[ Home Page ] [ Eiffel Archive ] [ Eiffel Classes and Clusters ]

Numeric cluster 0.2

Written by Nicolas Waquier (nicolas@esclare.fdn.org)

11 January 1995.
Targets TowerEiffel 1.3.1

numeric-0.2.tar.gz (7731 bytes)

Licensing

You can use/modify it for your own use, but not redistribute your modifications as a library/class/cluster.
No restriction about executables.

Status

Will be enhanced.
Please report every problem.

matrix/ -- matrix of doubles

MATRIX matrix of doubles

LINEAR_SYS linear system to be resolved with gauss method

MATRIXER an ancestor to both matrix and vector

VECTOR a vector of doubles

M_TEST test the cluster

matrix_num/ -- matrix of rationals

MATRIX matrix of rationals

LINEAR_SYS linear system to be resolved with gauss method

MATRIXER an ancestor to both matrix and vector

VECTOR a vector of rationals

M_TEST test the cluster

number/ -- rational numbers

RATIONAL

N_TEST test the class

Note

The algorithms are not !! optimal.

Please buy specialized books if you need such algorithms. (Or send me new versions of the classes.)

Target Use

The target use I had in mind while developing was a class of matrix to do spatial transformation of point/line/planes. You create a identity matrix, then rotate, translate it.

I had to develop it in C for a medical application, and wanted the same in eiffel.

Future Plans

clean up the classes
add dantzig algorithm for inequation linear systems
replace rationals (integers) with rationals (factorial) - see cluster factorial
make it interoperate with different compilers (let me know if you've modified the classes to achieve this!)
format the output more attractively depending on the width of the columns in matrix_num.

Requirements

Compiler: Tower Eiffel 1.3.1.0
Clusters: class FORMAT from the tower/.../sig cluster
Time to compile: number/ 09 sec Eiffel -> C
matrix/ 18 sec Eiffel -> C
matrix_num/ 20 sec Eiffel -> C

Development platform

nextstep/486