Manuel Bärenz
Manuel Bärenz
That implementation seems to accumulate all values it ever evaluated, so I can't adapt it directly. I need to find out how to implement Gaussian processes without growing memory indefinitely.
Another great example would be adding a kind of CRDT constraint on the internal state. Then it is conceivable that we could run two copies of the same program on...
Thanks a lot! This is awesome!
The test failure may simply be the Github MacOS runner. For some reason, some concurrency and timing stuff from monad-schedule doesn't work on MacOS so well.
I added a few commits to document everything. I needed this in order to fully understand what's going on. I believe I have it now. In case you can spot...
Fantastic! Again, thanks a lot, this makes rhine development a lot more productive. And I've learned a lot along the way.
> Now it seems reasonable to want inferenceRhine to tick at a significantly lower rate than modelRhine, since it is more computationally intensive I doubt this assumption. Let me try...
In such a situation, when we really don't have the computational power to process all observations, I guess we have to throw some away. This is what's done with the...
I can definitely see the motivation. Still, when I try to incorporate a measurement from a past timestamp, then it is about conditioning the state of that time. The straightforward...
As expected, this fails before https://github.com/turion/rhine/pull/343 is merged.