inconsistency between the documentation and implementation of use_ndim

[aras-y]

For multivariate time series (use_ndim=True), there is an inconsistency between the documentation and the implementation in the cost computation.

The documentation and the docstring says Euclidean distance is used, but the implementation (both in the pip (2.3.10) and Github versions) uses squared Euclidean distance by default (as it should be, for consistency with the squared difference in the univariate case).

So, I suggest the documentation to be updated to the squared Euclidean distance.

[kingwe-stack]

That's right,I also found that.I suggest to add Euclidean distance calculation function .

[wannesm]

Fixed by @aras-y in 4b23b99da4400025e52804bf683a3146a721b1d3 .

Euclidean distance is supported using the inner_dist argument.