Vectors and matrices

Vectors: Indexing starts with 0. Access to a single element via [].

• classic static vector  (static == at compile time)
• std:array<T,N>  (see www)
• static
• some methods:  [], at(), size(), constructors ()
• dynamic vector via pointer (dynamic == at run time)
• allocation (new T []) , deallocation (delete [])
• Example skalar (tar, doc)
• std::vector<T>   (see www)
  
• some methods:  [], at(), size(), constructors (), resize(), push_back(), pop_back(), front(), back()
•           method data()  returns a row pointer to the allocated memory.
• Example skalar_stl (tar, doc)
• dynamic vector in example intro_secretnumber (tar, doc)
• std::list<T>   (see www)
• some methods: constructors (), resize(), push_back(), pop_back()

Dense matrices with N nows and M colums:

• not supported as data class in C++
• 2D storage:
• vector<vector<T>>
• non continuous memory for matrix
• fragile dynamic allocation  
• access via mm[i][j]
• not compatible with most LinAlg libraries, e.g., BLAS, LAPACK, which require POD (plain old data)
• 1D storage:
• vector<T>
• vector length N*M needed
• continuous memory for matrix
• access via mm[i*M+j]
• via mm.data(), N, M compatible with LinAlg libraries.
• Example intro_vector_densematrix (tar, doc)

Sparse matrices (www):

• not supported
• many different matrix formats (CRS,COO, ...)
• we will used CRS (also named CSR) as described in the FEM code (plain version, oo version)
• Example intro_crsmatrix(tar, doc)