igraph_error_t igraph_vector_init(igraph_vector_t *v, igraph_integer_t size);

Every vector needs to be initialized before it can be used, and there are a number of initialization functions or otherwise called constructors. This function constructs a vector of the given size and initializes each entry to 0. Note that igraph_vector_null() can be used to set each element of a vector to zero. However, if you want a vector of zeros, it is much faster to use this function than to create a vector and then invoke igraph_vector_null().

Every vector object initialized by this function should be destroyed (ie. the memory allocated for it should be freed) when it is not needed anymore, the igraph_vector_destroy() function is responsible for this.

Arguments: 

Returns: 

Time complexity: operating system dependent, the amount of “time” required to allocate O(n) elements, n is the number of elements.

igraph_error_t igraph_vector_init_array(
        igraph_vector_t *v, const igraph_real_t *data, igraph_integer_t length);

Arguments: 

Returns: 

Time complexity: operating system specific, usually O(length).

igraph_error_t igraph_vector_init_copy(
    igraph_vector_t *to, const igraph_vector_t *from
);

The contents of the existing vector object will be copied to the new one.

Arguments: 

Returns: 

Time complexity: operating system dependent, usually O(n), n is the size of the vector.

igraph_error_t igraph_vector_init_range(igraph_vector_t *v, igraph_real_t start, igraph_real_t end);

The vector will contain the numbers start, start+1, ..., end-1. Note that the range is closed from the left and open from the right, according to C conventions.

Arguments: 

Returns: 

Time complexity: O(n), the number of elements in the vector.

void igraph_vector_destroy(igraph_vector_t *v);

All vectors initialized by igraph_vector_init() should be properly destroyed by this function. A destroyed vector needs to be reinitialized by igraph_vector_init(), igraph_vector_init_array() or another constructor.

Arguments: 

Time complexity: operating system dependent.

void igraph_vector_null(igraph_vector_t *v);

Note that igraph_vector_init() sets the elements to zero as well, so it makes no sense to call this function on a just initialized vector. Thus if you want to construct a vector of zeros, then you should use igraph_vector_init().

Arguments: 

Time complexity: O(n), the size of the vector.

void igraph_vector_fill(igraph_vector_t *v, igraph_real_t e);

Sets each element of the vector to the supplied constant.

Arguments: 

Time complexity: O(n), the size of the vector.

igraph_error_t igraph_vector_range(igraph_vector_t *v, igraph_real_t start, igraph_real_t end);

Sets the elements of the vector to contain the numbers start, start+1, ..., end-1. Note that the range is closed from the left and open from the right, according to C conventions. The vector will be resized to fit the range.

Arguments: 

Returns: 

Time complexity: O(n), the number of elements in the vector.

#define VECTOR(v)

Usage:

 VECTOR(v)[0] 

to access the first element of the vector, you can also use this in assignments, like:

 VECTOR(v)[10] = 5; 

Note that there are no range checks right now.

Arguments: 

Time complexity: O(1).

igraph_real_t igraph_vector_get(const igraph_vector_t *v, igraph_integer_t pos);

Unless you need a function, consider using the VECTOR macro instead for better performance.

Arguments: 

Returns: 

See also: 

Time complexity: O(1).

igraph_real_t* igraph_vector_get_ptr(const igraph_vector_t *v, igraph_integer_t pos);

Unless you need a function, consider using the VECTOR macro instead for better performance.

Arguments: 

Returns: 

See also: 

Time complexity: O(1).

void igraph_vector_set(igraph_vector_t *v, igraph_integer_t pos, igraph_real_t value);

Unless you need a function, consider using the VECTOR macro instead for better performance.

Arguments: 

See also: 

igraph_real_t igraph_vector_tail(const igraph_vector_t *v);

It is an error to call this function on an empty vector, the result is undefined.

Arguments: 

Returns: 

Time complexity: O(1).

const igraph_vector_t* igraph_vector_view(const igraph_vector_t *v,
        const igraph_real_t *data, igraph_integer_t length);

This is a special igraph_vector_t constructor. It allows to handle a regular C array as a igraph_vector_t temporarily. Be sure that you don't ever call the destructor (igraph_vector_destroy()) on objects created by this constructor.

Arguments: 

Returns: 

Time complexity: O(1)

void igraph_vector_copy_to(const igraph_vector_t *v, igraph_real_t *to);

The C array should have sufficient length.

Arguments: 

Time complexity: O(n), n is the size of the vector.

igraph_error_t igraph_vector_update(igraph_vector_t *to,
                                    const igraph_vector_t *from);

After this operation the contents of to will be exactly the same as that of from. The vector to will be resized if it was originally shorter or longer than from.

Arguments: 

Returns: 

Time complexity: O(n), the number of elements in from.

igraph_error_t igraph_vector_append(igraph_vector_t *to,
                                    const igraph_vector_t *from);

The target vector will be resized (except when from is empty).

Arguments: 

Returns: 

Time complexity: O(n), the number of elements in the new vector.

igraph_error_t igraph_vector_swap(igraph_vector_t *v1, igraph_vector_t *v2);

Arguments: 

Returns: 

Time complexity: O(1).

igraph_error_t igraph_vector_swap_elements(igraph_vector_t *v,
        igraph_integer_t i, igraph_integer_t j);

Note that currently no range checking is performed.

Arguments: 

Returns: 

Time complexity: O(1).

igraph_error_t igraph_vector_reverse(igraph_vector_t *v);

The first element will be last, the last element will be first, etc.

