[Help] Can't get the right model for 2d system about pysindy

[antono0106]

Hey again @Jacob-Stevens-Haas , thanks for the answer and sorry for bothering you before Christmas.

1. Yes, I tried to do this. Even tried to remove a denominator, but no success here.
2. Ah, I see. Swapped, you're right. Anyway, plots are not the issue =/
3. Thx, agree
4. Tried with different threshold and tol. Even tried to swap SINDyPILibrary on PDELibrary, but no success here too.

I think that in case when there is no "jump" ability using this library, such a way is not working for me.

Thank you again! Have a good holidays! May all your days be merry and bright this season.

[antono0106]

@Jacob-Stevens-Haas sorry for disturbing, could you please help me to see what’s wrong with my code in order to resolve the issue? Thank you.

[Jacob-Stevens-Haas]

Hey @antono0106, I'd love to. I'm going to be spending a bit less time on github over the holidays, however, and SINDy-PI is the part of the codebase I'm least familiar with. So I'm going to ask a bit more rather than diving fully in:

1. Can you get SINDy-PI to work with your code, but on a simpler system? E.g. with $\epsilon=0$, or fewer terms?
2. did you mean to plot x1 equation as model X and x0_dot as model_derivative_X'? Likewise, looks like you swapped x0 and x1 in your system and system derivative
3. plt.legend() and use a color other than yellow so I don't have to scroll between the plot and the code to read the plot.
4. You have models with the correct terms, they just have too many other terms. Increase your optimizer threshold. There's no great way to "jump" to the correct model in SINDy-PI AFAIK.