The itbase library is the core of IT++ and it contains classes and functions for mathematics with scalars, vectors, and matrices. This document does not cover all the aspects of the itbase library. It does however explain the most important things you need to know in order to start using IT++. Once you are more familiar with the itbase library you will find the online reference manual more useful.
Apart from the standard C++ types e.g. char, short, int, long, double, float, and long long, the following types are specific for IT++:
complex<double>: Contains real and imaginary parts of type double bin: Used for binary (0,1) dataA vector can in principle be of arbitrary type (that support addition, subtraction, multiplication and division), since the general vector class Vec<TYPE> is templated. However, the most commonly used vector types are predefined. These predefined vector types are:
vec: Basic vector type containing double cvec: Vector type containing complex<double> ivec: Vector type containing int bvec: Vector type containing bin svec: Vector type containing short The general vector class is used to define the specialized classes above. The vec class is actually a Vec<double>. We urge you to use these predefined classes instead of Vec<TYPE> when ever possible.
The general matrix class is called Mat<TYPE>. These predefined matrix types are:
mat: Basic matrix type containing doublecmat: Matrix type containing complex<double> imat: Matrix type containing int bmat: Matrix type containing bin smat: Matrix type containing short As with vector, the general matrix class is used to define the specialized classes above. The mat class is thus a Mat<double>. We urge you to use these predefined classes instead of Mat<TYPE> whenever possible.
Vectors and matrices in IT++ are very similar. We therefore begin to describe the vector class in detail and then briefly explain the differences regarding matrices in the next section.
A vector containing elements of type double is defined with:
vec my_vector;
However, this will not assign a size (memory) to the vector. To assign size 10 to the vector we may use:
vec my_vector(10);
or
vec my_vector;
my_vector.set_size(10,false);
where the second parameter in the set_size call (true or false) determines if you want to copy the contents of the old data area into the new resized one, or not. This may be useful when down-sizing a vector, but in this case it is not. It is also equivalent to use
my_vector.set_length(10,false);
instead of set_size.
Observe that a declared vector (or matrix) is not cleared (the element values are undefined). To clear a vector we simply write
my_vector.clear();
or
my_vector.zeros();
To fill the vector with ones we write
my_vector.ones();
It is possible to retrieve the length (size) of a vector in any of the following ways:
length_of_vector = my_vector.length(); length_of_vector = my_vector.size(); length_of_vector = length(my_vector);
To assign values to a vector
vec a = "0 0.7 5 9.3"; // that is a = [0 0.7 5 9.3] ivec b = "0:5"; // that is b = [0 1 2 3 4 5] vec c = "3:2.5:13"; // that is c = [3 5.5 8 10.5 13] ivec d = "1:3:5,0:2:4"; // that is d = [1 3 5 0 2 4] vec e("1.2,3.4,5.6"); // that is e = [1.2 3.4 5.6] vec f; f.set("1.0 2.0 3.0 4.0"); // that is f = [1.0 2.0 3.0 4.0] vec g; g = f; // that is g is a copy of f
A comma or a space character separates the vector elements. When assigning or retrieving a specific vector element use
a(i) = 3.14; a(i);
for element number i. Vector elements are numbered such that a(0) denotes the first element. It is also possible to use square brackets as in the C language, i.e.
