13.07.2015 Views

A Proposal of a Mathematical Roadmap for Scilab DRAFT - Projects

A Proposal of a Mathematical Roadmap for Scilab DRAFT - Projects

A Proposal of a Mathematical Roadmap for Scilab DRAFT - Projects

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

13 ConclusionThis is a rough, unfinished, incomplete, filled with spell checking and basic errors,draft <strong>of</strong> a proposal <strong>of</strong> a mathematical roadmap <strong>for</strong> <strong>Scilab</strong>.We have analyzed <strong>Scilab</strong> by decomposing it by fundamental mathematical domains,including elementary and special functions, sparse linear algebra, optimization,modelization, statistics, design <strong>of</strong> experiments, geometry, partial differentialequations, neural networks, dense linear algebra, ordinary differential equations andcontrol.The problems associated with completing this document is nontrivial. Indeed,defining the current state <strong>of</strong> <strong>Scilab</strong> in each <strong>of</strong> its computational domains is a difficulttask. In order to define the features to update, we need to determiner their per<strong>for</strong>manceand their accuracy: this is unknown to us <strong>for</strong> many features. By per<strong>for</strong>mance,we mean the mathematical per<strong>for</strong>mance <strong>of</strong> the algorithms (that is, their state-<strong>of</strong>the-artnature) as well as their CPU speed. Obtaining fast computations requires touse efficiently the s<strong>of</strong>tware and hardware currently available (e.g. multi-core architectures).Moreover, gathering the user feedback is difficult, perhaps because <strong>of</strong> theopen source and free nature <strong>of</strong> <strong>Scilab</strong>. We emphasize that the user feedback wouldbe a very efficient way <strong>of</strong> driving <strong>Scilab</strong>, since some users have a strong expertise intheir particular domain.In an hypothetical context with unlimited time and unlimited task <strong>for</strong>ce (withunlimited expertise), the following sequence <strong>of</strong> steps may be followed, in that order.1. Create or update technical reports describing the current state <strong>of</strong> <strong>Scilab</strong>. Thegoal <strong>of</strong> these reports would not be to analyze the features in great detail, butonly to give an overview <strong>of</strong> the existing features, the current implementations,there per<strong>for</strong>mance and accuracy. This includes the test <strong>of</strong> the current implementationsin order to know their per<strong>for</strong>mance and accuracy.2. Update this roadmap and take decisions.3. Update the existing features which may provide poor per<strong>for</strong>mance or pooraccuracy.4. Create new functions which may provide missing features.In order to updating existing features or to create new ones, we have to find (andselect) a open-source libraries providing the feature under study, or to develop ourselfthe missing feature. This may be based on the internal work <strong>of</strong> the Consortium orthe work <strong>of</strong> external contributors. These tasks are not straight<strong>for</strong>ward.Driving the roadmap might be driven by:• the existing use cases provided by the users,• the will to explore new use cases.The reason behind this is that choosing based on the only proposals <strong>of</strong> the Consortiumis a somewhat impossible task, leading either to arbitrary choices (may bebecause <strong>of</strong> limited knowledge), or to a large and heterogeneous list <strong>of</strong> existing bugreports to fix or new mathematical domains to explore.21

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!