engleski

Using open-source numerical computation software in education - basic performance comparison and lab examples

Even though commercial and production-proven tools such as Matlab provide attractive education licenses, our university and high-school educational experience also includes usage of open-source alternatives - freely available software for numerical computations such as Scilab and GNU Octave. These tools are quite highly compatible with Matlab, offering similar programming languages allowing students to develop custom "applications" and use them to solve different problems. Due to the interpretive nature of the code execution, both tools covered in this paper - Scilab and GNU Octave, will offer lower performance when compared with Matlab, where the code is compiled and executed (not a topic of this paper). However, that makes them suitable candidates for a performance comparison of basic computations and procedures used in our everyday lab exercises. This paper provides a basic performance comparison of Scilab and GNU Octave, with additional remarks, where applicable, of certain performance improvements that can be achieved using a specific implementation approach (most often, so-called vectorization methods). In addition to a performance comparison, the paper gives an in- depth coverage of specific lab examples including simple signal generation and spectrum analysis, describing the similarities and differences among the tools. The examples are extended with simple execution time measurements and results obtained in the same computer environment. It is concluded that open-source computation tools offer quite functional environments allowing students and teachers to develop different kinds of programs, solving different calculation problems and find solutions in a satisfactory time. Even though using interpreters to execute the code, usual educational problems and labs can be solved without any issues and extreme time consummation. Of course, the scalability and slow execution may be a problem when working on larger size problems, but that is not an issue for undergraduate or even graduate studies.

Scilab ; GNU Octave ; performances ; interpreted execution ; open-source ; lab examples

nije evidentirano