Mahendra Mariadassou

Hi @Bastien-mva, the checks are not valid due to a parsing error when reading the error message. It probably has to do with nested quotes (consider using `"` for outer...

Hi, Thank you for your interest in PLNmodels. There are actually two backends used for the optimization part: nlopt (default) and torch. With the code you used, you relied on...

You can in fact completely avoid the inception by choosing a smaller learning rate for the torch backend to solve the convergence problem, leading to a simpler and more elegant...

`nlopt` doesn't always converge. For example, we encountered instances where `torch` (with default parameters) converged while `nlopt` didn't. It all boils down to the optimization strategy (starting point and optimization...

Hi, for PLNPCA it's a bit involved but the crux is here: https://github.com/PLN-team/PLNmodels/blob/2657e506675ad26e77df94f9c40f02ae11aa1042/R/PLNPCAfamily-class.R#L41-L53 For PLNPCA, we have fit a family of models (one per number of dimension in the tested...

Hi @papayoun, thanks for finding the bug. I'll have a look into it (it's probably due to the "non standard" formulas used in ZIPLN). Can you just check something for...

For grobs, `vdffir::expect_doppelganger` returns a success when run via `devtools::test_file()` in the console but a failure when run via the *Run Tests* button or via R CMD CHECK...

Hello, Thank you for reporting this bug. This seems to arise from automatic conversion of one-column matrices to vectors but I'm not quite sure where it happens nor why it...

Dear user Thank you for your interest in `PLNModels`. I would like to point out that the PLNmodels is built for *count data* as it relies heavily on the Poisson...

I was indeed a bit hasty. Let me investigate and get back to you.