Difference in results between tslearn and DBA.py implementation

[filehelp]

I've been trying to use the average DBA function provided by tslearn to compute the average of a set of time series. However, I've noticed that the results obtained with tslearn's implementation and with the original DBA.py module by Francois Petitjean (https://github.com/fpetitjean/DBA/blob/master/DBA.py) seem to differ. I've attached graphs that illustrate the discrepancies. I've used the exact same data for both implementations.

I'm unsure whether this is due to a bug in the tslearn code or if I might have missed some modification that's required for the tslearn implementation. Could someone help me investigate this issue, please?
Thank you.

[charavelg]

Would you mind sharing some code and data? It would be useful to investigate your problem.

Also, note that the tslearn.barycenters package provides a dtw_barycenter_averaging_petitjean implementation.