Number of components in case of unknown number of sources

[turovapolina]

Dear @akcarsten ,

Thank you very much for your ICA tutorial, it was very helpful for me!
I just have small question: how do you think what can be done in case if there is no exact information of number of sources in data? Probably you are familiar with any methods to calculate a relative error for each number of component?
What I have found is only this article: https://www.sciencedirect.com/science/article/abs/pii/S1386142516301585 and authors suggested this equation:

but unfortunately I haven’t find any implementation of these calculations on python or any other solutions.
Have you ever had such problem or may be you can recommend me something?
Thank you very much in advance!

Best regards,
Polina