>> h = cos(60*deg);
>> h
h = 0.50000
Computational Finance – p. 4
Vectors
row vector: >> v=[3 5 1 -2]; #or v=[3, 5, 1, -2];
column vector: >> v=[-1; 1.5; 3; 0];
colon notation: start[:increment]:end (range)
>> v=3:6
v =
3 4 5 6
>> x=0:.2:1
x =
0.00000 0.20000 0.40000 0.60000 0.80000 1.00000
evenly spaced vector:
linspace(x1,x2,N); #N elements between x1 and x2
logarithmically spaced vector:
logspace(x1,x2,N); #N elements between 10^x1 and 10^x2
Computational Finance – p. 5
Matrices
Generating:
>> A = [1 2 3; 4 5 6]
A =
1 2 3
4 5 6
>> B = [A; 3:-1:1]
B =
1 2 3
4 5 6
3 2 1
>> [A A]
ans =
1 2 3 1 2 3
4 5 6 4 5 6
Computational Finance – p. 6
Matrices
Special matrices:
eye() # create matrix with ones on the main diagonal
ones() # create matrix of ones
zeros() # create matrix of zeros
rand() # create random matrix
# with entries uniformly distributed in (0,1)
diag() # create diagonal matrix from the given vector
# or extract diagonal of the given matrix
‘Equation solving’ operators:
>> X = A \ B; # X is solution of AX=B
>> X = B / A; # X is solution of XA=B
Computational Finance – p. 7
Vectors and matrices
Indexing (indices of vectors and matrices start at one!):
>> v = [-1 3:2:8 0];
>> v(4)
ans = 7
>> A = [1 2 3; 4 5 6];
>> A(2,2)
ans = 5
>> A(1,:)
ans =
1 2 3
Assigning values:
>> A(1,:)=v(1:2:5)
A =
-1 5 0
4 5 6
Computational Finance – p. 8
Vectors and matrices
Dimensions:
>> size(v)
ans =
1 5
>> size(A)
ans =
2 3
Empty matrix:
>> P = [];
>> P = [P; [1 2]]
P =
1 2
Built-in functions can take vector and matrix arguments:
>> t = 0:.5:2;
>> sin(t)
ans =
0.00000 0.47943 0.84147 0.99749 0.90930
Computational Finance – p. 9
Operators
arithmetic: + - * / ^ ++ --
matrix: + - * ^ ’ .’
element-wise: .+ .- .* ./ .\ .^
logical: < <= > >= == != ! & | && ||
>> v = [5 8 1 -3 0 4 -8];
>> v < 2
ans =
0 0 1 1 1 0 1
>> v(v<2)
ans =
1 -3 0 -8
Computational Finance – p. 10
User-defined functions and script files
Defining a function:
function [out1, out2, ...] = name (input1, input2, ...)
sequence of commands
endfunction
Calling the function:
[outvar1, outvar2, ...] = name(invar1, invar2, ...);
Each function is stored in a different file, which must have the
same name as the function. Alternatively, several functions
can be defined in a script file with the rest of the program.
Script file is a normal text file (it can not begin with the
command function). This is the basic form of Octave
program. Scripts are run by typing the name of the script
(without the extension) in the Octave command window.
Both function and script files must have an extension .m
Computational Finance – p. 11
Control statements
if selection
if (condition) # brackets are not necessary
commands
elseif (condition)
commands
else
commands
endif
switch selection
switch expression # different than in C
case label
commands
case label
commands
...
otherwise
commands
end
Computational Finance – p. 12
Control statements
for loop
for variable = expression # expr.: vector or matrix
commands
endfor
while loop
while (condition)
commands
endwhile
do-until loop
do
commands
until (condition)
Computational Finance – p. 13
Input and output
save data var1 [var2 ...]
saves the variables var1 etc. into the file data
load data
restores the variables from the file data
fprintf, printf
resembles C syntax for formatted output
var = input("Text");
prints the text text and waits for the user to enter a value
format long; format short;
display 5 or 15 significant digits
Computational Finance – p. 14
Graphics
x = linspace(0,pi,200);
y = cos(x);
plot(x,y); # plot a line through the points (x,y)
plot(x,y,"go"); # use green circles instead
hold on;
plot(x,sin(x),"r"); # add red sinusoid
hold off;
title("Sample plot");
legend("cosine","sine")
print("sample.eps","-deps")
figure # create new window
Computational Finance – p. 15
Other useful functions
timing the execution of the statements:
tic;
# some calculations later ...
toc; # prints the number of seconds since tic
cputime can be used also.
getting help:
>> help
>> help log
Computational Finance – p. 16
Sparse matrices
Sparse matrix — matrix with many zeros, only non-zero
elements are stored in memory.
To convert a full matrix to a sparse one use:
B=sparse(A);
Use sparse matrices whenever the size is large!
Building band matrix (sparse diagonal matrix):
B=spdiags(V,C,m,n);
where column of V are diagonals of B represented by C
(negative values — diagonals below the main diagonal,
positive values — above the main diagonal) and m, n are
dimensions of matrix.
Computational Finance – p. 17
Efficiency
Octave is designed to process matrices and vectors and this is
its most powerful feature.
Vectorization is an essential programming skill in Octave.
A few tips on efficient programming in Octave:
avoid using loops (especially while and do-until)
if possible, vectorize all operations
if necessary, rewrite the problem to use matrices and
vectors
use ranges (colon notation) for vectors, whenever possible
Computational Finance – p. 18
Efficiency example
Problem: calculate the sum of squares of numbers 1, 2, ..., n
Solution 1:
n = input ("Type a value for n: ");
tic;
nn = 1;
s = 0;
while (nn <= n)
s += nn^2;
nn++;
endwhile
toc;
disp ("Result:"), disp(s);
Computational Finance – p. 19
Efficiency example
Problem: calculate the sum of squares of numbers 1, 2, ..., n
Solution 2:
n = input ("Type a value for n: ");
tic;
s = 0;
for nn = 1:n
s += nn^2;
endfor
toc;
disp ("Result:"), disp(s);
Computational Finance – p. 20
Efficiency example
Problem: calculate the sum of squares of numbers 1, 2, ..., n
Solution 3:
n = input ("Type a value for n: ");
tic;
s = sum((1:n).^2);
toc;
printf("Result: %f\n",s);
Computational Finance – p. 21
