Add utility to configure PDHG for with explicit TGV or TV regularisation by paskino · Pull Request #1766 · TomographicImaging/CIL

paskino · 2024-03-28T13:23:44Z

Description

Add utility to configure PDHG for with explicit TGV or TV regularisation with Least Squares.

Example Usage

An example is on CIL-Demos TGV_tomography.py

Changes

adds code to create the operators and function to configure PDHG with TGV explicit regularisation
adds max to BlockDataContainer
allows dot between a BlockDataContainer and a DataContainer

Testing you performed

Related issues/links

closes Create utility to configure TV and TGV regularisation with explicit PDHG #1593

Checklist

I have performed a self-review of my code
I have added docstrings in line with the guidance in the developer guide
I have updated the relevant documentation
I have implemented unit tests that cover any new or modified functionality
CHANGELOG.md has been updated with any functionality change
Request review from all relevant developers
Change pull request label to 'Waiting for review'

Contribution Notes

Please read and adhere to the developer guide and local patterns and conventions.

The content of this Pull Request (the Contribution) is intentionally submitted for inclusion in CIL (the Work) under the terms and conditions of the Apache-2.0 License
I confirm that the contribution does not violate any intellectual property rights of third parties

samdporter · 2024-05-08T14:59:46Z

Hey all,

This still needs a few changes. Currently the BlockOperator.direct method still returns a BlockDataContainer with shape (1,1) rather than a DataContainer. I'll do a few more tests to see if there's anything else that needs doing.

Would you like me to do a PR or let you guys take care of it?

… add_TGV_utility

lauramurgatroyd

I have added unit tests for setting up TV
I would appreciate a review of the code and the unit tests for that please before I write the TGV unit tests

MargaretDuff · 2024-08-12T16:15:37Z

Hi @lauramurgatroyd - I like the style of the unit tests for explicit TV. They are comprehensive without being expensive. I have suggested a few changes to them and made some comments on the new utility file.

MargaretDuff · 2024-08-12T15:49:05Z

+from cil.optimisation.functions import MixedL21Norm, BlockFunction, L2NormSquared, ScaledFunction
+from cil.optimisation.operators import BlockOperator, IdentityOperator, GradientOperator, \
+    SymmetrisedGradientOperator, ZeroOperator
+


Can we call this file something other than PDHG.py? e.g. set_up_PDHG.py or something similar?

MargaretDuff · 2024-08-12T15:49:44Z

+        Forward operator.
+    data : AcquisitionData
+    alpha : float
+        Regularisation parameter.


Regularisation parameter for which part?

MargaretDuff · 2024-08-12T15:50:24Z

+
+def setup_explicit_TGV(A, data, alpha, delta=1.0, omega=1):
+    '''Function to setup LS + TGV problem for use with explicit PDHG
+


Need a TGV equation in here, defining alpha and beta and omega

MargaretDuff · 2024-08-12T15:52:42Z

+    delta : float, default 1.0
+        The Regularisation parameter for the symmetrised gradient, beta, can be controlled by delta
+        with beta = delta * alpha.
+    omega : float, default 1.0


Least squares uses c instead of omega, perhaps we could follow suit?

MargaretDuff · 2024-08-12T15:57:36Z

+        The constant in front of the data fitting term. Mathematicians like it to be 1/2 but it is 1 by default, 
+        i.e. it is ignored if it is 1.
+
+    Returns:


Could we add a code snippet to explain how to use the returned K and F in PDHG? Also need to explain briefly what we mean by "explicit"

MargaretDuff · 2024-08-12T16:11:18Z

+                            function = F[0].function
+
+                        np.testing.assert_equal(type(function), L2NormSquared)
+                        np.testing.assert_array_equal(function.b.as_array(), ad.as_array())


I would also be tempted to evaluate your functions on a small image, so reduce the size of the geometry to save computational cost, create a random test_data and then do
self.assertAlmostEqual(function(test_data), omega*LeastSquares(b=ad)(test_data))

MargaretDuff · 2024-08-12T16:11:52Z

+                        np.testing.assert_equal(type(K[1].operator), GradientOperator)
+                        ig = ad.geometry.get_ImageGeometry()
+                        expected_grad = alpha*GradientOperator(ig)
+                        np.testing.assert_allclose(expected_grad.direct(ad)[1].as_array(), expected_grad.direct(ad)[1].as_array())


Is that comparing the same thing?

