-
Notifications
You must be signed in to change notification settings - Fork 105
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
Automatic range inference of RayTransform breaks for unweighted space #1525
Comments
That sounds very shady indeed. What |
I have tried all implementations, the numbers are slightly different, but the inconsistency persists. The same happens, while using Parallel2dGeometry and FanBeamGeometry. |
What happens with 14.0? Could this be an integer issue? Are you running python 2 or 3? |
Python 3. Same problem with 14.0. I discovered this problem when I was trying to remove the influence of space cell volume on the operator norm. The dependency correspond to my theoretical considerations apart from the case when operator.domain.cell_volume = 1, which seem to be an outlier. |
Oh god damnit, now I know what it is. I knew this would happen :D @kohr-h! This is because we no longer have properly unweighted spaces, only spaces with "no weight" so to say. This is as mentioned here: I don't see how the weightings of domain and range are connected for ray transforms. They're totally different spaces. See e.g. our old conversation:
From: #1088 (comment) |
Yeah, that's one of those cases where earlier mistakes come back to bite us. I still think that "no weighting" is a broken concept and I don't regret removing it 😉 But the immediate issue is that |
I am trying to estimate the norm of the Radon transform:
The result is 11.8375811049. However, when I am changing xlim to 14.0000000000001, the result is 9.22141281723, which does not correspond to my expectations that an operators norm should be continuous with respect to the input.
The text was updated successfully, but these errors were encountered: