Starting with the ancora project

I’ve finally done it: I created the public repository for ancora, my new side project! Of course, ancora is not CORA, but my aim is to implement at least all the set operations that can be solved in a safe way.

Because that’s what ancora is supposed to do: solve numerical problems with hard guarantees, even accounting for floating-point errors at every step of the way.

Now, this paradigm will only be one facet of ancora: since my colleagues at TUM are currently deciding on what language to move CORA to (current winner is C++), to convince them C is the way I will also implement a “fast” version of ancora, which has fewer guarantees but is significantly faster. This will make ancora applicable in the context of machine learning, but the main aim will still be to develop the “safe” version further.

Right now, the repository is quite empty, but stay tuned - if I’m fast enough, this week could see the first real version of ancora.