a[i] = 3.14; a[i];
Parts or a vector are retrieved by
a.left(3); // a vector containing the first 3 elements of a a.right(2); // a vector containing the last 2 elements of a a.mid(1,2); // a vector containing the 2 elements starting with a(1) a(2,4); // a vector containing all elements from a(2) to a(4)
If you have a vector called index_list containing indexes (ivec) you may write
// this gives a vector containing all elements in index_list
a(index_list);
If you have a bvec called e.g. bin_list you may write
a.get(bin_list); // this gives a vector containing all elements a(i) // for which bin_list(i) equals 1
Have a look at the following example:
#include "itbase.h" int man() { vec a = linspace(0,1,11); ivec index_list = "3 5 2 2"; bvec bin_list = "1 0 1 0 1 0 1 0 1 0 1"; cout << "a = " << a << endl; cout << "a(index_list) = " << a(index_list) << endl; cout << "a.get(bin_list) = " << a.get(bin_list) << endl; }
When you run this program you will see
a = [0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0] a(index_list) = [0.3 0.5 0.2 0.2] a(bin_list) = [0.0 0.2 0.4 0.6 0.8 1.0]
Below follows a listing of the most common vector manipulation commands that are available. All examples are given for an ivec denoted my_ivec, but of course this will work for other vector types as well.
shift_right: // Shift in scalar data (10) at position 0 my_ivec.shift_right(10); // Shift in vector at position 0 my_ivec.shift_right("10 4");
shift_left: // Shift in scalar data (10) at position Size()-1 my_ivec.shift_left(10); // Shift in vector at position Size()-1 my_ivec.shift_left("10 4");
replace_mid: // Replace part of vector from position (10) with the vector "11 13" my_ivec.replace_mid(10,"11 13");
del: // Delete element at index (10), making vector size one less
my_ivec.del(10);
ins: // Insert element at index (10), making vector size one extra
my_ivec.ins(10);
split: // Splits vector at pos (10). Returns first part, keep second part.
ivec first_part = my_ivec.split(10);
elem_mult: // Multiply two vectors element wise.
ivec my_product = elem_mult(my_invec1,my_ivec2);
elem_div: // Dividide two vectors element wise.
ivec my_product = elem_div(my_invec1,my_ivec2);
a = my_ivec1 + my_ivec2; // Addition of vectors a = my_ivec + 10; // Addition of vector and scalar a = my_ivec1 - my_ivec2; // Subtraction of vectors a = my_ivec - 10; // Subtraction of vector and scalar a = my_ivec * 10; // Multiplication of vector and scalar a = my_ivec / 10; // Division of vector and scalar
a += my_ivec; // Addition of vectors (a = a+my_ivec) my_ivec += 10; // Addition of vector and scalar (10) a -= my_ivec; // Subtraction of vectors (a = a-my_ivec) my_ivec -= 10; // Subtraction of vector and scalar (10) my_ivec *= 10; // Multiplication of vector and scalar (10) my_ivec /= 10; // Divsion of vector and scalar (10) my_ivec |= a; // Element wise division
concat a = concat(my_ivec1, my_ivec2); // concatenation of two vectors
In order to convert e.g an ivec to a vec we can write some thing like my_vec = to_vec(my_ivec). The following converters are available:
to_bvec,to_svec,to_ivec,to_vec,to_cvec.There are several functions that operate on vectors. Some examples are: max, max_index, min, min_index, product, energy, geometric_mean, mean, median, norm, round, variance, ceil_i, floor_i, round_i, find.
Examples of functions that generate different kinds of vectors are: linspace, ones_b, ones_c, ones_i, ones zeros_b. There are several more than these. Please refer to the IT++ reference manual for a description of these.
Matrices are two-dimensional arrays, and most of their functionality is similar to that of vectors. The predefined matrix types are:
mat,cmat,imat,smat,bmat.Below follows some examples that are specific for matrices only:
Define a matrix of type double with 3 rows and 4 columns
mat a(3,4);
Define a matrix of type int with 2 rows and 3 columns. A comma (,) or space is used to separate columns and a semicolon (;) is used to separate rows.
