Extract common functionality from SINDy to _BaseSINDy
#562
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This will eventually enable SSSindy.print, as well as a simpler discrete and SINDyPI class
SINDy.equations(), equations(), print_model() were written with extra generality that never ended up being used. This commit removes such generality and throws away these indirected helper functions in favor of a single method.
Jacob-Stevens-Haas
force-pushed
the
simultaneous
branch
from
5182f24 to
79f43bb
Compare
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #562 +/- ##
==========================================
- Coverage 95.28% 95.26% -0.02%
==========================================
Files 38 39 +1
Lines 4069 4099 +30
==========================================
+ Hits 3877 3905 +28
- Misses 192 194 +2 ☔ View full report in Codecov by Sentry. |
Jacob-Stevens-Haas
force-pushed
the
simultaneous
branch
from
79f43bb to
b7eac64
Compare
Also: * fix sphinx cross-linking error * remove unused print_model
Jacob-Stevens-Haas
force-pushed
the
simultaneous
branch
from
b7eac64 to
2b3871b
Compare
|
Hey Alan, I put you as a reviewer in case you want an eye into this process. This is the beginning of what we talked about in the last meeting. @AHsu98, @MalachiteWind, I'm going to add you as collaborators so you can review these changes too. |
Jacob-Stevens-Haas
force-pushed
the
simultaneous
branch
2 times, most recently
from
98a475a to
0482105
Compare
This includes things like reduce() and coef_
This was thought to be easy, because in many cases jax arrays were an almost drop-in replacement for numpy arrays. However, they are far less amenable to subclassing. Why does this matter? The codebase gained a lot of readability with AxesArray allowing arrays to dynamically know what their axes meant, even after indexing changed their shape. However, extending AxesArray to dynamically subclass either numpy.ndarray or jax.Array is impossible - even a static subclass of the latter is impossible. Long term, we will need our own metadata type that carries around an array, it's type package (numpy or jax.numpy or cvxpy.numpy), its bidirectional mapping between axis index and axis meaning, and maybe even something from sympy. Short term, we should expose our general expectations for axis definitions as global constants. This is still error prone, as the constants are incorrect for arrays that have changed shape due to indexing, but will be far more readable than magic numbers.
Jacob-Stevens-Haas
force-pushed
the
simultaneous
branch
from
0482105 to
22329dd
Compare
|
I'm going to merge it, and turn the last commit into an issue |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Begin #351
This pull request begins the abstraction of the
SINDyclass, which handles a lot of methods with special cases, to a variety of classes that reflect fundamental differences between problem formulations. Here, I pull out_BaseSINDy, which contains basic information about the feature library, system dimension, and printing the equations.Future PRs will make Weak and discrete their own classes, modify feature libraries to be amenable to jax, numpy, and cvxpy arrays, and introduce the Single-Step SINDy class and associated optimizer/problem setup.