Arguments: 

Returns: 

See also: 

Time complexity: O(n), the number of elements.

void igraph_vector_reverse_section(igraph_vector_t *v, igraph_integer_t from, igraph_integer_t to);

Arguments: 

See also: 

Time complexity: O(to - from), the number of elements to reverse.

void igraph_vector_rotate_left(igraph_vector_t *v, igraph_integer_t n);

Rotates the elements of a vector to the left by the given number of steps. Element index n will have index 0 after the rotation. For example, rotating (0, 1, 2, 3, 4, 5) by 2 yields (2, 3, 4, 5, 0, 1).

Arguments: 

Time complexity: O(n), the number of elements.

igraph_error_t igraph_vector_shuffle(igraph_vector_t *v);

The Fisher-Yates shuffle ensures that every permutation is equally probable when using a proper randomness source. Of course this does not apply to pseudo-random generators as the cycle of these generators is less than the number of possible permutations of the vector if the vector is long enough.

Arguments: 

Returns: 

Time complexity: O(n), n is the number of elements in the vector.

References:

#include <igraph.h>

#define R_INTEGER(a,b) (igraph_rng_get_integer(igraph_rng_default(), (a), (b)))
#define R_UNIF(a,b) (igraph_rng_get_unif(igraph_rng_default(), (a), (b)))

int main(void) {
    igraph_real_t d;
    igraph_vector_t u, v;
    igraph_integer_t i, k, n;

    /********************************
     * Example usage
     ********************************/

    /* Sequences with one element. Such sequences are trivially permuted.
     * The result of any Fisher-Yates shuffle on a sequence with one element
     * must be the original sequence itself.
     */
    n = 1;
    igraph_vector_init(&v, n);
    igraph_rng_seed(igraph_rng_default(), 42); /* make tests deterministic */
    k = R_INTEGER(-1000, 1000);
    VECTOR(v)[0] = k;
    igraph_vector_shuffle(&v);
    if (VECTOR(v)[0] != k) {
        return 1;
    }
    d = R_UNIF(-1000.0, 1000.0);

    VECTOR(v)[0] = d;
    igraph_vector_shuffle(&v);
    if (VECTOR(v)[0] != d) {
        return 2;
    }
    igraph_vector_destroy(&v);

    /* Sequences with multiple elements. A Fisher-Yates shuffle of a sequence S
     * is a random permutation \pi(S) of S. Thus \pi(S) must have the same
     * length and elements as the original sequence S. A major difference between
     * S and its random permutation \pi(S) is that the order in which elements
     * appear in \pi(S) is probably different from how elements are ordered in S.
     * If S has length n = 1, then both \pi(S) and S are equivalent sequences in
     * that \pi(S) is merely S and no permutation has taken place. If S has
     * length n > 1, then there are n! possible permutations of S. Assume that
     * each such permutation is equally likely to appear as a result of the
     * Fisher-Yates shuffle. As n increases, the probability that S is different
     * from \pi(S) also increases. We have a probability of 1 / n! that S and
     * \pi(S) are equivalent sequences.
     */
    n = 100;
    igraph_vector_init(&u, n);
    igraph_vector_init(&v, n);

    for (i = 0; i < n; i++) {
        k = R_INTEGER(-1000, 1000);
        VECTOR(u)[i] = k;
        VECTOR(v)[i] = k;
    }

    igraph_vector_shuffle(&v);
    /* must have same length */
    if (igraph_vector_size(&v) != n) {
        return 3;
    }
    if (igraph_vector_size(&u) != igraph_vector_size(&v)) {
        return 4;
    }
    /* must have same elements */
    igraph_vector_sort(&u);
    igraph_vector_sort(&v);
    if (!igraph_vector_all_e(&u, &v)) {
        return 5;
    }
    igraph_vector_destroy(&u);
    igraph_vector_destroy(&v);

    /* empty sequence */
    igraph_vector_init(&v, 0);
    igraph_vector_shuffle(&v);
    igraph_vector_destroy(&v);

    return 0;
}

igraph_error_t igraph_vector_permute(igraph_vector_t *v, const igraph_vector_int_t *index);

This function takes a vector v and a corresponding index vector ind, and permutes the elements of v such that v[ind[i]] is moved to become v[i] after the function is executed.

It is an error to call this function with an index vector that does not represent a valid permutation. Each element in the index vector must be between 0 and the length of the vector minus one (inclusive), and each such element must appear only once. The function does not attempt to validate the index vector.

The index vector that this function takes is compatible with the index vector returned from igraph_vector_sort_ind(); passing in the index vector from igraph_vector_sort_ind() will sort the original vector.

As a special case, this function allows the index vector to be shorter than the vector being permuted, in which case the elements whose indices do not occur in the index vector will be removed from the vector.

Arguments: 

Returns: 

Time complexity: O(n), the size of the vector.

void igraph_vector_add_constant(igraph_vector_t *v, igraph_real_t plus);

plus is added to every element of v. Note that overflow might happen.

Arguments: 

Time complexity: O(n), the number of elements.

void igraph_vector_scale(igraph_vector_t *v, igraph_real_t by);

Arguments: 

Returns: 

Added in version 0.2.

Time complexity: O(n), the number of elements in a vector.

igraph_error_t igraph_vector_add(igraph_vector_t *v1,
                                 const igraph_vector_t *v2);

Add the elements of v2 to v1, the result is stored in v1. The two vectors must have the same length.

