Romain Tavenard

On this point, I can just inform about the choice made for [tslearn](https://github.com/rtavenar/tslearn) (probably not ideal). Our time series are 3d numpy arrays (to allow for multidimensional time series). As...

@nilesh05apr great! Feel free to open a PR and ask for help at any point!

Hi @tgermain25 Can you `pip install numpy` ? It seems it's the numpy install that fails, not the `tslearn` one.

I would say the main constraint we have on `numpy` version is that it should match the constraints set by `numba`, so maybe a more recent `numpy` could be used.

Hi @wzpy Thanks for your feedback. This would indeed make sense. We should add it to our TODO list. However, I cannot tell when it can be added to `tslearn`...

Hi @blafabregue Thanks for your hints. Have you been able to reproduce the published results for fixed values of the hyper-parameters?

Hi @brk21 Thanks for pointing out this deprecation warning! It seems to me that it is not even necessary to cast the list into a numpy array, or is it?...

I would like to avoid the `object=True` since our algorithms in `tslearn` are supposed to operate on datasets represented as 3d arrays, which is no longer the case when arrays...

> @rtavenar I would be strongly opposed to that because one of the main values of `softdtw` is that it can operate on time series of different lengths! This `ValueError`...

Hi @londumas Technically speaking, this would probably be feasible. If anyone has time to work on this, feel free to suggest help, I'd be glad to review a PR on...