Changelog

Latest Changes

v0.3.33

Fixed

Fixed FutureWarning messages from pandas.
Set default numeric type for prior knowledge to float.

Changed

Updated logging statements to more applicable levels (e.g. INFO to DEBUG).
Changed tight layout to constrained layout for figures with subplots in visualization.py.
Simplified OS test matrix by skipping Python 3.9, 3.11, 3.13.
Disabled preview for HSVDinitializer in tests/step2_simple.ipynb to reduce CI/CD failures.

Thanks to @bastigw for contributing this version in PR #10.

v0.3.32

Added

Added support for Python 3.13 and 3.14.
Modernized CI/CD workflows with reusable workflows and new triggers.
Created a legacy branch legacy2025 and labeled it as v0.3.29-legacy2025 for users who need to maintain compatibility with Python 3.6 and 3.7.

Fixed

Fixed a bug that .vscode is not ignored in the .gitignore file.

Changed

Updated minimum Python version requirement from 3.6 to 3.8 across all configuration files
Updated recommended Python version from 3.8 to 3.12.
Expanded test matrix to cover Python 3.8-3.14 on Ubuntu, Windows, and macOS.
Removed older Intel-based macOS (macos-13) from test matrix.
Changed the default python version for readthedocs from 3.8 to 3.12.
Updated documentation to reflect the new minimum and recommended Python versions, and added notes for legacy users to download the legacy2025 branch when needed.

Thanks to @bastigw for contributing this version in PR #9.

v0.3.31

Fixed

Fixed a bug in the web interface where non-.txt format FIDs were incorrectly identified as .txt files, causing program crashes.
Fixed a bug in both the web interface and command-line script where the code attempted to access non-existent result_sum and simple_pd from the fitted result object.

v0.3.30

Added

Added a new “Simple FID Simulation” mode to the Streamlit app script/amaresfit_gui.py. This mode allows users to simulate FIDs using prior knowledge dataset and perturbed the parameters.

s needed

Added extra_line_broadening to fid.simulate_fid to allow users to add extra line broadening to the simulated FID before adding noise. This avoid SNR bias introduced by extra global linebroadening factors.

Fixed

Fixed a bug in the Streamlit app where editing prior knowledge (PK) would remove # symbols from comment rows, causing file parsing to fail when the comment line appears as the first line

v0.3.29

Added

Added a streamlit app script/amaresfit_gui.py to provide a graphical user interface for pyAMARES.

v0.3.28

Fixed

Fixed a bug in CRLB calculation that caused failures when multiple fitting parameters were fixed.
Fixed logger print format typos across multiple files.

v0.3.27

Added

Added logger.py module to API documentation.
Added more comprehensive ruff checking configurations in pyproject.toml.

Changed

Replaced remaining print and warning statements with the new logging system.
Reorganized import statements across multiple files to follow best practices (alphabetical ordering, standard library imports first).
Excluded Jupyter notebooks and third-party code from ruff linting.
Updated long description in setup.py with proper line breaks for PEP-8 standards.

v0.3.26

Added

Added comprehensive code formatting with ruff as mentioned in Issue #5.
Added ruff requirements to setup.py with dedicated install option: pip install -e ".[ruff]"
Added dev dependencies in setup.py to include jupyter, documentation, and code quality tools.
Added GitHub CI/CD configuration with ruff.yml for automated code quality checks.
Added CONTRIBUTING.rst to documentation with guidelines for contributors.

Changed

Significantly reformatted codebase using ruff for consistent code style.
Updated documentation structure with inclusion of CONTRIBUTING.rst.
Revised README.rst with improved contributor guidelines.
Updated docs/source/index.rst to include links to contribution information.

v0.3.25

Added

Added centralized logging system with pyAMARES/libs/logger.py module for better debugging and output management.
Implemented configurable log levels (DEBUG, INFO, WARNING) via get_logger() and set_log_level() functions.
Replaced print statements with proper logging in several modules (PriorKnowledge.py, fid.py, lmfit.py, crlb.py).
Added ability to control output verbosity, especially useful when processing multiple spectra in loops.

Changed

Improved code consistency by standardizing status and warning messages across the package.
Enhanced user experience with cleaner console output and better filtering options.

Fixed

Addressed console flooding issues when processing large batches of spectra.

Thanks to @andrewendlinger for contributing this version in PR #4.

v0.3.24

Added

Added tests directory to the package with Jupyter notebook test examples.
Implemented CI/CD pipeline testing against multiple Python versions (3.7, 3.8, 3.10, and 3.12) on Ubuntu-22.04.

