Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Remove unnecessary variables, clean up warp_image #625

Closed
wants to merge 1 commit into from

Conversation

ZackAttack614
Copy link
Collaborator

Cleaned up some of the logic in warp_image directly. Expected speedup ~4-5%

@ZackAttack614 ZackAttack614 requested a review from psavery April 2, 2024 20:37
Comment on lines +236 to +238
gvec_angs = np.vstack([
self.angular_grid[1].flatten(),
self.angular_grid[0].flatten(),
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
gvec_angs = np.vstack([
self.angular_grid[1].flatten(),
self.angular_grid[0].flatten(),
angular_grid = self.angular_grid
gvec_angs = np.vstack([
angular_grid[1].flatten(),
angular_grid[0].flatten(),

Since self.angular_grid is a computed property, you will probably get better performance by adding back the variable so it is only computed once.

@@ -233,42 +233,35 @@ def warp_image(self, image_dict, pad_with_nans=False,

"""

angpts = self.angular_grid
dummy_ome = np.zeros((self.ntth*self.neta))
gvec_angs = np.vstack([
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Moving gvec_angs outside the loop is a great idea here.

@@ -233,42 +233,35 @@ def warp_image(self, image_dict, pad_with_nans=False,

"""

angpts = self.angular_grid
dummy_ome = np.zeros((self.ntth*self.neta))
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Removing unnecessary variables isn't always helpful, because they can help with readability (in this case, dummy_ome indicates that we are making dummy omega values).

xypts = np.nan*np.ones((len(gvec_angs), 2))
valid_xys, rmats_s, on_plane = _project_on_detector(*args,
**kwargs)
valid_xys, _, on_plane = _project_on_detector(*args, **kwargs)
Copy link
Collaborator

@psavery psavery Apr 4, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could we just pass valid_xys into the interpolation function directly, rather than setting xypts[on_plane, :] = valid_xys below and passing xypts? It appears that the interpolation functions ignore invalid xys, and only operate on a subset of the valid xys (in their functions, they check which xys are on the panel, and ignore xys not on the panel, and they also apply the panel buffer mask to remove even more xys).

Can you verify that you get the same result that way? It might be faster also...

More thoughts: the interpolation functions call clip_to_panel(xy, buffer_edges=True), which consists of two parts:
~~ 1. Removing any xys that are not on the panel ~~
~~ 2. Applying the panel buffer mask to remove any masked xys

Since we already know that valid_xys are on the panel, it would be nice if we could just skip the check for whether the xys are on the panel. But we still need to apply the panel buffer mask (it hasn't been applied yet). Maybe we should make a separate way to apply the panel buffer mask so we can skip the extra check that clip_to_panel() performs... but only if we verify that the result stays the same.

Correction: it looks like the original valid_xys are on the detector plane, but not necessarily on the detector itself. So we do actually need to keep the clip_to_panel() call.

However, I think it still might be better to just pass the valid_xys into the interpolation function directly, since we don't need to pass in any xys that are not valid (i. e., nan).

@ZackAttack614 ZackAttack614 deleted the warp-image-speedup branch July 11, 2024 16:36
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants