Just a question: Would anyone have a clue, or link – not the same thing Wink –, whether and how machine learning has been applied to provide a ‘solution’ or counter-example to the smoothness of 3D Navier-Stokes equations ..? As here, and here (and onward!) in particular.
Given the pattern of extreme sensitivity of starting parameters for the ‘outcome’ predictions of ‘full’ Navier-Stokes equations (for the moment abstracting away from this…), does this somehow mean that the ‘unpredictability’ of convergence of neural networks, is a neural network to be trained and started off with some random weight seeding maybe just a somewhatto a degree yet to be established on the range of 0-1 or 1-∞-simplified N-S equation ..? #askingforafriend of course.
The methodological congruities are just too beautiful to miss, and just think about how the ‘solution’ or heuristics achievements throughout the decades, might translate!
Just out of interest. In return:
[The clearness of the cuts helps, contrastingly; Strasbourg off some side street]