This vignette is adapted from the official Armadillo documentation.
The linspace() function generates a vector of linearly spaced values from start to end (it includes end). The arguments can be start, end or start, end, N, where N is optional and indicates the number of elements in the vector (N is 100 by default).
The usage is:
vec v = linspace(start, end, N)
vector_type v = linspace<vector_type>(start, end, N)[[cpp11::register]] doubles linspace1_(const int& n) {
vec a = linspace(1, 2, n);
rowvec b = linspace<rowvec>(3, 4, n);
vec res = a + b.t();
return as_doubles(res);
}For N = 1, the generated vector will have a single element equal to end.
The logspace() function generates a vector of logarithmically spaced values from 10^start to 10^end (it includes 10^end). The arguments can be start, end or start, end, N, where N is optional and indicates the number of elements in the vector (N is 50 by default).
The usage is:
vec v = logspace(start, end, N)
vector_type v = logspace<vector_type>(start, end, N)[[cpp11::register]] doubles logspace1_(const int& n) {
vec a = logspace(1, 2, n);
rowvec b = logspace<rowvec>(3, 4, n);
vec res = a + b.t();
return as_doubles(res);
}The regspace() function generates a vector of regularly spaced values start, start + delta, start + 2*delta, ..., start + M * delta where M is M = floor((end - start) / delta). The arguments can be start, end or start, delta, end, where delta is optional (delta = 1 if start <= end and delta = -1 if start > end by default).
The usage is:
vec v = regspace(start, end)
vec v = regspace(start, delta, end)
vector_type v = regspace<vector_type>(start, end)
vector_type v = regspace<vector_type>(start, delta, end)The output vector will be empty if any of the following conditions are met:
start < end and delta < 0start > end and delta > 0delta = 0[[cpp11::register]] doubles regspace1_(const double& delta) {
vec a = regspace(1, delta, 2);
rowvec b = regspace<rowvec>(3, delta, 4);
vec res = a + b.t();
return as_doubles(res);
}regspace() to specify ranges for contiguous submatrix views, use span() instead.The randperm() function generates a vector of permutation of integers from 0 to N-1. The argument can be empty, N, or N, M, where N (N = 10 by default) is the range of integers and M (M = N by default) is the length of the output.
The usage is:
uvec v = randperm(N)
uvec v = randperm(N, M)[[cpp11::register]] integers randperm1_(const int& n, const int& m) {
uvec a = randperm(n);
uvec b = randperm(n, m);
uvec res = a + b;
return as_integers(res);
}The eye() function generates a matrix of size n x m. The argument can be n_rows, n_cols or size(X). When n_rows = n_cols, the output is an identity matrix.
The usage is:
mat X = eye(n_rows, n_cols)
matrix_type X = eye<matrix_type>(n_rows, n_cols)
matrix_type X = eye<matrix_type>(size(X))[[cpp11::register]] doubles_matrix<> eye1_(const int& n) {
mat A = eye(5,5); // or: mat A(5,5,fill::eye);
fmat B = 123.0 * eye<fmat>(5,5);
cx_mat C = eye<cx_mat>( size(B) );
return as_doubles(res);
}The ones() function generates a vector, matrix or cube. The arguments can be n_elem, n_rows, n_cols, n_rows, n_cols, n_slices, or size(X). The
The usage is:
vector_type v = ones<vector_type>(n_elem)
matrix_type X = ones<matrix_type>(n_rows, n_cols)
matrix_type Y = ones<matrix_type>(size(X))
cube_type Q = ones<cube_type>(n_rows, n_cols, n_slices)
cube_type R = ones<cube_type>(size(Q))[[cpp11::register]] doubles_matrix<> ones2_(const int& n) {
vec v = ones(n); // or: vec v(10, fill::ones);
uvec u = ones<uvec>(n);
rowvec r = ones<rowvec>(n);
mat A = ones(n, n); // or: mat A(n, n, fill::ones);
fmat B = ones<fmat>(n, n);
cube Q = ones(n, n, n + 1); // or: cube Q(n, n, n + 1, fill::ones);
mat res = diagmat(v) + diagmat(conv_to<vec>::from(u)) + diagmat(r) + A + B +
Q.slice(0);
return as_doubles_matrix(res);
}Specifying fill::ones during object construction is more compact. For example, mat A(5, 6, fill::ones).
