Added titles of all the figures. Changed name of the following functions:
- Module MISC:
- SNR → snr
- SNR_2D → snr_2D
- Module PROCESSING
- Cadzow → cadzow
- Cadzow_2D → cadzow_2D
- EAE → eae
- LRD → lrd
- MCR → mcr
- MCR_ALS → mcr_als
- MCR_unpack → mcr_unpack
- stack_MCR → mcr_stack
- SIMPLISMA → simplisma
- Module FIT
- LR → lr
- LSP → lsp
- SINC_phase → sinc_phase
Several bug corrections.
- misc.merge_dict
- misc.zero_crossing
- fit.make_iguess_auto
- Spectrum_1D.add_noise
- Spectrum_1D.scan
- Spectrum_2D.add_noise
- Spectrum_2D.scan
- Pseudo_2D.add_noise
- Pseudo_2D.scan
- fit.Voigt_Fit.iguess: added "auto" flag to use fit.make_iguess_auto
- Spectrum_1D.plot: added tracker for distance measurement
- Spectrum_2D.plot: added tracker for distance measurement
- fit.Voigt_Fit: added attribute SW, computed on the ppm scale
Several changes to allow compatibility with newer versions of matplotlib. Added the attribute CM_2D in config, a dictionary that collects only the colormaps to be used for displaying 2D spectra. Modified dtype of empty arrays in Spectra. Added new functions for convolution, and its reverse. The latter does not work yet. Bug corrections.
- processing.convolve
- processing.inv_convolve - DOES NOT WORK
- fit.test_randomsign
- fit.test_correl
- fit.test_ks
- misc.select_traces: changed identifier of mouse buttons
- misc.show_cmap: added parameter filename to save the figure, changed figure size
- processing.MCR: modified parameters to use the new processing.stack_MCR and processing.MCR_unpack
- processing.MCR_unpack: now uses positioning matrix
- processing.stack_MCR: now uses positioning matrix
- figures.dotmd_2D: now uses the new CM_2D
- figures.redraw_contour: change for compatibility with matplotlib>3.7
- fit.test_residuals: employs three new functions, improved performance
- Spectrum_2D.init: reshape of the fid is now made with last dimension implicit
Added new functions to transform FID in audio files! New function for computing Hilbert transform. Adaptation of some codes for the new versions of numpy and scipy. Bug correction.
- misc.data2wav
- processing.hilbert
- fit.Peak.get_fid
- Spectrum_1D.to_wav
- Spectrum_2D.to_wav
- Pseudo_2D.to_wav
- misc.binomial_triangle: employs the function from the math library
- misc.edit_checkboxes: adapted for matplotlib>=3.7
- processing.ft, processing.ift: warnings for non-complex data are now suppressed
- processing.qpol: the polynomion is computed by solving the linear system with fit.LSP
- sim.load_sim_1D, sim.load_sim_2D: more intelligent handling of spacings, now writing input files is easier
- sim.t_voigt, sim.t_2Dvoigt: corrected the parameter x_G in order to preserve the linewidth of the signal
- Pseudo_2D.mount: the FIDs have now priority on the .npy file
- fit.CostFunc: all methods now compute the array of residuals
Added new functions for fitting pseudo-2D experiments. Bug corrections. Improved fitting routines.
22/01/2024
- fit.make_iguess_P2D
- fit.plot_fit_P2D
- fit.read_vf_P2D
- fit.voigt_fit_P2D
- fit.write_vf_P2D
- fit.Voigt_Fit.res_histogram
- fit.Voigt_Fit_P2D.res_histogram
- Spectrum_1D.pknl
- Spectrum_2D.pknl
- Spectrum_1D.to_vf
- fit.histogram: changed figure size to figures.figsize_large
- .write_ser for all classes in Spectra: fixed positional arguments order
- fit.Voigt_Fit_P2D
- fit.Peak: instead of zero-filling, the acqusition timescale is extended to match the shape of the spectrum with misc.extend_taq
- fit.Voigt_Fit: acqusition timescale extended with misc.extend_taq
- Pseudo_2D: added new attribute F as a wrapper for a fit.Voigt_Fit_P2D object. The self.F.S attribute is overwritten every time self.S is changed by a function such as adjph.