imat a = "1 2 3;4 5 6";
Access to rows and columns with get_row and get_col
a.get_row(1); // Returns the second row of the matrix b a.get_col(0); // Returns the first column of the matrix b
Set rows and columns with set_row and set_col
a.set_row(1,"9 8 7"); // Set second row to "9 8 7" a.set_col(0,"7 2"); // Set first column to "7 2"
The size of a matrix
// Set the size. "false" means "do not copy" a.set_size(4,5,false); int nr_of_rows = a.rows(); // return the number of rows int nr_of_columns = a.cols(); // return the number of columns
Access to parts of a matrix
a(r,c); // Access to a single element. a(i); // Access to a single element. Linear addressing, by rows. // Returns the sub-matrix from rows r1 to r2 and columns c1 to c2. a(r1,r2,c1,c2);
Copy rows and columns
// Copy row number "from" to row number "to" a.copy_row(to,from) // Copy column number "from" to column number "to" a.copy_col(to,from)
Swap rows and columns
// Swap rows number r1 and r2 a.swap_row(r1,r2) // Swap columns number c1 and c2 a.swap_col(c1,c2)
// Equivalent to the MATLAB command c = [a b] c = concat_horizontal(a,b); // Equivalent to the MATLAB command c = [a;b] c = concat_vertical(a,b);
The following converters are available:
to_mat,to_imat,to_cmat,to_bmat.The itbase library contains, among other things, the Array class. An Array can contain any type of data. Below is an example of an Array containing vectors (vec):
#include <itpp/itbase.h> using namespace itpp; using namespace std; int main() { Array<vec> my_vec_array(2); my_vec_array(0) = linspace(0,1,4); my_vec_array(1) = "0.1 0.2 0.3 0.4 0.3 0.2 0.1"; cout << "my_vec_array = " << my_vec_array << endl; return 0; }
Random vectors and matrices are easily obtained by using these predefined functions:
randb: Generates a random bit vector or matrixrandu: Generates a random uniform vector or matrixrandi: Generates a random index vector or matrixrandray: Generates a random Rayleigh vector or matrixrandrice: Generates a random Rice vector or matrixrandexp: Generates a random Exponential vector or matrixrandn: Generates a random Gaussian vector or matrixrandn_c: Generates a random complex Gaussian vector or matrixThe following discrete valued random number generators are available. More information about these can be found in the IT++ reference manual.
Bernoulli_RNG I_Uniform_RNG The following continuous valued random number generators are available.
Uniform_RNG Exponential_RNG Normal_RNG Complex_Normal_RNG AR1_Normal_RNG Weibull_RNG Rayleigh_RNG Rice_RNG The following deterministic sources are available:
Sine_Source Square_Source Triangle_Source Sawtooth_Source Impulse_Source Pattern_Source The following filter classes are available:
AR_Filter MA_Filter ARMA_Filter Freq_Filt The following filter functions are available:
filter The following signal processing functions are available:
a2k, k2a, a2lar, k2lar, lpc, levinsson, lerouxguegen fft, ifft, fft_real, ifft_real dct, idct spectrum cov, xcorr chirp dht, dht2, dwht, dhwt2, self_dht, self_dwht filter_spectrum filter_whiteness The Real_Timer class can be used to measure execution time of a program as in the following example:
#include <itpp/itbase.h> using namespace itpp; using namespace std; int main() { long sum = 0; Real_Timer my_timer; my_timer.tic(); for (int i=0; i<10000000; i++) { sum += i; } my_timer.toc_print(); cout << "The sum is " << sum << endl; return 0; }
The following example saves the variable a to the file my_file_name.it:
#include <itpp/itbase.h> using namespace itpp; int main() { it_file my_file("my_file_name.it"); vec a = "1.0 2.0 3.0 4.0"; my_file << Name("a") << a; return 0; }
The following example reads the variable a from the file my_file_name.it and prints it:
#include <itpp/itbase.h> using namespace itpp; using namespace std; int main() { it_file my_file("my_file_name.it"); vec a; my_file >> Name("a") >> a; cout << "a = " << a << endl; return 0; }
Note that *.it files can be read and written in Matlab/Octave by using the itload.m and itsave.m functions.
Also available is the class it_ifile that can only be used for reading of files.