Data structures

A few data structures are available within BCP and COIN. Two of them will appear in the example: a class implementing vectors and a class implementing matrices.

The class BCP_vec is an implementation of an array with elements of generic type T with facilities for resizing. If V is a BCP_vec$<$T$>$, the following functions are used:

• BCP_vec<T> V$(k)$ : creates a vector with $k$ entries of type $<$T$>$.
• V.size() : returns the number of elements stored in V.
• V[$i$] : access the element stored at position $i$.
• V.push_back(x) : inserts element x to the end of the vector; if the space allocated to V is filled, then V is resized before inserting x.
• V.clear() : removes all entries in V.
• CoinFillN(V, $n$, elem) : put elem in the first $n$ entries of V.

See the documentation (file COIN/Bcp/include/BCP_vector.hpp) for the full description of this class.

The class CoinPackedMatrix implements a two dimensional matrix stored either by rows or by columns. For a CoinPackedMatrix M stored by rows, the following functions are used in the examples:

• M = new CoinPackedMatrix(false, $a$, $b$) : creates a matrix stored by rows (first boolean parameter ``false''). Roughly speaking, the parameter $0 \le a \le 1$ is a percentage of extra space to be allocated when a reallocation of the matrix occur: When trying to insert $k$ new rows in a matrix having $m$ rows and allocated space for $p$ rows, with $k+m > p$, the matrix will be reallocated to store $(k+m) (1+a)$ rows. The parameter $b$ is similar, for reallocation when columns are added. Note that if a matrix stored by columns is created using CoinPackedMatrix(true, $a$, $b$), then $a$ is used for the reallocation of columns and $b$ for the reallocation of rows.

• submatrixOf(M, nb_ind, v_ind) : extracts from M the submatrix formed by the rows of M with indices in the vector v_ind (having nb_ind entries).

• M.times(v, v_res) : Computes the matrix-vector product (M v) and puts the result in v_res.

See the documentation (file COIN/Coin/include/coinPackedMatrix.hpp) for the full description of this class.