Arguments: 

Returns: 

Time complexity: O(n), the number of elements.

igraph_error_t igraph_vector_sub(igraph_vector_t *v1,
                                 const igraph_vector_t *v2);

Subtract the elements of v2 from v1, the result is stored in v1. The two vectors must have the same length.

Arguments: 

Returns: 

Time complexity: O(n), the length of the vectors.

igraph_error_t igraph_vector_mul(igraph_vector_t *v1,
                                 const igraph_vector_t *v2);

v1 will be multiplied by v2, elementwise. The two vectors must have the same length.

Arguments: 

Returns: 

Time complexity: O(n), the number of elements.

igraph_error_t igraph_vector_div(igraph_vector_t *v1,
                                 const igraph_vector_t *v2);

v1 is divided by v2, elementwise. They must have the same length. If the base type of the vector can generate divide by zero errors then please make sure that v2 contains no zero if you want to avoid trouble.

Arguments: 

Returns: 

Time complexity: O(n), the length of the vectors.

igraph_error_t igraph_vector_floor(const igraph_vector_t *from, igraph_vector_int_t *to);

Flooring means rounding down to the nearest integer.

Arguments: 

Returns: 

Time complexity: O(n), where n is the number of elements in the vector.

igraph_bool_t igraph_vector_all_e(const igraph_vector_t *lhs,
                                  const igraph_vector_t *rhs);

Checks element-wise equality of two vectors. For vectors containing floating point values, consider using igraph_matrix_all_almost_e().

Arguments: 

Returns: 

Time complexity: O(n), the length of the vectors.

igraph_bool_t igraph_vector_all_almost_e(const igraph_vector_t *lhs,
                                         const igraph_vector_t *rhs,
                                         igraph_real_t eps);

Checks if the elements of two vectors are equal within a relative tolerance.

Arguments: 

Returns: 

igraph_bool_t igraph_vector_all_l(const igraph_vector_t *lhs,
                                  const igraph_vector_t *rhs);

Arguments: 

Returns: 

Time complexity: O(n), the length of the vectors.

igraph_bool_t igraph_vector_all_g(const igraph_vector_t *lhs,
                                  const igraph_vector_t *rhs);

Arguments: 

Returns: 

Time complexity: O(n), the length of the vectors.

igraph_bool_t igraph_vector_all_le(const igraph_vector_t *lhs,
                                   const igraph_vector_t *rhs);

Arguments: 

Returns: 

Time complexity: O(n), the length of the vectors.

igraph_bool_t igraph_vector_all_ge(const igraph_vector_t *lhs,
                                   const igraph_vector_t *rhs);

Arguments: 

Returns: 

Time complexity: O(n), the length of the vectors.

igraph_bool_t igraph_vector_is_equal(const igraph_vector_t *lhs,
                                     const igraph_vector_t *rhs);

This is an alias of igraph_vector_all_e() with a more intuitive name.

Arguments: 

Returns: 

Time complexity: O(n), the length of the vectors.

igraph_error_t igraph_vector_zapsmall(igraph_vector_t *v, igraph_real_t tol);

Vector elements which are smaller in magnitude than the given absolute tolerance will be replaced by exact zeros. The default tolerance corresponds to two-thirds of the representable digits of igraph_real_t, i.e. DBL_EPSILON^(2/3) which is approximately 10^-10.

Arguments: 

Returns: 

See also: 

int igraph_vector_lex_cmp(
    const igraph_vector_t *lhs, const igraph_vector_t *rhs
);

If the elements of two vectors match but one is shorter, the shorter one comes first. Thus {1, 3} comes after {1, 2, 3}, but before {1, 3, 4}.

This function is typically used together with igraph_vector_list_sort().

Arguments: 

Returns: 

See also: 

Time complexity: O(n), the number of elements in the smaller vector.

#include <igraph.h>
#include <stdio.h>

int main(void) {
    igraph_t graph;
    igraph_vector_int_list_t cliques;
    igraph_integer_t i, n;

    /* Set a random seed to make the program deterministic */
    igraph_rng_seed(igraph_rng_default(), 31415);

    /* Create a random graph with a given number of vertices and edges */
    igraph_erdos_renyi_game_gnm(&graph, 15, 80, IGRAPH_UNDIRECTED, IGRAPH_NO_LOOPS);

    /* Find all maximal cliques in the graph */
    igraph_vector_int_list_init(&cliques, 0);
    igraph_maximal_cliques(&graph, &cliques, -1, -1);

    /* Print the cliques in lexicographical order */
    printf("Maximal cliques in lexicographical order:\n");
    igraph_vector_int_list_sort(&cliques, igraph_vector_int_lex_cmp);
    n = igraph_vector_int_list_size(&cliques);
    for (i=0; i < n; ++i) {
        igraph_vector_int_print(igraph_vector_int_list_get_ptr(&cliques, i));
    }

    /* Print the cliques in colexicographical order */
    printf("\nMaximal cliques in colexicographical order:\n");
    igraph_vector_int_list_sort(&cliques, igraph_vector_int_colex_cmp);
    n = igraph_vector_int_list_size(&cliques);
    for (i=0; i < n; ++i) {
        igraph_vector_int_print(igraph_vector_int_list_get_ptr(&cliques, i));
    }

    /* Destroy data structures when we no longer need them */
    igraph_vector_int_list_destroy(&cliques);
    igraph_destroy(&graph);

    return 0;
}

int igraph_vector_lex_cmp_untyped(const void *lhs, const void *rhs);

If the elements of two vectors match but one is shorter, the shorter one comes first. Thus {1, 3} comes after {1, 2, 3}, but before {1, 3, 4}.

