Implementation of a C + + library for data exchange with matlab R ⃝ Bachelor

matlab ⃝ stores matrices and others data structures into binary files which we call MAT-files. They represent an efficient and powerful way for data exchange between C++ programs and matlab ⃝ or other C++ programs. We propose an easy-to-use C++ library for input and output operations to share data in matrix form. 1 Description of the problem In the field of numerical simulations the need to share data between a C++ algorithm and other programs like matlab R ⃝, Octave or even other C++ programs is a common problem. An efficient way to do this is through MAT-files, the matlab R ⃝ standard for data saving. This standard has many advantages: • It supports dense and sparse matrix storage. • It supports other data structures, such as rectangular tensors, structs, objects and cell arrays. • The data are stored in a binary format, i.e. data precision is preserved. • The standard does not impose a limit on the number of data objects. • The standard allows data compression using zlib. • The standard is portable. Nevertheless, there is a lack of modern, powerful, object oriented, simple to use C++ libraries for input-output operation with this type of files. matlab R ⃝ does provide within External interfaces [2] a C/C++ API for that purpose, but its C based functional programming does not conform to the object-oriented ∗Seminar for Applied Mathematics, Swiss Federal Institute of Technology, Rämistrasse 101, Zurich †Seminar for Applied Mathematics, Swiss Federal Institute of Technology, Rämistrasse 101, Zurich