shimming-toolbox / shimming-toolbox / 13526392470

90%

Build Type

push

github

Committed by web-flow

Commit Message

Adding Gradient-based optimizer (BFGS) and pseudo huber loss function  (#515)

* adding the pseudo huber cost functon to the existing lsq_optimizer

* modifiying the b0_shim cli to include added opt_criteria and opt_method

* correct formatting

* adding a gradient based optmiziation method and pseudo huber minimization criteria

* adjusting the max character per line

* minor modification on formatting

* adding the gradient optimizer to list of supported optimizers

* optimizing the way delta is calculated for pseudo huber for whole brain shimming

* updating the gradient_optimizer to inherit from lsq_optimizer

* removing redundant init function in gradient optimizer

* Update shimmingtoolbox/cli/b0shim.py

Co-authored-by: Alex Dastous <47249340+po09i@users.noreply.github.com>

* removing constraint from gradient optimizer

* modifying help text in the CLI

* Apply suggestions from code review

modifying some typos.

Co-authored-by: Alex Dastous <47249340+po09i@users.noreply.github.com>

* removed the redundant codes

* Rename gradient optimizer to bfgs, squash bugs

* Refactor bound generation for real-time shimming

* Add BFGS RT PMU optimizer

* Fix test

* Remove comments

* Fig bug with signal recovery opt if using a single slice fmap

* Fix ps_huber bug

---------

Co-authored-by: Alex Dastous <47249340+po09i@users.noreply.github.com>
Co-authored-by: Alexandre D'Astous <po09i@hotmail.com>

Run Details

71 of 78 new or added lines in 5 files covered. (91.03%)

2 existing lines in 2 files now uncovered.

5404 of 6018 relevant lines covered (89.8%)

4.49 hits per line