User:Urathai: Difference between revisions

*: My main interest lies in improving the support for doing validated numerics in Octave. I think that validated numerics will grow, we get access to more and more computing power and  we can in many cases shift our focus from quickly generating results to generating guaranteed correct results. The interval package is intended for exactly this and I would like to improve on that. I am also interested in implementing Taylor arithmetic, a method for calculating arbitrary order derivatives, in Octave which I think would be a addition for the interval package in particular. I have implemented a proof-of-concept for how this would work [https://github.com/Urathai/octave-taylor-POC]. I would be very work with this during GSoC but I am not sure if it is reasonable to implement a whole new Octave-forge package during GSoC and I am not sure who could mentor such a project, if this is interesting please let me know.

*: As mentioned I am also very interested in improving the interval package. There are several possible projects listed and I could work on anyone of them. The one I have thought most about is about implementing support for N-dimensional vectors. This would be a very important step to allow for, for example, implementation of higher dimensional Taylor arithmetic. It would also make the transition from non-validated to validated numerics easier which I think is very important. Below is a rough time line for this project.