Made LinearOperator independent of spLinearOperator

[rohanbabbar04]

For issue #328

• Added _AdjointLinearOperator, _TransposeLinearOperator and _ProductLinearOperator
• Added methods __matmul__,__rmatmul__ , transpose and adjoint
• Made all classes extend the LinearOperator

Still need to change/update the docs, if code seems fine...

[mrava87]

Thank you!

I'll try to review this in coming days and let you know :)

[mrava87]

Here are some comments. I have also made a commit where I propose changes according to my comments.

@rohanbabbar04, feel free to go over and let me know what you think.

[rohanbabbar04]

Thank You @mrava87 for reviewing my PR, I have updated the docs and agree with all your changes.
Do take a look and give your opinion

[mrava87]

Thank You @mrava87 for reviewing my PR, I have updated the docs and agree with all your changes.
Do take a look and give your opinion

Will do soon :)

[mrava87]

I just identified a problem with describe due to some of the internal linop methods being moved from scipy to native pylops ones...

For example:

A = pylops.MatrixMult(np.ones((10, 5)))
A.name = "A"
C = pylops.MatrixMult(np.ones((10, 5)))

# Sum
D = A + C
describe(D)

now returns A where: {'A': '_SumLinearOperator'} instead of A+M where: {'A': 'MatrixMult', 'M': 'MatrixMult'}.

Currently investigating why...

[rohanbabbar04]

Is the bug fixed?

[rohanbabbar04]

Just tested the code
Now the describe works fine

A + M
where: {'A': 'MatrixMult', 'M': 'MatrixMult'}

[rohanbabbar04]

Just tested the code Now the describe works fine

A + M
where: {'A': 'MatrixMult', 'M': 'MatrixMult'}

So I am removing the docstrings to make the Codacy Static Code Analysis successful
Removing this and making a commit

[mrava87]

Yes the code is fixed :)

[rohanbabbar04]

Yes the code is fixed :)

@mrava87 Made the commit all checks are now successful...

[mrava87]

I made some extra changes as you completely removed any mention to scipy LinearOperator, which instead is worth still mentioning (also because we can still mix our operators with theirs - and with their solvers!)

[mrava87]

@rohanbabbar04 I am satisfied with the current status of the PR.

Since we are making changes to a core method of PyLops, I want to get the feedback of another core developer before merging.. please be patient, we will let you know soon when this is ready to go :)

[cako]

I'll be trying this patch very soon! Sorry for the delays!

[cako]

This is great stuff. Looks good to me!!