Cosqueric nicolas

3rd year CFD student

MASTERING INDUSTRIAL CODES AND PARRALELISM

Scilab Instalation and Evaluation

Introduction

Installation

Evaluation

Overview of SCILAB possibilities

Conclusion

Scilab is a scientific software package for numerical computations in a user friendly environment. It was developed by the INRIA, a french research laboratory. ( www.inria.fr ) It is a free software (no lisence for utilisation). The applications are various. Indeed, it computes the following problems : elaborate data structure, polynomials, matrixes, programming, graphics 2d and 3d, interfacing with C or Fortran routines, linear algebra, signal processing ...

Moreover, it an evolutive code because user can create functions and send them to the INRIA which distribute them to all users. It is called contributions. There are a lots of contributions which perform computations in the following fields : finite elements...

The purpose
of this report is to explain installation of scilab and evaluate it. However,
the possibilities of SCILAB are so huge therefore, we can only give an
overview of this software

Scilab is available
for windows and linux machines. It is distributed in source code format.
You can obtain it at the following adrees : ftp.inria.fr where you can
also get the current contributions.

In this course,
SCILAB was installed on linux machines. The file README that you can find
in the SCILAB package explain the installation and the compilation.

At the beginning, the use of SCILAB is pretty difficult. If you allready know MATLAB, it is easier because there are lots of similarities.

A good way for practise SCILAB is to realize the example of the INRIA, which you can find at the adress :

www-rocq.inria.fr/scilab/doc/demos-html

Indeed, this demos are well made and it permits to you practise the manipulation of vectors, matrixes, functions, polynomials, graph and file management.

Then, you have the scilab help on line, which contains a description of all the primitives which are available. The help is available from SCILAB windows and is composed of two parts. Bellow the help window, you have the main fields such as polynomials, programming and above, you have the main functions of the corresponding field. Moreover, you have a box dialog in the bottom of the help window where you can write a key word and get the corresponding information. For all variables, there is a description of the syntax and of the input/output variables. Moreover, most often the time, there are some examples of utilisation.

Bellow, you
can find a presentation of the main functions which have been tested in
this course. The conclusion shows demonstration of the possibilities of
this software.

**Matrixes and polynomials**

In SCILAB,
there is two class of variables : matrix and polynomial. Indeed, constant
and vector are defined as matrixes. There are lots of predifined functions.
For example, inverse or diagonalise a matrix, solve equations or linear
systems, integrate or derivate polynomials or functions. The manipulation
of these objects and functions is relatively easy.

**Graphics**

The graphic interface allows to vizualize 2d and 3d plot. Graphic are based on matrixes. There are lots of predifined functions and their utilisations require a certain training.

Example of graphs :

**Programming**

SCILAB is not a programming language such as C++ or FORTRAN. Nevertheless,
you can define functions and all the classical instructions such as conditionnals
or loops are available. There is not a text editor as in MATLAB but you
can write functions in file with a text editor such as Xemacs or VI. Your
file must have the extension .sci. To compile, you must use the function
getf(myfile.sci). It is very usefull for define small program. You can
also write a series of instructions.

**Data processing**

** **SCILAB
disposes of primitives which allow to achieve write or read operations
in file. At the beginning, it is pretty difficult because there a lot of
primitives and some primitives are reserved for binary file and the other
for ASCII file.

**Signal processing**

All the classical instructions are available such as fast fourier transform.
Moreover, you can get the equivalent of Simulink of Matlab : SCICOS that
you can get at the inria web site.

**Comparison with MATLAB**

** **There
is a lot of similarities with Matlab. Moreover, there is a primitive on
SCILAB, mfile2sci() which translate m-files in sci-files. Nevertheless,
this function is limited because it does not recognise all the function
off matlab and strong skills of matlab are required.

You can find this results in scilab, in the section demo.

Finite elements

Advanced graphics

As a conclusion, we can say that
this sofware offers strong possibilities.** **The utilisation is pretty
easy for simple problem. For advanced problem, it is more complex. Moreover,
you can access to introduction course on internet. But for advanced functions,
it is more complex for find books or article. Moreover, there is some books
and user manual but there are paying and expensive. There is a lot of similarities
with Matlab. The graphic interface on scilab is less advanced : it is a
simple interface, less developed than Matlab. For example, computation
on matrix have similar syntax. Nevertheless, some functions proposed in
scilab are limited or complex for a beginner. For example, the matlab to
scilab translator is not very efficient and require strong skills of matlab.
Some functions such as graphics or data interfacing are complex at the
beginning.