The zeros() function generates a vector, matrix or cube. The arguments can be n_elem, n_rows, n_cols, n_rows, n_cols, n_slices, or size(X).
The usage is:
vector_type v = zeros<vector_type>(n_elem)
matrix_type X = zeros<matrix_type>(n_rows, n_cols)
matrix_type Y = zeros<matrix_type>(size(X))
cube_type Q = zeros<cube_type>(n_rows, n_cols, n_slices)
cube_type R = zeros<cube_type>(size(Q))[[cpp11::register]] doubles_matrix<> zeros2_(const int& n) {
vec v = zeros(n); // or: vec v(10, fill::zeros);
uvec u = zeros<uvec>(n);
rowvec r = zeros<rowvec>(n);
mat A = zeros(n, n); // or: mat A(n, n, fill::zeros);
fmat B = zeros<fmat>(n, n);
cube Q = zeros(n, n, n + 1); // or: cube Q(n, n, n + 1, fill::zeros);
mat res = diagmat(v) + diagmat(conv_to<vec>::from(u)) + diagmat(r) + A + B +
Q.slice(0);
return as_doubles_matrix(res);
}Specifying fill::zeros during object construction is more compact. For example, mat A(5, 6, fill::zeros).
The randu() function generates a vector, matrix or cube with the elements set to random floating point values uniformly distributed in the [a,b] interval. The arguments can be distr_param(a,b), n_elem, n_elem, distr_param(a,b), n_rows, n_cols, n_rows, n_cols, distr_param(a,b), n_rows, n_cols, n_slices, n_rows, n_cols, n_slices, distr_param(a,b), size(X), or size(X), distr_param(a,b).
The usage is:
// the scalar type can be: float, double, cx_float, or cx_double
scalar_type s = randu<scalar_type>()
scalar_type s = randu<scalar_type>(distr_param(a,b))
vector_type v = randu<vector_type>(n_elem)
vector_type v = randu<vector_type>(n_elem, distr_param(a,b))
matrix_type X = randu<matrix_type>(n_rows, n_cols)
matrix_type X = randu<matrix_type>(n_rows, n_cols, distr_param(a,b))
cube_type Q = randu<cube_type>(n_rows, n_cols, n_slices)
cube_type Q = randu<cube_type>(n_rows, n_cols, n_slices, distr_param(a,b))[[cpp11::register]] doubles_matrix<> randu3_(const int& n) {
double a = randu();
double b = randu(distr_param(10, 20));
vec v1 = randu(n); // or vec v1(n, fill::randu);
vec v2 = randu(n, distr_param(10, 20));
rowvec r1 = randu<rowvec>(n);
rowvec r2 = randu<rowvec>(n, distr_param(10, 20));
mat A1 = randu(n, n); // or mat A1(n, n, fill::randu);
mat A2 = randu(n, n, distr_param(10, 20));
fmat B1 = randu<fmat>(n, n);
fmat B2 = randu<fmat>(n, n, distr_param(10, 20));
mat res = diagmat(v1) + diagmat(v2) + diagmat(r1) + diagmat(r2) + A1 + A2 +
B1 + B2;
res.each_col([a](vec& x) { x += a; });
res.each_row([b](rowvec& y) { y /= b; });
return as_doubles_matrix(res);
}To generate a matrix with random integer values instead of floating point values, use randi() instead.