Bug corrections. Added support for Topspin 4.3.0, i.e. a reshape for the FIDs. Convdta seems to not work anymore, hence we introduced a way to remove the digital filter effects on the spectra by using the pnkl function.
- misc.load_ser
- processing.integrate
- processing.stack_fids
- fit.calc_R2
- Spectrum_2D.pknl
- Pseudo_2D.mount
- Pseudo_2D.cal
- Pseudo_2D.pknl
- misc.get_ylim: changed method of calculations of plot edges
- misc.set_ylim: now uses modified misc.get_ylim
- misc.polyn: use the Vandermonde matrix to compute the polynomion instead of a for loop
- fit.LR: now treated as a polynomial fit, added option to force the intercept
- fit.gen_iguess: brand new implementation
- fit.make_iguess: added an interactive text to show the actual sensitivity value, and a tutorial for the keyboard shortcuts (thanks to Tino Golub for requesting for this)
- Spectrum_1D.integrate: corrected integral computation by using new function processing.integrate
- Pseudo_2D.init: self.fid = None by default, added acqus parameter "TD1", removed procs parameter "cal_1", renamed procs parameter "cal_2" into "cal"
- Spectrum_2D: Added control for FnMODE="QF-nofreq"
It is now possible to work with Oxford Instrument NMR data, saved in .jdx files. The class fit.Voigt_Fit was completely rewritten using new, better performant functions. The default format for all figures became ".tiff". Minor bugs corrected.
- misc.makeacqus_1D_oxford
- processing.blp
- fit.plot_fit
- fit.voigt_fit_indep
- fit.write_vf
- fit.read_vf
- Spectrum_1D.blp
- fit.Peak
- figures.dotmd: when the figure is closed, returns the scaling intensity factor for each spectrum
- figures.dotmd_2D: when the figure is closed, returns the contour level start for each spectrum.
- fit.make_iguess: infinite components!, new format for the output file
- fit.Voigt_Fit: completely rewritten using the new functions.
Extension of the Spectra classes for the management of other-than-Bruker data formats. Modifications to some attributes and methods to make everything more user-friendly. See the dedicated paragraph for details.
The linalg package, previously imported from scipy, was now substituted with its numpy counterpart. Problems due to missing funcions were not tested, hence errors might arise.
New algorithms for phase correction, baseline computation, smoothing, linear least-squares fitting were also implemented.
All the functions that save figures, i.e. almost the whole figures package, have now two additional parameters: ext, which allows to save the figure in different formats than PNG, and dpi, to specify the desired resolution.
The user guide at the beginning of the documentation pdf file was updated to instruct the reader about the recent modifications of the package.
The attribute datadir has been split in two parts: datadir, which only contains the path to the directory that contains the file, and filename, which is actually the name of the file, without any extension. They are generated by functions in os.path, which should make the reading of the spectra cross-platform. The trimming of the extension is achieved by right-splitting at the last "." character of the filename, and retaining the left part.
Right after having read the FID and/or the processed data, the program looks for a file named filename.procs in the same directory of the input file, which contains the procs dictionary in a format that the eval function is able to interpret as a dictionary. If there is, this file is transformed in a dictionary and stored in the procs attribute; otherwise, the procs dictionary is initialized with default values and then saved in a file named filename.procs in the same directory of the input file.
The processing function, after the Fourier transform, applies the phase correction and calibrates the spectrum according to the values that are currently stored in the procs dictionary. With respect to the previous point, this means that you can store your phase angles to remember the processing.