This function is typically used together with igraph_vector_ptr_sort().

Arguments: 

Returns: 

See also: 

Time complexity: O(n), the number of elements in the smaller vector.

int igraph_vector_colex_cmp(
    const igraph_vector_t *lhs, const igraph_vector_t *rhs
);

This comparison starts from the last element of both vectors and moves backward. If the elements of two vectors match but one is shorter, the shorter one comes first. Thus {1, 2} comes after {3, 2, 1}, but before {0, 1, 2}.

This function is typically used together with igraph_vector_list_sort().

Arguments: 

Returns: 

See also: 

Time complexity: O(n), the number of elements in the smaller vector.

#include <igraph.h>
#include <stdio.h>

int main(void) {
    igraph_t graph;
    igraph_vector_int_list_t cliques;
    igraph_integer_t i, n;

    /* Set a random seed to make the program deterministic */
    igraph_rng_seed(igraph_rng_default(), 31415);

    /* Create a random graph with a given number of vertices and edges */
    igraph_erdos_renyi_game_gnm(&graph, 15, 80, IGRAPH_UNDIRECTED, IGRAPH_NO_LOOPS);

    /* Find all maximal cliques in the graph */
    igraph_vector_int_list_init(&cliques, 0);
    igraph_maximal_cliques(&graph, &cliques, -1, -1);

    /* Print the cliques in lexicographical order */
    printf("Maximal cliques in lexicographical order:\n");
    igraph_vector_int_list_sort(&cliques, igraph_vector_int_lex_cmp);
    n = igraph_vector_int_list_size(&cliques);
    for (i=0; i < n; ++i) {
        igraph_vector_int_print(igraph_vector_int_list_get_ptr(&cliques, i));
    }

    /* Print the cliques in colexicographical order */
    printf("\nMaximal cliques in colexicographical order:\n");
    igraph_vector_int_list_sort(&cliques, igraph_vector_int_colex_cmp);
    n = igraph_vector_int_list_size(&cliques);
    for (i=0; i < n; ++i) {
        igraph_vector_int_print(igraph_vector_int_list_get_ptr(&cliques, i));
    }

    /* Destroy data structures when we no longer need them */
    igraph_vector_int_list_destroy(&cliques);
    igraph_destroy(&graph);

    return 0;
}

int igraph_vector_colex_cmp_untyped(const void *lhs, const void *rhs);

This comparison starts from the last element of both vectors and moves backward. If the elements of two vectors match but one is shorter, the shorter one comes first. Thus {1, 2} comes after {3, 2, 1}, but before {0, 1, 2}.

This function is typically used together with igraph_vector_ptr_sort().

Arguments: 

Returns: 

See also: 

Time complexity: O(n), the number of elements in the smaller vector.

igraph_real_t igraph_vector_min(const igraph_vector_t *v);

The vector must not be empty.

Arguments: 

Returns: 

Time complexity: O(n), the number of elements.

igraph_real_t igraph_vector_max(const igraph_vector_t *v);

The vector must not be empty.

Arguments: 

Returns: 

Time complexity: O(n), the number of elements.

igraph_integer_t igraph_vector_which_min(const igraph_vector_t* v);

The vector must not be empty. If the smallest element is not unique, then the index of the first is returned. If the vector contains NaN values, the index of the first NaN value is returned.

Arguments: 

Returns: 

Time complexity: O(n), the number of elements.

igraph_integer_t igraph_vector_which_max(const igraph_vector_t *v);

The vector must not be empty. If the largest element is not unique, then the index of the first is returned. If the vector contains NaN values, the index of the first NaN value is returned.

Arguments: 

Returns: 

Time complexity: O(n), n is the size of the vector.

void igraph_vector_minmax(const igraph_vector_t *v,
                          igraph_real_t *min, igraph_real_t *max);

Handy if you want to have both the smallest and largest element of a vector. The vector is only traversed once. The vector must be non-empty. If a vector contains at least one NaN, both min and max will be NaN.

Arguments: 

Time complexity: O(n), the number of elements.

void igraph_vector_which_minmax(const igraph_vector_t *v,
        igraph_integer_t *which_min, igraph_integer_t *which_max);

Handy if you need the indices of the smallest and largest elements. The vector is traversed only once. The vector must be non-empty. If the minimum or maximum is not unique, the index of the first minimum or the first maximum is returned, respectively. If a vector contains at least one NaN, both which_min and which_max will point to the first NaN value.

Arguments: 

Time complexity: O(n), the number of elements.

igraph_bool_t igraph_vector_empty(const igraph_vector_t *v);

Arguments: 

Returns: 

Time complexity: O(1).

igraph_integer_t igraph_vector_size(const igraph_vector_t *v);

Returns the number of elements stored in the vector.

Arguments: 

Returns: 

Time complexity: O(1).

igraph_integer_t igraph_vector_capacity(const igraph_vector_t *v);

Note that this might be different from the size of the vector (as queried by igraph_vector_size()), and specifies how many elements the vector can hold, without reallocation.

Arguments: 

Returns: 

See also: 

Time complexity: O(1).

igraph_real_t igraph_vector_sum(const igraph_vector_t *v);

For the empty vector 0 is returned.

