Feature: new Sliding and Patch operators #578
Merged
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
Changed sliding1d from using other operators to being directly implemented (including also an option to apply Op simultaneously to all windows).
Changed sliding2d and sliding3d from using other operators to being directly implemented (including also an option to apply Op simultaneously to all windows).
Changed patch2d from using other operators to being directly implemented (including also an option to apply Op simultaneously to all windows).
Changed patch3d from using other operators to being directly implemented (including also an option to apply Op simultaneously to all windows).
|
@cako not sure if you have time to take a look at this, any comment/suggestion will be welcome 😄 |
|
Will have a look at it! Could you make this repo public: https://github.com/mrava87/pylops_slidingpatching |
|
Ops, yes :) I also moved it to the PyLops organization |
cako
approved these changes
May 19, 2024
|
@cako thanks a lot for the comments, and great to hear that you also run the benchmarks on your machine with similar results! I’ll address your comments in the next few days and then merge this PR 😀 |
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Outdated suggestions cannot be applied.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.
In this PR, I provide a new implementation for all Sliding and Patch operators that presents the following features:
_matvecand_rmatvecmethods. The old implementation, although elegant in that it relied on using other pylops operators, showed to be inefficient especially in cases with very large number of windows.Opcan be provided: the first applies a single transformation to each window separately (as default from before); the second applies the transformation to all of the windows at the same time.nprocin Sliding3D is deprecated since we do not use anymore HStack/VStack operators that have a multi-threading option. Currently, all of the new implementations are single threaded (although inside they may call numpy methods that are multi-threaded).savetaperin added to all operators: ifTrue, all of the tapers are created and saved in a numpy array so that they can be efficiently applied in one go as element-wise multiplication of the numpy arrays (this is fast but can be memory demanding); ifFalse, only unique tapers are saved and applied one by one (this is a bit slower but more memory friendly)All of the new operators have been benchmarked against the hold one, and an analysis of both the timings and memory usage of the different operators has been performed here: https://github.com/PyLops/pylops_slidingpatching