All functions that change a processing parameter, such as adjph or the baseline computation, call for the method write_procs, which update the file filename.procs with the new values stored in the procs dictionary. In order to comply with this new aspect, the methods adjph and cal have now a new parameter "update" which allows to disable the update of the procs dictionary with the new values. Its default value is "True", hence it behaves as it did previously unless it is explicitely declared.
More detailed comments added in the source code.
- fit.fit_int: now it computes both intensity and offset, with a slightly modified formula which takes into account also the less relevant term.
- figures.ax2D: corrected AssertionError when xscale and yscale have the same dimension
- figures.figure2D: corrected AssertionError when xscale and yscale have the same dimension
- figures.figure2D_multi: employs better colormaps. You can superimpose a maximum of 6 spectra (positive only) or 3 spectra (both positive and negative).
- fit.get_region: automatic zoom can be enabled/disabled by pressing "z" on the keyboard
- fit.histogram: now it calls for fit.ax_histogram to compute the histogram.
- spectra.Pseudo_2D.plot_md: default value of the "which" changed from "all" to None
- spectra.Pseudo_2D.plot_stacked: default value of the "which" changed from "all" to None
- spectra.Spectrum_1D.init: now the parameter "in_file" can also be an acqus dictionary, in this case the function sim.load_sim_1D is not called
- spectra.Spectrum_1D.init: added new parameter "spect", that allows to read also Varian and Magritek datasets.
- spectra.Spectrum_1D.basl now subtracts self.baseline from self.S, then unpacks it into self.r and self.i. Its previous behaviour is copied in the new function Spectrum_1D.baseline_correction
- spectra.Spectrum_1D.process now reverses automatically non-Bruker data after FT for correct display of data
- spectra.Spectrum_2D.eae: the behaviour of the function is now related to the value of "self.eaeflag"
- spectra.Spectrum_nD.save_ser changed name into .write_ser for consistency with other functions. Also the definition of the destination pathway has changed.
- misc.makeacqus_1D_varian
- misc.makeacqus_1D_spinsolve
- processing.acme
- processing.whittaker_smoother
- processing.RPBC
- processing.align
- fit.LR
- fit.LSP
- fit.polyn_basl
- fit.SINC_phase
- spectra.Spectrum_nD.write_procs
- spectra.Spectrum_nD.read_procs
- spectra.Spectrum_1D.acme
- spectra.Spectrum_1D.baseline_correction
- spectra.Spectrum_1D.rpbc
- fit.CostFunc
- fit.SINC_ObjFunc
Serious bug found! The functions processing.ft and processing.ift did not return the correct frequency. They have been corrected by converting them to "normal" FT and iFT, but reversing the data manually.
- processing.ft: so to use np.fft.fftshift(np.fft.fft(data))
- processing.ift: so to use np.fft.ifft(np.fft.ifft(data))
The docstrings of the functions, as well as the comments and the Latex documentation, were corrected, expanded, and uniformed to a common standard. This should make them more readable and useful.
- misc.in2px
- misc.px2in
- misc.trim_data_2D
- processing.interactive_echo_param
- fit.peak_pick
- fit.gen_iguess_2D
- fit.build_2D_sgn
- fit.voigt_fit_2D
- fit.Voigt_Fit_2D (class)
- processing.split_echo_train: made more python-friendly and not limited to 1D and 2D data. It now treats the first decay separately from the true echoes.
- processing.sum_echo_train: now calls for processing.split_echo_train and sum on the first dimension
- processing.ft: removed the "Numpy" parameter.
- processing.ift: removed the "Numpy" parameter.
- replaced the "print" statement with "warnings.warn" in processing.ft and processing.ift
- decorator function cron added in config and imported by init
- Spectra, all classes: the attributes BYTORDA, DTYPA and grpdly were removed. Three keys, BYTORDA, DTYPA and GRPDLY, were added to the attribute acqus
- Spectra, all classes: the method write_ser became a @staticmethod
- Spectra.Pseudo_2D: added method write_integrals
First release of KLASSEZ.