Arguments: 

Returns: 

Time complexity: O(n), the size of the vector.

igraph_real_t igraph_vector_prod(const igraph_vector_t *v);

For the empty vector one (1) is returned.

Arguments: 

Returns: 

Time complexity: O(n), the size of the vector.

igraph_bool_t igraph_vector_isininterval(const igraph_vector_t *v,
        igraph_real_t low,
        igraph_real_t high);

Arguments: 

Returns: 

Time complexity: O(n), the number of elements in the vector.

igraph_real_t igraph_vector_maxdifference(const igraph_vector_t *m1,
        const igraph_vector_t *m2);

The element with the largest absolute value in m1 - m2 is returned. Both vectors must be non-empty, but they not need to have the same length, the extra elements in the longer vector are ignored. If any value is NaN in the shorter vector, the result will be NaN.

Arguments: 

Returns: 

Time complexity: O(n), the number of elements in the shorter vector.

igraph_error_t igraph_vector_is_nan(const igraph_vector_t *v, igraph_vector_bool_t *is_nan);

Arguments: 

Returns: 

Time complexity: O(n), the number of elements.

igraph_bool_t igraph_vector_is_any_nan(const igraph_vector_t *v);

Arguments: 

Returns: 

Time complexity: O(n), the number of elements.

igraph_bool_t igraph_vector_is_all_finite(const igraph_vector_t *v);

Arguments: 

Returns: 

Time complexity: O(n), the number of elements.

igraph_bool_t igraph_vector_contains(const igraph_vector_t *v,
        igraph_real_t what);

Check whether the supplied element is included in the vector, by linear search.

Arguments: 

Returns: 

Time complexity: O(n), the length of the vector.

igraph_bool_t igraph_vector_search(const igraph_vector_t *v,
        igraph_integer_t from, igraph_real_t what, igraph_integer_t *pos);

The supplied element what is searched in vector v, starting from element index from. If found then the index of the first instance (after from) is stored in pos.

Arguments: 

Returns: 

Time complexity: O(m), the number of elements to search, the length of the vector minus the from argument.

igraph_bool_t igraph_vector_binsearch(const igraph_vector_t *v,
        igraph_real_t what, igraph_integer_t *pos);

It is assumed that the vector is sorted. If the specified element (what) is not in the vector, then the position of where it should be inserted (to keep the vector sorted) is returned. If the vector contains any NaN values, the returned value is undefined and pos may point to any position.

Arguments: 

Returns: 

Time complexity: O(log(n)), n is the number of elements in v.

igraph_bool_t igraph_vector_binsearch_slice(const igraph_vector_t *v,
        igraph_real_t what, igraph_integer_t *pos, igraph_integer_t start, igraph_integer_t end);

It is assumed that the indicated slice of the vector, from start to end, is sorted. If the specified element (what) is not in the slice of the vector, then the position of where it should be inserted (to keep the slice sorted) is returned. Note that this means that the returned index will point inside the slice (including its endpoints), but will not evaluate values outside the slice. If the indicated slice contains any NaN values, the returned value is undefined and pos may point to any position within the slice.

Arguments: 

Returns: 

Time complexity: O(log(n)), n is the number of elements in the slice of v, i.e. end - start.

igraph_bool_t igraph_vector_contains_sorted(const igraph_vector_t *v, igraph_real_t what);

It is assumed that the vector is sorted.

Arguments: 

Returns: 

Time complexity: O(log(n)), n is the number of elements in v.

void igraph_vector_clear(igraph_vector_t* v);

This function simply sets the size of the vector to zero, it does not free any allocated memory. For that you have to call igraph_vector_destroy().

Arguments: 

Time complexity: O(1).

igraph_error_t igraph_vector_reserve(igraph_vector_t *v, igraph_integer_t capacity);

igraph vectors are flexible, they can grow and shrink. Growing however occasionally needs the data in the vector to be copied. In order to avoid this, you can call this function to reserve space for future growth of the vector.

Note that this function does not change the size of the vector. Let us see a small example to clarify things: if you reserve space for 100 elements and the size of your vector was (and still is) 60, then you can surely add additional 40 elements to your vector before it will be copied.

Arguments: 

Returns: 

Time complexity: operating system dependent, should be around O(n), n is the new allocated size of the vector.

igraph_error_t igraph_vector_resize(igraph_vector_t* v, igraph_integer_t new_size);

Note that this function does not free any memory, just sets the size of the vector to the given one. It can on the other hand allocate more memory if the new size is larger than the previous one. In this case the newly appeared elements in the vector are not set to zero, they are uninitialized.

Arguments: 

Returns: 

See also: 

Time complexity: O(1) if the new size is smaller, operating system dependent if it is larger. In the latter case it is usually around O(n), n is the new size of the vector.

void igraph_vector_resize_min(igraph_vector_t *v);

This function attempts to deallocate the unused reserved storage of a vector. If it succeeds, igraph_vector_size() and igraph_vector_capacity() will be the same. The data in the vector is always preserved, even if deallocation is not successful.

Arguments: 

See also: 

Time complexity: operating system dependent, O(n) at worst.

igraph_error_t igraph_vector_push_back(igraph_vector_t *v, igraph_real_t e);

This function resizes the vector to be one element longer and sets the very last element in the vector to e.

