The MPScilab toolbox provides a datatype and functions to perform
arbitrary precision matrix arithmetic. This release of MPScilab should be
considered a preview and only basic functions are implemented. Bugs and features
requests can be filed in the Scilab forge issue tracker for the project :
Main features :
The toolbox introduce a new arbitrary precision full matrix datatype. By using
GMP and MPFR as the computing backend the toolbox also provides a set of basic
functions to manipulate this new datatype. Most functions are able to mix
multi-precision data with the native Scilab matrix of double.
Overview of available functions :
- Basic element-wise addition, subtraction, multiplication and division
- Element wise exponential, square root etc...
- Matrix transposition, multiplication and inversion
- Trigonometric and hyperbolic functions
- Basic comparisons and min/max
A short tutorial is included in the online help available from within Scilab.
WINDOWS NOTE : On Windows the Microsoft Visual C++ 2010 Runtime may be required
to run the toolbox.
IMPORTANT NOTE : Even if written as such it is not possible to install the
toolbox automatically by running atomsInstall("MPScilab"). To install MPScilab
from the binary distribution one must first download the correct file for the
targeted platform. Then from within Scilab the module can be install by running
atomsInstall with the compressed file as argument.
For example on 64 bit Linux :
Summary of the changes in version 0.2 :
The full list of changes in version 0.2 is available in the NEWS file of the
The main change in this version is the possibility of calling mps functions with
the output argument on the left side like this :
rop = mps_sin( op )
Calling mps functions with an explicit output argument like in version 0.1 is
still possible as is the overloaded functions. In addition to this major change
about 20 new functions are available. Finally the argument checking and
validation was revised for most functions to be more complete and consistent.