Manuel Bärenz

Ok, if the switching is memoryless, then this sounds a lot like a Poisson process where we throw an exception for every event. See https://github.com/turion/rhine/pull/218, it should be possible to...

I might have just fixed it, the variance looks a lot better now. The issue was again that I forgot that after a resampling step, all particles ~are already normalized~...

Consider adding a `Frames` clock as well (or instead): https://hackage.haskell.org/package/Frames

I've observed runtime errors with this approach in the wild. Possibly, the way I'm reading the file here isn't that good, and instead I should use the streaming interface https://hackage.haskell.org/package/cassava-0.5.3.0/docs/Data-Csv-Streaming.html#g:2...

> I've observed runtime errors with this approach in the wild. I definitely need hardened tests that would have detected these errors.

It seems 1.2 failed because it didn't support GHC 9.6. 1.3 went through fine. So this is not urgent, but still it would be great to do this.

Thanks for finding this! I can sort of reproduce, but I'm getting a different (yet still incorrect) scheduling: ``` ghci> runScheduleIO $ flow $ (f300 +@+ f500) @>-^ arrM (liftIO...

I have an inkling that this might be an upstream issue in https://github.com/turion/monad-schedule. Let me investigate.

The problem is not related to the clock structures, since it can be reproduced with automata: ``` ghci> a300 = Data.Automaton.constM (wait 300 >> pure 300) >>> accumulateWith (+) 0...

Which versions of `rhine` and `monad-schedule` are you using? (I'm using `rhine-1.4.0.1` and `monad-schedule-0.2`)