Arguments: 

Returns: 

Time complexity: operating system dependent. What is important is that a sequence of n subsequent calls to this function has time complexity O(n), even if there hadn't been any space reserved for the new elements by igraph_vector_reserve(). This is implemented by a trick similar to the C++ vector class: each time more memory is allocated for a vector, the size of the additionally allocated memory is the same as the vector's current length. (We assume here that the time complexity of memory allocation is at most linear.)

igraph_real_t igraph_vector_pop_back(igraph_vector_t *v);

It is an error to call this function with an empty vector.

Arguments: 

Returns: 

Time complexity: O(1).

igraph_error_t igraph_vector_insert(
        igraph_vector_t *v, igraph_integer_t pos, igraph_real_t value);

Note that this function does not do range checking. Insertion will shift the elements from the position given to the end of the vector one position to the right, and the new element will be inserted in the empty space created at the given position. The size of the vector will increase by one.

Arguments: 

void igraph_vector_remove(igraph_vector_t *v, igraph_integer_t elem);

Note that this function does not do range checking.

Arguments: 

Time complexity: O(n-elem), n is the number of elements in the vector.

void igraph_vector_remove_section(
        igraph_vector_t *v, igraph_integer_t from, igraph_integer_t to);

Arguments: 

Time complexity: O(n-from), n is the number of elements in the vector.

igraph_error_t igraph_vector_complex_real(const igraph_vector_complex_t *v,
                               igraph_vector_t *real);

Arguments: 

Returns: 

Time complexity: O(n), n is the number of elements in the vector.

igraph_error_t igraph_vector_complex_imag(const igraph_vector_complex_t *v,
                               igraph_vector_t *imag);

Arguments: 

Returns: 

Time complexity: O(n), n is the number of elements in the vector.

igraph_error_t igraph_vector_complex_realimag(const igraph_vector_complex_t *v,
                                   igraph_vector_t *real,
                                   igraph_vector_t *imag);

Arguments: 

Returns: 

Time complexity: O(n), n is the number of elements in the vector.

igraph_error_t igraph_vector_complex_create(igraph_vector_complex_t *v,
                                 const igraph_vector_t *real,
                                 const igraph_vector_t *imag);

Arguments: 

Returns: 

Time complexity: O(n), n is the number of elements in the vector.

igraph_error_t igraph_vector_complex_create_polar(igraph_vector_complex_t *v,
                                       const igraph_vector_t *r,
                                       const igraph_vector_t *theta);

Arguments: 

Returns: 

Time complexity: O(n), n is the number of elements in the vector.

igraph_bool_t igraph_vector_complex_all_almost_e(const igraph_vector_complex_t *lhs,
                                                 const igraph_vector_complex_t *rhs,
                                                 igraph_real_t eps);

Checks if the elements of two complex vectors are equal within a relative tolerance.

Arguments: 

Returns: 

igraph_error_t igraph_vector_complex_zapsmall(igraph_vector_complex_t *v, igraph_real_t tol);

Similarly to igraph_vector_zapsmall(), small elements will be replaced by zeros. The operation is performed separately on the real and imaginary parts of the numbers. This way, complex numbers with a large real part and tiny imaginary part will effectively be transformed to real numbers. The default tolerance corresponds to two-thirds of the representable digits of igraph_real_t, i.e. DBL_EPSILON^(2/3) which is approximately 10^-10.

Arguments: 

Returns: 

See also: 

void igraph_vector_sort(igraph_vector_t *v);

If the vector contains any NaN values, the resulting ordering of NaN values is undefined and may appear anywhere in the vector.

Arguments: 

Time complexity: O(n log n) for n elements.

void igraph_vector_reverse_sort(igraph_vector_t *v);

If the vector contains any NaN values, the resulting ordering of NaN values is undefined and may appear anywhere in the vector.

Arguments: 

Time complexity: O(n log n) for n elements.

igraph_error_t igraph_vector_sort_ind(
        const igraph_vector_t *v,
        igraph_vector_int_t *inds,
        igraph_order_t order);

Takes an unsorted array v as input and computes an array of indices inds such that v[ inds[i] ], with i increasing from 0, is an ordered array (either ascending or descending, depending on order). The order of indices for identical elements is not defined. If the vector contains any NaN values, the ordering of NaN values is undefined.

Arguments: 

Returns: 

This routine uses igraph's built-in qsort routine. Algorithm: 1) create an array of pointers to the elements of v. 2) Pass this array to qsort. 3) after sorting the difference between the pointer value and the first pointer value gives its original position in the array. Use this to set the values of inds.

igraph_error_t igraph_vector_intersect_sorted(const igraph_vector_t *v1,
        const igraph_vector_t *v2, igraph_vector_t *result);

The elements that are contained in both vectors are stored in the result vector. All three vectors must be initialized.

For similar-size vectors, this function uses a straightforward linear scan. When the vector sizes differ substantially, it uses the set intersection method of Ricardo Baeza-Yates, which takes logarithmic time in the length of the larger vector.

The algorithm keeps the multiplicities of the elements: if an element appears k1 times in the first vector and k2 times in the second, the result will include that element min(k1, k2) times.

Reference:

Baeza-Yates R: A fast set intersection algorithm for sorted sequences. In: Lecture Notes in Computer Science, vol. 3109/2004, pp. 400--408, 2004. Springer Berlin/Heidelberg. https://doi.org/10.1007/978-3-540-27801-6_30

Arguments: 

Returns: 

Time complexity: O(m log(n)) where m is the size of the smaller vector and n is the size of the larger one.

igraph_integer_t igraph_vector_intersection_size_sorted(
        const igraph_vector_t *v1,
        const igraph_vector_t *v2);

Counts elements that are present in both vectors. This is particularly useful for counting common neighbours of two vertices.