The randn() function generates a vector, matrix or cube with the elements set to random floating point values normally distributed with mean 0 and standard deviation 1. The arguments can be n_elem, distr_param(mean, stddev), n_elem, n_elem, distr_param(mean, stddev), n_rows, n_cols, n_rows, n_cols, distr_param(mean, stddev), n_rows, n_cols, n_slices, n_rows, n_cols, n_slices, distr_param(mean, stddev), size(X), or size(X), distr_param(mean, stddev).
The usage is:
// the scalar type can be: float, double, cx_float, or cx_double
scalar_type s = randn<scalar_type>()
scalar_type s = randn<scalar_type>(distr_param(mean, stddev))
vector_type v = randn<vector_type>(n_elem)
vector_type v = randn<vector_type>(n_elem, distr_param(mean, stddev))
matrix_type X = randn<matrix_type>(n_rows, n_cols)
matrix_type X = randn<matrix_type>(n_rows, n_cols, distr_param(mean, stddev))
cube_type Q = randn<cube_type>(n_rows, n_cols, n_slices)
cube_type Q = randn<cube_type>(n_rows, n_cols, n_slices, distr_param(mean, stddev))[[cpp11::register]] doubles_matrix<> randn3_(const int& n) {
vec v1 = randn(n); // or vec v1(n, fill::randn);
vec v2 = randn(n, distr_param(10, 20));
rowvec r1 = randn<rowvec>(n);
rowvec r2 = randn<rowvec>(n, distr_param(10, 20));
mat A1 = randn(n, n); // or mat A1(n, n, fill::randn);
mat A2 = randn(n, n, distr_param(10, 20));
fmat B1 = randn<fmat>(n, n);
fmat B2 = randn<fmat>(n, n, distr_param(10, 20));
mat res = diagmat(v1) + diagmat(v2) + diagmat(r1) + diagmat(r2) + A1 + A2 +
B1 + B2;
return as_doubles_matrix(res);
}The randg() function generates a vector, matrix or cube with the elements set to random floating point values gamma distributed with shape a and scale b. The arguments can be distr_param(a, b), n_elem, n_elem, distr_param(a, b), n_rows, n_cols, n_rows, n_cols, distr_param(a, b), n_rows, n_cols, n_slices, n_rows, n_cols, n_slices, distr_param(a, b), size(X), or size(X), distr_param(a, b).
The usage is:
// the scalar type can be: float, double, cx_float, or cx_double
scalar_type s = randg<scalar_type>()
scalar_type s = randg<scalar_type>(distr_param(a, b))
vector_type v = randg<vector_type>(n_elem)
vector_type v = randg<vector_type>(n_elem, distr_param(a, b))
matrix_type X = randg<matrix_type>(n_rows, n_cols)
matrix_type X = randg<matrix_type>(n_rows, n_cols, distr_param(a, b))
cube_type Q = randg<cube_type>(n_rows, n_cols, n_slices)
cube_type Q = randg<cube_type>(n_rows, n_cols, n_slices, distr_param(a, b))[[cpp11::register]] doubles_matrix<> randg3_(const int& n) {
int a = randi();
int b = randi(distr_param(-10, +20));
imat A1 = randi(n, n);
imat A2 = randi(n, n, distr_param(-10, +20));
mat B1 = randi<mat>(n, n);
mat B2 = randi<mat>(n, n, distr_param(-10, +20));
mat res = A1 + A2 + B1 + B2;
res.each_col([a](vec& x) { x *= a; });
res.each_row([b](rowvec& y) { y -= b; });
return as_doubles_matrix(res);
}The randi() function generates a vector, matrix or cube with the elements set to random integer values uniformly distributed in the [a,b] interval. The arguments can be distr_param(a, b), n_elem, n_elem, distr_param(a, b), n_rows, n_cols, n_rows, n_cols, distr_param(a, b), n_rows, n_cols, n_slices, n_rows, n_cols, n_slices, distr_param(a, b), size(X), or size(X), distr_param(a, b). The default values are a = 0 and b = maximum_int.
