slimgroup / JUDI.jl

Julia Devito inversion.
https://slimgroup.github.io/JUDI.jl
MIT License
98 stars 32 forks source link

Update docs/tutorials #53

Closed ziyiyin97 closed 2 years ago

ziyiyin97 commented 3 years ago

Open a PR so that @rafaelorozco and I could check the tutorials and make sure they are up to date

codecov[bot] commented 3 years ago

Codecov Report

Merging #53 (c9bda12) into master (fef5821) will decrease coverage by 4.75%. The diff coverage is 96.15%.

Impacted file tree graph

@@            Coverage Diff             @@
##           master      #53      +/-   ##
==========================================
- Coverage   83.69%   78.94%   -4.76%     
==========================================
  Files          35       35              
  Lines        2300     2322      +22     
==========================================
- Hits         1925     1833      -92     
- Misses        375      489     +114     
Impacted Files Coverage Δ
...c/TimeModeling/Modeling/twri_objective_parallel.jl 0.00% <ø> (ø)
src/TimeModeling/Types/ModelStructure.jl 86.36% <ø> (ø)
src/TimeModeling/Utils/auxiliaryFunctions.jl 80.12% <96.15%> (-1.98%) :arrow_down:
.../TimeModeling/Modeling/lsrtm_objective_parallel.jl 0.00% <0.00%> (-100.00%) :arrow_down:
...rc/TimeModeling/Modeling/lsrtm_objective_serial.jl 0.00% <0.00%> (-78.27%) :arrow_down:
...TimeModeling/LinearOperators/judiExtendedSource.jl 17.64% <0.00%> (-58.83%) :arrow_down:
...ling/LinearOperators/judiJacobianExtendedSource.jl 65.51% <0.00%> (-24.14%) :arrow_down:
...ing/Modeling/extended_source_interface_parallel.jl 55.55% <0.00%> (-22.23%) :arrow_down:
...rc/TimeModeling/LinearOperators/judiPDEextended.jl 54.54% <0.00%> (-20.46%) :arrow_down:
src/TimeModeling/LinearOperators/judiLRWF.jl 64.70% <0.00%> (-17.65%) :arrow_down:
... and 9 more

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update fef5821...c9bda12. Read the comment docs.

mloubout commented 3 years ago

Is this work in progress or did you finish the update?

ziyiyin97 commented 3 years ago

Still in progress. Will finish it by the end of the week

mloubout commented 3 years ago

status?

mloubout commented 3 years ago

ping

ziyiyin97 commented 3 years ago

@mloubout could you provide with me some suggestions on how to make the test set? I am thinking of adding isic to the argument and to the runtest.jl file but what about dft/optimal checkpointing etc?

For example, lsrtm_objective w/ isic=true should pass the gradient test (Taylor series one). However, lsrtm_objective w/ 'optimal_checkpointing=trueshould NOT pass the gradient test but only follow the correct algebra, i.e. objective function= 0.5 * ||J^T(Jx-d)||_2^2, gradient=J^T(Jx-d)`. It would be kind of messy to put all of these to the argument list

mloubout commented 3 years ago

Why would it fail with checkpointing??? It's supposed to give the exact same gradient. Do you know what checkpointing does?

ziyiyin97 commented 3 years ago

My bad. Checkpointing computes the true gradient. How about gradients computed by several frequencies?

mloubout commented 3 years ago

what's the status on this?

ziyiyin97 commented 2 years ago

Most of the things here were solved by previous PRs, e.g. #87 #64 #91 . I may open another one to address some remaining issues (if any). Closing.

mloubout commented 2 years ago

@rafaelorozco @ziyiyin97 one of the main goal of this PR was to update and fix the tutorials, examples and documentation. None of these happened in the end and just a few PRs adding additional un-documented features isn't really helping.

ziyiyin97 commented 2 years ago

The leading order problem of the tutorial is about sim src, which is mentioned in #92 . It needs a bit work to implement a weighting operator. I will get back to it after SEG