Fixed

Fixed an image link in README.rst

v0.3.23

Fixed

Fixed compatibility issues with Numpy 2.0 and Scipy 1.14.0+ and above.
Fixed a critical bug introduced in v0.3.18 where the incorrectly included simple_df broke fitAMARES when using HSVD generated parameters.

v0.3.22

Fixed

Fixed a bug in initialize_FID where J-coupling constants in the priori knowledge spreadsheet could not be float numbers.

v0.3.21

Added

Added delta_phase parameter to initialize_FID for applying additional phase offset (in degrees) to the prior knowledge phase values.
Added delta_phase to script amaresfit.py

v0.3.20

Added

Added scale_amplitude to initialize_FID to scale the amplitude of input prior knowledge dataset.
Added scale_amplitude to script amaresfit.py
Added a bool notebook to run_parallel_fitting_with_progress to toggle the Progress Bar for jupyter notebook.

Fixed

Fixed an important but not critical bug where lowerbound (lval) parameter of linewidth dk read from prior knowledge dataset by initialize_FID was always ignored and set to 0 in all versions prior to v0.3.20.
Fixed a critical bug introduced in v0.3.19 (commit e319a5c) that broke the function run_parallel_fitting_with_progress.
Install hlsvdpro only when platform.machine() returns x86_64 or amd64.

v0.3.19

Added

Added remove_zero_padding function to eliminate zero-filled data points that could cause incorrect SNR calculations.

v0.3.18

Added

Added simple_df Dataframe to the fid_parameters.

Fixed

Fixed a typo in the equation in what.rst.

v0.3.17

Added

Added objective_func parameter to multiprocessing.run_parallel_fitting_with_progress and multiprocessing.fit_dataset functions
Fixed minor typos

v0.3.16

Added

Added params_to_result_pd, which is the inverse function of params_to_result_pd.

v0.3.15

Fixed

Fixed a critical bug where J-coupling expressions ending with Hz were incorrectly interpreted as ppm.
Fixed a critical bug that prevented correct parsing of prior knowledge when there was a space in J-coupling strings, such as “0.125 ppm” and “15 Hz”.
Loosen the bounds of chemical shift of ATP peaks in the attached example prior knowledge datasets of human brain at 7T.
Updated the simple_tutorial.ipynb to use the new prior knowledge dataset and the new API.

v0.3.14

Added

Added print_lmfit_fitting_results, a function to print key lmfit fitting results from the fitting_results.out_obj.

Fixed

Changed the version number from 0.4.0 to 0.3.10 to better manage version increments.

v0.3.13

Added

Added result_pd_to_params, a function that converts fitted results from a DataFrame format into a Parameters object for use with simulate_fid.

Fixed

Set normalize_fid=False to be turn it off for initialize_FID by default.

v0.3.12

Fixed

Fixed a bug in the sum_multiplets function that prevented the SNR multiplets from being added.
Revised the printouts for when initialize_with_lm is enabled.

v0.3.11

Fixed

Updated the result["phase"] and result["phase_sd"] to be wrapped according to the minimum and maximum degree constraints defined in the prior knowledge dataset.

v0.3.10

Added

Added the initialize_with_lm option to both fitAMARES and run_parallel_fitting_with_progress functions.
Added a highlight_dataframe function that highlights rows in a DataFrame based on the values of a specified column.

Fixed

Updated docstrings in numerous functions to ensure they render properly.
Add result["phase"] = (result["phase"] + 180) % 360 - 180 to report.py to wrap ~360 degrees to ~0
Fixed a bug in readmat.py
Fix a bug that the internal initializer initialize_with_lm always uses the input method to initialize. Now it uses leastqs as the internal initializer.

v0.3.9

Added

The peak-wise Signal-to-Noise Ratio (SNR) is now added to each result_pd. The Standard Deviation (SD) of the noise is obtained from the last 10% of points in the FID.

Fixed

Mute __version__ and __author__ printouts.

v0.3.8

Added

Add a read_fidall function to read GE MNS Research Pack fidall generated MAT-files.

v0.3.7

Fixed

Instead of try .. catch, use def is_mat_file_v7_3(filename) to identify if a file is V-7.3

v0.3.6

Added

The readmrs function now supports any MAT-files containing either an fid or data variable. This enhancement makes it compatible with GE fidall reconstructed MAT-files as well as Matlab formats written by jMRUI.

v0.3.5

Fixed

Fixed a bug where, if the ppm needs to be flipped while the carrier frequency is not 0 ppm, the resulting spectrum looks wrong with a fftshift().

v0.3.4

Added

An argument noise_var to initialize_FID that allows users to select CRLB estimation methods based on user-defined noise variance. By default, it employs the noise variance estimation method used by OXSA, which estimates noise from the residual. Alternatively, users can opt for jMRUI’s default method, which estimates noise from the end of the FID.

v0.3.3

Added

Fixed the carrier placeholder. If carrier is not 0 ppm, shift the center frequency accordingly.

v0.3.2

Added

Updated the generateparameter to allow a single number in the bounds region to fix a parameter. This update resolves issues with parameter bounds specification.

v0.3.1

Added

Introduced a read_nifti placeholder to facilitate future support for the NIFTI file format.

This document describes all notable changes to pyAMARES.