David Métivier

Results 19 comments of David Métivier

Hi, I am just curious do you have a reference for this type of HMM? So if you observe a chain of size T, you assume that the transition matrix/emission...

Thanks for the answer. I don't know if a separate add-on is needed or not. Inhomogeneous chains (and periodic ones) are definitely interesting for me!, so thanks for your work....

Thanks! You probably are working on that, but I wonder, how do you plan to manage multiple observations of the same chain of length `T`. Would you add the multiple...

I did that [PeriodicHiddenMarkovModels.jl](https://github.com/dmetivie/PeriodicHiddenMarkovModels.jl) based on `HMMBase.jl` package. Here, the transition matrices `A(t)` and emission distribution `B(t)` are periodic over `t=1:T`. One observes `Y` over several periods `n=1:N`. This can...

I think that in ``` choice,state=rand(hmm, Ntrials, seq = true) #create synthetic data ``` it should be ``` state,choice=rand(hmm, Ntrials, seq = true) #create synthetic data ``` where `choice` are...

It is used in the test function of the [QuasiMonteCarlo.jl](https://github.com/SciML/QuasiMonteCarlo.jl) package, (see [here](https://github.com/SciML/QuasiMonteCarlo.jl/blob/2dce9905e564a85e1280115cc8af071674fc7d80/test/runtests.jl#L34)). It provides a convenient way to build the “elementary intervals” used in the definition of some low...

I agree with your points 1) and 3). I am not sure why a lot of package do like that. It feels like the literature (including a lot of paper...

On Windows with` pyplot()` and heatmap, I get the same problem with `savefig ` in pdf (and not in png). Does someone have a solution for this issue?

I ran into the same problem with the current official [doc tutorial](https://fluxml.ai/Flux.jl/stable/tutorials/2021-10-08-dcgan-mnist/#What-are-GANs?). After a lot of investigation (and being unfamiliar with Flux and API changes), I figured that the code...

Yes it does solve the problem! I tried the code. In particular, it includes change in API like `setup` optimizer and `update(model,...)` not `update(params, ...)`. I am not sure what...