AttributeError: module 'ot' has no attribute 'weak_optimal_transport'

[zikai1]

Hi Alexander Korotin,

Thanks for your great work! I am quite interesting in your work that utilizes neural network to learn optimal maps and plans.

When I run the illustration part in demo NOT_toy_1D.ipynb, the following error emerge:

It seems that the function weak_optimal_transport in module OT is missing. Could you please help solve the above error? Thank you very much.

[iamalexkorotin]

Hi,

it seems like they changed the structure of the POT library which is used to get the discrete OT solution in the plotting code.

To fix the issue, in src/plotters.py, you should change
ot.weak_optimal_transport
to
ot.weak.weak_optimal_transport.

Hope this helps!

Best wishes,
Alex

[zikai1]

Hi,

it seems like they changed the structure of the POT library which is used to get the discrete OT solution in the plotting code.

To fix the issue, in src/plotters.py, you should change ot.weak_optimal_transport to ot.weak.weak_optimal_transport.

Hope this helps!

Best wishes, Alex

Hi Alex,

Thanks! As suggested, I have changed ot.weak_optimal_transport in src/plotters.py to ot.weak.weak_optimal_transport, as shown in the following. But the above issue still exists.

Do you know why? Maybe I should ask for POT for some solution.

Best wishes,
Mingyang

[zikai1]

Hi,

it seems like they changed the structure of the POT library which is used to get the discrete OT solution in the plotting code.

To fix the issue, in src/plotters.py, you should change ot.weak_optimal_transport to ot.weak.weak_optimal_transport.

Hope this helps!

Best wishes, Alex

Hi Alex,

I have successfully solved the above issue. The problem is that the POT version POT==0.4.0 in requirement.txt is a little bit old.

I have updated it to pip install POT==0.8.2 and the plotter.py works.

Thank you again for you work!

Best wishes,
Mingyang