Skip to content

Latest commit

 

History

History
77 lines (74 loc) · 7.47 KB

CHANGE.rst

File metadata and controls

77 lines (74 loc) · 7.47 KB

Version 1.2 (unreleased)

Version 1.1 (2019-11-26)

  • #217 - Add pip install to docs. Clean-up setup script.

Version 1.0 (2019-04-20)

  • #216 - Correct the PCA spatial and spectral width definitions. Update unit test data to reflect change.
  • #215 - Generalize WCS pixel conversions. Fixed units for the delta-variance break.
  • #208 - Fix #207; description of NaN handling for the delta-variance tutorial from @keflavich. Change to requiring astropy >v2!
  • #171 - Expand PCA tutorial; fix unit conversion error in astropy 3.2dev
  • #170 - Add tutorial pages for distance metrics and updated distance metric inputs. API CHANGES IN DISTANCE METRICS!
  • #192 - Add stddev and weighted fitting for Wavelet. Correct bug in fitting weights for StatisticBase_PSpec2D.
  • #191 - Use robust covariance for linear models. Add residual plots. Add residual bootstrapping for linear models.
  • #190 - Add spectral downsampling by averaging for VCA; remove downsampling for VCS.
  • #189 - Fixing elliptical p-law parameters for isotropic fields.
  • $187 - Correct normalization in 3D power-law fields.
  • #186 - Generate mock PPV cubes and 3D power-law fields; added tests for generated power-law in 2D and 3D; renamed data_reduction to moments; removed masking procedures from Mask_and_Moments and renamed to Moments
  • #185 - Correct SCF weighting for the distance metric. Addresses #184.
  • #183 - Add progress bars for long operations. Addresses #180.
  • #182 - Separate out plotting functions that were implemented in the run functions of the statistics.
  • #181 - Add apodizing kernels and beam correction for spatial power-spectrum methods.
  • #177 - Add pyfftw as an optional dependency. Allow using pyfftw in all FFT-based statistics for multi-thread support.
  • #179 - Remove use of matplotlib._cntr (removed in their version 2.2) for PCA contours. Added scikit-image as a dependency to use their find_contours function.1
  • #174 - Set the fraction of valid data points to compute the moments in a region for StatsMoments. Updated those test values for the default setting of min_frac=0.8.
  • #171 - Fix issues with NaN interpolation in Delta-Variance for observational data; added a min. weighting for Delta-Variance; allow segmented fitting for Delta-Variance; added support for weights and WLS in Lm_Seg
  • #165 - Added azimuthal constraints when creating 1D power-spectra. Also added radial and azimuthal slicing for the bispectrum. Fixed a critical bug in the bootstrapping for the 2D elliptical power-law model.
  • #166 - Removed testing python 2.7 with the dev version of astropy.
  • #164 - Add segmented linear fitting for the wavelet transform; updated unit test values to include segmented fitting slopes for wavelets
  • #163 - Use Hermitian eigenvalue decomposition for PCA (fixes getting imaginary components); Update the ellipse contour fitting to use the robust fitter from skimage (fixes previous PCA test failures); updated unit test values for the change in PCA
  • #161 - python 3 support; use ci-helpers; updated astropy-helpers; consistent convolution results with astropy 2.x; new delta-variance regression test values after convolution updates; DeltaVariance_Distance.curve_distance now restricts to the region between xlow and xhigh
  • #159 - Add an elliptical power-law model for fitting 2D power-spectra. Added 2D fitting for PowerSpectrum, MVC, VCA, and SCF, along with tests for each. Unit test values updated to include the 2D slopes. Also updated the tutorials of those 4 methods to demonstrate the new fitting procedure.
  • #156 - Complete statistic tutorials; complete angular + physical unit handling; pass break kwargs for all power-spectrum methods; change several APIs to be consistent throughout the package; removed stats_wrapper; added tests to test all cases with a slope and unit conversions; reworked slice thickness choice for VCA and VCS; restrict VCS fitting to real frequencies only; default to periodic boundaries

