Christopher Rackauckas
Christopher Rackauckas
I went into the Conda.jl folder and directly ran that Python and got: ``` Python 3.8.5 (default, Sep 3 2020, 21:29:08) [MSC v.1916 64 bit (AMD64)] :: Anaconda, Inc. on...
Agreed on this. Some kind of remake functionality could improve the robustness of pass writing.
https://mtk.sciml.ai/dev/basics/FAQ/#Getting-the-index-for-a-symbol Just grab the index and use it in the loss function.
Yeah I think the issue is that we need a way to support differentiation with observed functions. The non-observed case can always use indices so that's fine. Observed functions are...
@YingboMa how is it able to run with the wrong number of params? Is that inbounds stuff?
Where is that allowed? It's not on purpose.
Yes, the keyword arguments are splatted onwards.
The default in the jump stuff should probably disable it because yes that would have terrible performance.
1. is sparse truly better here? 2. autodiff=false shouldn't matter if jac=true? 3. Did you try FBDF? That's close to CVODE/IDA 4. Which linear solver? Did you try KLU? 5....