set up acceptance tests for toolbox-to-be-developed

psycharo-zz / factor-graph

matlab/c++ factor graph framework

32 stars 8 forks source link

set up acceptance tests for toolbox-to-be-developed #10

Closed ghost closed 12 years ago

ghost commented 12 years ago

Formulate the result of evaluation of SumProductLab as a set of acceptance tests for the toolbox to be developed.

psycharo-zz commented 12 years ago

By a set of tests you mean requirements? E.g. "The framework should support multivariate Gaussians"?

ghost commented 12 years ago

Yes. At this moment we only have the one-page project proposal to describe what we will develop. I thin it woudl be good to revisit the original plan after having played with SumProductLab (SPL) and rewrite the plan as a set of requirements for the toolbox to be developed. Eg,

As a user,

I want to design factor graphs in matlab notation
I want to execute factor graphs in real-time on a PC
The framework should support multivariate Gaussians (Kalman filters, etc)

etc.

If this turns out to be a clear extention of the SPL toolbox, we can call it SPL-4.0 (check teh license); otherwise we have to think of a new name for the toolbox (any suggestions?)

psycharo-zz commented 12 years ago

I don't think that it will be an extension of the SPL. The name doesn't matter for me, factorgraphlib would be ok. btw, currently I am developing a simple C++ version of sum-product, in order to understand which language constructions and datastructures might be useful. Later this code can be mimicked in matlab, or if that's impossible used inside of a wrapper. To say, even if matlab was able to convert SPL to C-code, it would be really inefficient (there are lots of array manipulations that look nice in matlab but in C they would be a bottleneck for sure).

ghost commented 12 years ago

OK, sounds great. See you tomorrow.

ghost commented 12 years ago

I've started a user acceptance tests document in dropbox (doc/DeVries - Oct2012 - Factor Graph Acceptance Tests)