Version 0.2 (2017-07-03)

  • #147 - Changed Cramer normalization to the spectral norm.
  • #146 - Allow weights to be passed in StatsMoments.
  • #144 - Fix the MVC: subtract mean velocity dispersion, not the array of dispersions.
  • #143 - Set periodic boundaries for dendrograms.
  • #141 - Added fitting for delta-variance. Added setting of fit limits for delta-variance, wavelets, power spectrum, MVC, and VCA.
  • #138 - Add fitting routines and more normalization options for PDFs. Added testing for numpy 1.12 and astropy 1.3.
  • #136 - Normalize data to between 0 and 1 for the Cramer statistic.
  • #134 - Add rolling back for computing the SCF when shifts are integers (and so don't need the FFT approach)
  • #133 - Fix indexing when creating 1D power spectra.
  • #131 - Another correction for centroid offset issue. Cleaned moment calcs up a bit.
  • #130 - Corrections to moment error calculations.
  • #128 - log-log plotting for SCF spectrum. Fitting for SCF spectrum.
  • #127 - Standardize all data used for calculating the delta-variance distance. Also allows the allow_huge flag to be set in DeltaVariance when working on large images (>1 Gb).
  • #85 - Added full PCA treatment from the Heyer & Brunt papers. Also implements general handling of physical (given distance) and spectral scale conversions (only working on PCA, needs to be added on to the rest).
  • #120 - Minor plotting changes to PDF and bispectrum. Common angular area for Genus normalization. Fix issue with NaNs in the weights for PDFs.
  • #118 - Fix SCF surface indexing with non-integer shifts.
  • #117 - Use common angular scales for SCF, Moments, Delta-Variance comparison. Dropped using nanmean, nanstd from scipy.stats. Added fourier shifting for SCF.
  • #116 - Normalize Genus distance by the image area for comparing unequally-sized images
  • #115 - Alter the PCA and Genus metrics to more closely follow the rest of the suite.
  • #112 - Stop importing everything at the top level of the module. Also fixes minor input bugs in BiSpectrum_Distance and registers "check_deps" as a command for setup.py.
  • #109 - Limit VCA spectrum frequencies below 0.5.
  • #106 - Support for multiple data inputs (FITS HDU, SpectralCube, etc). Switched to pytest.
  • #105 - Bug in wrapper function. Added test for wrapper.
  • #102 - Homogenized the structure of the statistics and distance metrics. Several API changes and two metrics had their names changed: StatMomentDistance -> StatMoment_Distance and MVC_distance -> MVC_Distance.
  • #104 - Common base class for 2D power spectra
  • #103 - implement a corrected version of the wavelet transform
  • #101 - Fixed some bugs brought in with #92 for segmented fitting. Using OLS instead of WLS.
  • #100 - Fix plotting in PDF_Distance
  • #99 - Fixed moment array loading when a full path to the cube is given.
  • #98 - Removed scripts for the paper results from TurbuStat. Now available in a separate repo.
  • #96 - Fix error calculation in Delta Variance
  • #95 - remove double rounding and clean-up unused code in Bispectrum
  • #93 - Updated this file :).
  • #92 - Re-wrote pspec to use binned_statistic; now returns the proper spatial frequencies
  • #89 - Make signal_id and astrodendro optional packages
  • #86 - Fix ECDFs for standardized data. Added 'find_percentile' and 'find_at_percentile' functions.
  • #80 - Disabled Anderson-Darling test for PDFs due to unpredictable errors.
  • #79 - Fixed issue in PDF_Distance where all standardized data below 0 was cut-off. Also fixed the plotting in PDF_Distance.
  • #78 - Fixed using stddev for the 1D power spectra (MVC, SCF, VCA, PSpec). Pickling for SCF results.