For similar-size vectors, this function uses a straightforward linear scan. When the vector sizes differ substantially, it uses the set intersection method of Ricardo Baeza-Yates, which takes logarithmic time in the length of the larger vector.

The algorithm keeps the multiplicities of the elements: if an element appears k1 times in the first vector and k2 times in the second, the result will include that element min(k1, k2) times.

Reference:

Baeza-Yates R: A fast set intersection algorithm for sorted sequences. In: Lecture Notes in Computer Science, vol. 3109/2004, pp. 400--408, 2004. Springer Berlin/Heidelberg. https://doi.org/10.1007/978-3-540-27801-6_30

Arguments: 

Returns: 

Time complexity: O(m log(n)) where m is the size of the smaller vector and n is the size of the larger one.

igraph_error_t igraph_vector_difference_sorted(const igraph_vector_t *v1,
        const igraph_vector_t *v2, igraph_vector_t *result);

The elements that are contained in only the first vector but not the second are stored in the result vector. All three vectors must be initialized.

Arguments: 

igraph_error_t igraph_vector_ptr_init(igraph_vector_ptr_t* v, igraph_integer_t size);

This is the constructor of the pointer vector data type. All pointer vectors constructed this way should be destroyed via calling igraph_vector_ptr_destroy().

Arguments: 

Returns: 

Time complexity: operating system dependent, the amount of “time” required to allocate size elements.

igraph_error_t igraph_vector_ptr_init_copy(igraph_vector_ptr_t *to, const igraph_vector_ptr_t *from);

This function creates a pointer vector by copying another one. This is shallow copy, only the pointers in the vector will be copied.

It is potentially dangerous to copy a pointer vector with an associated item destructor. The copied vector will inherit the item destructor, which may cause problems when both vectors are destroyed as the items might get destroyed twice. Make sure you know what you are doing when copying a pointer vector with an item destructor, or unset the item destructor on one of the vectors later.

Arguments: 

Returns: 

Time complexity: O(n) if allocating memory for n elements can be done in O(n) time.

void igraph_vector_ptr_destroy(igraph_vector_ptr_t* v);

The destructor for pointer vectors.

Arguments: 

Time complexity: operating system dependent, the “time” required to deallocate O(n) bytes, n is the number of elements allocated for the pointer vector (not necessarily the number of elements in the vector).

void igraph_vector_ptr_free_all(igraph_vector_ptr_t* v);

If an item destructor is set for this pointer vector, this function will first call the destructor on all elements of the vector and then free all the elements using igraph_free(). If an item destructor is not set, the elements will simply be freed.

Arguments: 

Time complexity: operating system dependent, the “time” required to call the destructor n times and then deallocate O(n) pointers, each pointing to a memory area of arbitrary size. n is the number of elements in the pointer vector.

void igraph_vector_ptr_destroy_all(igraph_vector_ptr_t* v);

This function is equivalent to igraph_vector_ptr_free_all() followed by igraph_vector_ptr_destroy().

Arguments: 

Time complexity: operating system dependent, the “time” required to deallocate O(n) pointers, each pointing to a memory area of arbitrary size, plus the “time” required to deallocate O(n) bytes, n being the number of elements allocated for the pointer vector (not necessarily the number of elements in the vector).

igraph_integer_t igraph_vector_ptr_size(const igraph_vector_ptr_t* v);

Arguments: 

Returns: 

Time complexity: O(1).

void igraph_vector_ptr_clear(igraph_vector_ptr_t* v);

This function resizes a pointer to vector to zero length. Note that the pointed objects are not deallocated, you should call igraph_free() on them, or make sure that their allocated memory is freed in some other way, you'll get memory leaks otherwise. If you have set up an item destructor earlier, the destructor will be called on every element.

Note that the current implementation of this function does not deallocate the memory required for storing the pointers, so making a pointer vector smaller this way does not give back any memory. This behavior might change in the future.

Arguments: 

Time complexity: O(1).

igraph_error_t igraph_vector_ptr_push_back(igraph_vector_ptr_t* v, void* e);

Arguments: 

Returns: 

See also: 

Time complexity: O(1) or O(n), n is the number of elements in the vector. The pointer vector implementation ensures that n subsequent push_back operations need O(n) time to complete.

void *igraph_vector_ptr_pop_back(igraph_vector_ptr_t *v);

It is an error to call this function with an empty vector.

Arguments: 

Returns: 

Time complexity: O(1).

igraph_error_t igraph_vector_ptr_insert(igraph_vector_ptr_t* v, igraph_integer_t pos, void* e);

Note that this function does not do range checking. Insertion will shift the elements from the position given to the end of the vector one position to the right, and the new element will be inserted in the empty space created at the given position. The size of the vector will increase by one.

Arguments: 

void *igraph_vector_ptr_get(const igraph_vector_ptr_t* v, igraph_integer_t pos);

Arguments: 

Returns: 

Time complexity: O(1).

void igraph_vector_ptr_set(igraph_vector_ptr_t* v, igraph_integer_t pos, void* value);

Arguments: 

Time complexity: O(1).

igraph_error_t igraph_vector_ptr_resize(igraph_vector_ptr_t* v, igraph_integer_t newsize);

Note that if a vector is made smaller the pointed object are not deallocated by this function and the item destructor is not called on the extra elements.