The usage is:
scalar_type s = randi<scalar_type>()
scalar_type s = randi<scalar_type>(distr_param(a, b))
vector_type v = randi<vector_type>(n_elem)
vector_type v = randi<vector_type>(n_elem, distr_param(a, b))
matrix_type X = randi<matrix_type>(n_rows, n_cols)
matrix_type X = randi<matrix_type>(n_rows, n_cols, distr_param(a, b))
cube_type Q = randi<cube_type>(n_rows, n_cols, n_slices)
cube_type Q = randi<cube_type>(n_rows, n_cols, n_slices, distr_param(a, b))[[cpp11::register]] integers_matrix<> randi3_(const int& n) {
uvec v1 = randi(n); // or uvec v1(n, fill::randi);
uvec v2 = randi(n, distr_param(10, 20));
umat A1 = randi(n, n); // or umat A1(n, n, fill::randi);
umat A2 = randi(n, n, distr_param(10, 20));
icube Q1 = randi(icube(n, n, n + 1)); // or icube Q1(n, n, n + 1, fill::randi);
icube Q2 = randi(icube(n, n, n + 1), distr_param(10, 20));
mat res = diagmat(conv_to<vec>::from(v1)) + diagmat(conv_to<vec>::from(v2)) +
A1 + A2 + Q1.slice(0) + Q2.slice(0);
return as_integers_matrix(res);
}To generate a matrix with random floating point values (e.g., float or double) instead of integers, use randu() instead.
The speye() function generates a sparse matrix of size n x n with the elements on the diagonal set to 1 and the remaining elements set to 0. The argument can be n_rows, n_cols or size(X). An identity matrix is generated when n_rows = n_cols.
The usage is:
sparse_matrix_type X = speye(n_rows, n_cols)
sparse_matrix_type X = speye<sparse_matrix_type>(size(X))[[cpp11::register]] doubles_matrix<> speye1_(const int& n) {
sp_mat A = speye<sp_mat>(n, n);
mat B = mat(A);
return as_doubles_matrix(B);
}The spones(X) function generates a sparse matrix with the same size as X and all the non-zero elements set to 1.
[[cpp11::register]] doubles_matrix<> spones1_(const int& n) {
sp_mat A = sprandu<sp_mat>(n, n, 0.1);
sp_mat B = spones(A);
mat C = mat(B);
return as_doubles_matrix(C);
}The sprandu() function generates a sparse matrix of size n_rows x n_cols with random floating point values uniformly distributed in the [0,1] interval. The arguments can be n_rows, n_cols, density or size(X), density.
The usage is:
sparse_matrix_type X = sprandu<sparse_matrix_type>(n_rows, n_cols, density)
sparse_matrix_type X = sprandu<sparse_matrix_type>(size(X), density)[[cpp11::register]] doubles_matrix<> sprandu1_(const int& n) {
sp_mat A = sprandu<sp_mat>(n, n, 0.05);
mat B = mat(A);
return as_doubles_matrix(B);
}The sprandn() function generates a sparse matrix of size n_rows x n_cols with random floating point values normally distributed with mean 0 and standard deviation 1. The arguments can be n_rows, n_cols, density or size(X), density.
The usage is:
sparse_matrix_type X = sprandn<sparse_matrix_type>(n_rows, n_cols, density)
sparse_matrix_type X = sprandn<sparse_matrix_type>(size(X), density)[[cpp11::register]] doubles_matrix<> sprandn1_(const int& n) {
sp_mat A = sprandn<sp_mat>(n, n, 0.05);
mat B = mat(A);
return as_doubles_matrix(B);
}The toeplitz() function generates a toeplitz matrix. The arguments can be a or a, b, where a is a vector that determines the first column and b is an optional vector that determines the first row.
Alternatively, circ_toeplitz() generates a circulant toeplitz matrix.
[[cpp11::register]] doubles_matrix<> toeplitz1_() {
vec a = linspace(1, 5, 5);
vec b = linspace(1, 5, 5);
mat A = toeplitz(a, b);
mat B = circ_toeplitz(a);
return as_doubles_matrix(A + B);
}