MargaretDuff · 2024-08-12T16:13:39Z

+                        np.testing.assert_allclose(expected_grad.direct(ad)[1].as_array(), K[1].direct(ad)[1].as_array(), 10**(-4))
+                        np.testing.assert_allclose(expected_grad.direct(ad)[0].as_array(), K[1].direct(ad)[0].as_array(), 10**(-4))


I think the expected_grad sould be operator on an image geometry not on acquisition data? Thus you should not be using ad

MargaretDuff · 2024-08-12T16:14:40Z

+
+                        # F[1]
+
+                        np.testing.assert_equal(MixedL21Norm, type(F[1]))


Similarly, can you test this on an object?

paskino and others added 9 commits December 7, 2023 14:19

add init for optimisation utilities

be825a7

BlockOperator to return simple geometry if shape is (N,1) or (1,N)

e0cdf4e

updates to BlockDataContainer

a918a7f

Merge remote-tracking branch 'origin' into add_TGV_utility

b3a32d7

Merge remote-tracking branch 'origin' into add_TGV_utility

3f9b1cc

add functions to setup explicit TGV

c6b9a37

added TGV utility

f82a2c4

Merge branch 'master' into add_TGV_utility

880674a

removed not working code for SPDHG

60ca836

MargaretDuff mentioned this pull request Apr 25, 2024

BlockDataContainers with (1,1) shape #1455

Closed

samdporter mentioned this pull request May 10, 2024

Direct method of BlockOperator returns a single DataContainer if the output shape is (1,1) #1802

Merged

9 tasks

paskino added 2 commits July 24, 2024 16:33

Pull static methods out of DataOrder and make them plain functions.

f091c97

added function to create TV + LS problem

07ebfbe

paskino changed the title ~~Add utility to configure PDHG for with explicit TGV regularisation~~ Add utility to configure PDHG for with explicit TGV or TV regularisation Jul 25, 2024

lauramurgatroyd added 2 commits July 31, 2024 17:01

Merge branch 'master' into add_TGV_utility

5485fc6

Merge branch 'master' into add_TGV_utility

78dfd0c

lauramurgatroyd self-assigned this Aug 9, 2024

lauramurgatroyd added 2 commits August 9, 2024 11:07

Add unit tests

eafcfd6

Merge branch 'add_TGV_utility' of https://github.com/paskino/CIL into…

64e1c68

… add_TGV_utility

lauramurgatroyd reviewed Aug 9, 2024

View reviewed changes

Comment thread Wrappers/Python/cil/optimisation/utilities/PDHG.py

MargaretDuff reviewed Aug 13, 2024

View reviewed changes

casperdcl force-pushed the master branch 3 times, most recently from a910556 to f8b513b Compare September 2, 2024 13:11

casperdcl force-pushed the master branch from 3d0ccd4 to 967f094 Compare April 30, 2025 05:11

Navigation Menu

Uh oh!

Add utility to configure PDHG for with explicit TGV or TV regularisation - #1766

Add utility to configure PDHG for with explicit TGV or TV regularisation#1766
paskino wants to merge 15 commits into
TomographicImaging:masterfrom
paskino:add_TGV_utility

paskino commented Mar 28, 2024 •

edited

Loading

Uh oh!

samdporter commented May 8, 2024

Uh oh!

lauramurgatroyd left a comment

Uh oh!

Uh oh!

MargaretDuff commented Aug 12, 2024

Uh oh!

Uh oh!

MargaretDuff Aug 12, 2024

Uh oh!

MargaretDuff Aug 12, 2024

Uh oh!

MargaretDuff Aug 12, 2024

Uh oh!

MargaretDuff Aug 12, 2024

Uh oh!

MargaretDuff Aug 12, 2024

Uh oh!

MargaretDuff Aug 12, 2024

Uh oh!

MargaretDuff Aug 12, 2024

Uh oh!

MargaretDuff Aug 12, 2024

Uh oh!

MargaretDuff Aug 12, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants


		def setup_explicit_TGV(A, data, alpha, delta=1.0, omega=1):
		'''Function to setup LS + TGV problem for use with explicit PDHG

		np.testing.assert_allclose(expected_grad.direct(ad)[1].as_array(), K[1].direct(ad)[1].as_array(), 10**(-4))
		np.testing.assert_allclose(expected_grad.direct(ad)[0].as_array(), K[1].direct(ad)[0].as_array(), 10**(-4))

Uh oh!

Conversation

paskino commented Mar 28, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Example Usage

Changes

Testing you performed

Related issues/links

Checklist

Contribution Notes

Uh oh!

samdporter commented May 8, 2024

Uh oh!

lauramurgatroyd left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

MargaretDuff commented Aug 12, 2024

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

paskino commented Mar 28, 2024 •

edited

Loading