Added all base method for the CMA-ES component of MTUQ. In the process of streamlining the parallel_cmaes.py implementation.

This will be the default file for development. Non-mpi version will be tested but might not be fully developed.

Corrected a few anomalies in the code. The path and cov update now correspond to purecma python package.

Based on GridSearch.FullMomentTensor.py, should provide a nice overview of the basic MT inversion with CMA-ES. Still work in progress.

Added a new function to automatically create the list of parameters for force inversion.

Work in progress Force CMA-ES application. Currently works with the "greens" mode in CMAES.Force.py

Fixed import error and cleaned up debug plot.

Tested the database mode on a depth-search.

This reverts commit 815d735.

This reverts commit aa596b3.

Overhauled the script description and populated with comments and recommendations on uses.

Local paths referring to my machine have been replaced with "generic" path to be changed by users.

Reset boundary handling method back to rand-based, instead of re-initialize.

Slight modification to allow for matplotlib backend plot for _plot_lune_matplotlib and the use of CMA-ES log as input for the main mtuq plot_misfit_lune() function

Added the CMA-ES algorithm description in the introduction comment for reference.

Added a Green's function caching option to account for local database use with fixed origin. This fix increases the performance significantly, even if the database mode is not the preferred way of searching solutions for fixed origin.

Combines the matplotlib backends misfit plots (lune and dc) into a figure for CMA-ES visualization.

Remove un-necessary default option preventing to change the colormap with **kwargs for some reason.

Adjustments for combined_plots

Added ability to plot deviatoric and dc on the lune by padding result DataArrays with v and w = 0 if needed

Added new plotting options, and got rid of temporary parts of the base script.

Workin version of the future "Solve" method. TODO: clean the deprecated sections of the code, add checks and tests, and docstrings.

Fix the latitude binning so that it can plot values centered on 0 (requires odd number of grid cells)

Fixed some of the plotting functions. Adding subroutines for readability and maintenance.

Reorganized script, changing docstring to unified format, rename the main "Solve" method.

Continued docstring reformating

Added new docstring for the Solve method and removed a useless input (replaced with inner CMA_ES_parallel mode attribute)

Remodeled the example script, and modified plotting function.

Make sure that colored plots are displayed above the Lune's grid.

Iteration printing + plot scaling tuning.

Removed unused source file.

Removed un-used source file (all functionalities are in the cmaes_parallel.py file, that will soon replace the base cmaes.py file)

Changed cmaes_parallel into cmaes for simplicity.

Corrected a bug that would prevent switching from the 5-column waveform plot to the 3-column waveform plot (if a single dataset/process was given)

Modified log print

Tweaked Polarity misfit class to improve compatibility with CMA-ES.
Integrated polarity misfit in CMA-ES, and detection of polarity misfit for waveform plotting.

Change checks order for polarity misfit avoidance in plot waveforms.

Pointing to incorrect variable in previous fix.

Fixed bug that prevented using polarity misfit with CMA-ES automatic Solve function (and plotting).

Changed dependency

Added Solve method input check and implemented basic data normalization.

Implemented weighted misfit combination and fixed bugs.

Added a workaround for CMA-ES plotting method, where we want to plot the mean instead of the minimum misfit.

Several bug-solve. Normalization works both in terms of inversion and plotting.

Allows plotting CMA-ES result (would fail otherwise when encountering a single sample in a bin)

Added a plot_type argument to allow more flexibility when plotting with the Matplotlib backend (allowing pcolormesh, scatter plots and contour plots).
Correct docstrings for the _plot_lune_matplotlib() method.

Added "plot_type" argument to force plots functions

Removed unnecessary argument that can go in **kwargs, to follow original syntax.

Implemented polarity misfit for point-force sources. A test function has also been added in order to verify that the USE convention is respected.

Removed unnecessary normalization.

Changed the default behavior for the v and w variable to "reinitialize" as it might helps with sampling close to the boundaries (even if making the convergence a bit slower).

Updated CMA-ES examples with new input syntax (for weighted misfit)

Fixed tri-contour bug encountered with NaN values in force plots.

Fixed colormesh on the sphere by splitting the plot in two to avoid projection artifacts.

Corrected the mis-rotation of DC data

Replaced dataset with Ezgi's example.

Changed default behavior for h to reinitialize for better exploration of both up and down hemispheres.

Defined default example for the Force CMA-ES, based on the Barry Arm glacier Landslide

Modified default weights for CMA.Solve in the database mode.

Changed naming convention for figures, and added event_id field in example CMA object initialization.

Fixed final naming convention.

Added data from the Barry Arm landslide for force inversion examples.

Allowing different naming conventions for long-period data (MXZ, MXR, MXT on top of BXZ, BXR, BXT)

Added the force sphere plot during the plotting phase of the Solve method. Also fixed the mean misfit location as the green circle in the plot (in-lieu of the minimum of the misfit map).

Added scatter plot functionality to force mode. The behavior will now be to return a different plot for each mode (Lune or Force sphere) based on CMA-ES mode (mt, mt_dc, mt_dev or force).

Added rank.0 plotting function (would otherwise cause problems on parallel jobs).

Final version of the CMA-ES force example.

Created the Force Example for MTUQ based on the Barry Arm glacier landslide.

Restored default .gitignore

Updated comment before final commit and PR.

Change the type of import to mimic import of mpi4py in example scripts.

Added a check for mpi environment.