Arguments: 

Returns: 

Time complexity: O(1) if the vector if made smaller. Operating system dependent otherwise, the amount of “time” needed to allocate the memory for the vector elements.

void igraph_vector_ptr_sort(igraph_vector_ptr_t *v, int (*compar)(const void*, const void*));

Sometimes it is necessary to sort the pointers in the vector based on the property of the element being referenced by the pointer. This function allows us to sort the vector based on an arbitrary external comparison function which accepts two void * pointers p1 and p2 and returns an integer less than, equal to or greater than zero if the first argument is considered to be respectively less than, equal to, or greater than the second. p1 and p2 will point to the pointer in the vector, so they have to be double-dereferenced if one wants to get access to the underlying object the address of which is stored in v.

Arguments: 

igraph_error_t igraph_vector_ptr_sort_ind(igraph_vector_ptr_t *v,
        igraph_vector_int_t *inds, cmp_t *cmp);

Takes an unsorted array v as input and computes an array of indices inds such that v[ inds[i] ], with i increasing from 0, is an ordered array (either ascending or descending, depending on \v order). The order of indices for identical elements is not defined.

Arguments: 

Returns: 

This routine uses the C library qsort routine. Algorithm: 1) create an array of pointers to the elements of v. 2) Pass this array to qsort. 3) after sorting the difference between the pointer value and the first pointer value gives its original position in the array. Use this to set the values of inds.

igraph_error_t igraph_vector_ptr_permute(igraph_vector_ptr_t* v, const igraph_vector_int_t* index);

This function takes a vector v and a corresponding index vector ind, and permutes the elements of v such that v[ind[i]] is moved to become v[i] after the function is executed.

It is an error to call this function with an index vector that does not represent a valid permutation. Each element in the index vector must be between 0 and the length of the vector minus one (inclusive), and each such element must appear only once. The function does not attempt to validate the index vector.

The index vector that this function takes is compatible with the index vector returned from igraph_vector_ptr_sort_ind(); passing in the index vector from igraph_vector_ptr_sort_ind() will sort the original vector.

As a special case, this function allows the index vector to be shorter than the vector being permuted, in which case the elements whose indices do not occur in the index vector will be removed from the vector.

Arguments: 

Returns: 

Time complexity: O(n), the size of the vector.

igraph_finally_func_t* igraph_vector_ptr_get_item_destructor(const igraph_vector_ptr_t *v);

The item destructor is a function which will be called on every non-null pointer stored in this vector when igraph_vector_ptr_destroy(), igraph_vector_ptr_destroy_all() or igraph_vector_ptr_free_all() is called.

Returns: 

Time complexity: O(1).

igraph_finally_func_t* igraph_vector_ptr_set_item_destructor(
    igraph_vector_ptr_t *v, igraph_finally_func_t *func);

The item destructor is a function which will be called on every non-null pointer stored in this vector when igraph_vector_ptr_destroy(), igraph_vector_ptr_destroy_all() or igraph_vector_ptr_free_all() is called.

Returns: 

Time complexity: O(1).

#define IGRAPH_VECTOR_PTR_SET_ITEM_DESTRUCTOR(v, func)

This macro is expanded to igraph_vector_ptr_set_item_destructor(), the only difference is that the second argument is automatically cast to an igraph_finally_func_t*. The cast is necessary in most cases as the destructor functions we use (such as igraph_vector_destroy()) take a pointer to some concrete igraph data type, while igraph_finally_func_t expects void*

igraph_real_t* igraph_vector_e_ptr(const igraph_vector_t *v, igraph_integer_t pos);

Arguments: 

Returns: 

See also: 

Time complexity: O(1).

igraph_error_t igraph_vector_init_seq(igraph_vector_t *v, igraph_real_t from, igraph_real_t to);

The vector will contain the numbers from, from+1, ..., to. Note that both endpoints are inclusive, contrary to typical usage of ranges in C.

Arguments: 

Returns: 

Time complexity: O(n), the number of elements in the vector.

igraph_error_t igraph_matrix_init(
        igraph_matrix_t *m, igraph_integer_t nrow, igraph_integer_t ncol);

Every matrix needs to be initialized before using it. This is done by calling this function. A matrix has to be destroyed if it is not needed any more; see igraph_matrix_destroy().

Arguments: 

Returns: 

Time complexity: usually O(n), n is the number of elements in the matrix.

igraph_error_t igraph_matrix_init_array(
        igraph_matrix_t *m, const igraph_real_t *data,
        igraph_integer_t nrow, igraph_integer_t ncol,
        igraph_matrix_storage_t storage);

The array is assumed to store the matrix data contiguously, either in a column-major or row-major format. In other words, data may store concatenated matrix columns or concatenated matrix rows. Constructing a matrix from column-major data is faster, as this is igraph's native storage format.

Arguments: 

Returns: 

Time complexity: operating system specific, usually O(nrow ncol).

igraph_error_t igraph_matrix_init_copy(igraph_matrix_t *to, const igraph_matrix_t *from);

Creates a matrix object by copying from an existing matrix.

Arguments: 

Returns: 

Time complexity: O(n), the number of elements in the matrix.

void igraph_matrix_destroy(igraph_matrix_t *m);

This function frees all the memory allocated for a matrix object. The destroyed object needs to be reinitialized before using it again.

Arguments: 

Time complexity: operating system dependent.