A comparative study of several numerical computational packages.

摘要

The commercial software package Matlab is the most widely used numerical computational package. GNU Octave, FreeMat, and Scilab are other numerical computational packages that have many of the same features as Matlab. Unlike Matlab, the other three packages are free of charge to the user and easily available to download to Linux, Windows, or Mac OS X operating systems. Do these packages give comparable computational outcomes as Matlab? We review in depth the basic operations of Matlab, such as solving a system of linear equations, computing eigenvalues and eigenvectors, and plotting in two-dimensions. In addition, we research a more complex test problem resulting from a finite difference discretization of the Poisson equation in two spatial dimensions. Then, we compare the results we receive from GNU Octave, FreeMat, and Scilab to our previously found Matlab results. All packages gave identical numerical results, though Scilab exhibited a limitation in the size of the linear system it could solve in the complex test problem and FreeMat was hampered by the lack of a conjugate gradient function. The available graphical functions differ in functionality, but give equivalent plots, though FreeMat has limited three-dimensional graphics capabilities at present. Overall, we are able to conclude that GNU Octave is the most compatible with Matlab due to its numerical abilities and the similarity of its syntax.