tqt_nmr package

Subpackages

tqt_nmr.experimental package

Submodules

tqt_nmr.collection module

Dummy pipeline:

>>> specs = NMRSpectraCollection.from_signals(signals, zf_amount=len(times) * 9, time_scale=1)
>>> spec = specs.map(NMRSpectrum.apply_harmonics_mask).brute_windowing(100).fuse('auto', verbose=True)
>>> spec.normalizes_frequencies(fast=False).correct_phase()
>>> spec.kivi_baseline_correction().plot()

class NMRSpectraCollection(spectra)

Bases: object

Parameters:: spectra (list[NMRSpectrum])

spectra: list[NMRSpectrum]

fuse(mode='mean', verbose=False)

Averages the spectra into a single one.

Parameters:

mode (Literal['mean', 'corr', 'auto']) – Either ‘mean’, ‘corr’, or ‘auto’. - mean: Averages the spectra element-wise. - corr: Aligns by correlating the spectra beforehand. - auto: Tries both mean and corr and uses the one that produces the most symmetric spectrum.
verbose (bool) – Whether to log which method is picked if mode="corr".

Returns:

The fused single spectrum.

Return type:

NMRSpectrum

classmethod from_raw(times, voltages, zf_amount=-1)

Converts a list of Signal objects into a collection of NMR spectra.

Parameters:

times (ndarray) – 2D array of shape (samples, time_points).
voltages (ndarray) – 2D array of shape (samples, voltages).
zf_amount (int) – Zero-filling parameter for the NMRSpectrum’s construction.

Return type:

Self

classmethod from_signals(signals, zf_amount=-1, cut_dead_time=False, **kwargs)

Converts a list of Signal objects into a collection of NMR spectra.

Parameters:

signals (list[Signal]) – A collection of signals to convert.
zf_amount (int) – Zero-filling parameter for the NMRSpectrum’s construction.
cut_dead_time (bool) – Whether to cut the dead time signal.

Return type:

Self

map(func, inplace=True, *args, **kwargs)

Applies a function to all of the spectra in the collection.

Parameters:

func (Callable[..., NMRSpectrum]) – Function to apply. Takes NMRSpectrum as input, returning a modified spectrum instance.
inplace (bool) – Whether to modify the current collection or to create a new one. Defaults to True.

Return type:

Self

brute_windowing(window_width=150)

Cuts a window of the specified size around the max abs peak of each spectrum. Shifts the frequencies to move the peak position to zero.

Parameters:: window_width (float) – Width of the window in Hz. Defaults to 100.
Return type:: Self

tqt_nmr.config module

tqt_nmr.fft module

apply_fft(times, values, length, full_fft=False, **kwargs)

Return type:

tuple[array, array]

Parameters:

times (ndarray)
values (ndarray)
length (int)
full_fft (bool)

tqt_nmr.signal module

class Signal(time_points, amplitudes)

Bases: object

Parameters:

time_points (ndarray)
amplitudes (ndarray)

time_points: ndarray

amplitudes: ndarray

plot(ax=None, **kwargs)

Plots the signal with matplotlib.

Parameters:

ax – Axis to plot on. Uses the default one is not provided.
kwargs – Additional arguments to be provided to ax.plot(…).

Returns:

The axis on which the signal is drawn.

get_upper_envelope(smoothing_kernel_factor=70)

Creates an upper-envelope of the signal

Return type:: Self

cut_dead_time(max_duration_millis=inf, threshold=0.5, extra_cut_millis=0)

Finds the first local maxima of the envelope and cuts the signal by that point.

Parameters:

max_duration_millis (float) – Max possible length of the dead time
threshold (float) – Normalized abs signal above this level is considered to be a signal.
extra_cut_millis (float) – Additional time to cut after the dead time.

Return type:

Self

Returns:

Self

apodize(func)

Applies the apodization function provided.

Parameters:

func (Callable[[ndarray], ndarray]) – Apodization function. Will take the range [0;1] as input.
output. (The signal is multiplied by the)

Return type:

Self

Returns:

Self

align_to(reference_signal, max_shift=0.02, res=3, fast=False)

Cuts the signal to align with the reference.

Parameters:

reference_signal (Self) – Signal to align to.
max_shift (float) – Max shift in seconds. Defaults to 0.02.
res (int) – Resolution of phase alignment, see phase_align(…). Defaults to 3.
fast (bool) – Whether to use a faster version of the algorithm. Defaults to False.

Return type:

Self

Returns:

Self

normalize()

Scales the signal to [-1 1] range.

Return type:: Self

quadrature_detection(F0=None)

Return type:: Self
Parameters:: F0 (float | None)

as_plotly(plot_type='line', **kwargs)

Parameters:: plot_type (str)

copy()

classmethod from_folder(folder, file_extension='.dat')

Reads a collection of signals from a folder containing raw *.dat files.

Return type:

list[Self]

Parameters:

folder (Path | str)
file_extension (str)

classmethod from_csv(file)

Return type:: Self
Parameters:: file (Path | str | IOBase)

save_bin(file, **kwargs)

Return type:: None
Parameters:: file (BytesIO | Path | str)

classmethod from_bin(file)

Return type:: Self
Parameters:: file (BytesIO | Path | str)

tqt_nmr.spectrum module

class NMRSpectrum(frequencies, amplitudes)

Bases: object

Parameters:

frequencies (ndarray)
amplitudes (ndarray)

frequencies: ndarray

amplitudes: ndarray

classmethod from_signal(signal, zero_fill_amount, time_scale=1e-06, **kwargs)

Shortcut for NMRSpectrum.from_voltages(…)

Return type:

Self

Parameters:

signal (Signal)
zero_fill_amount (int)
time_scale (float)

classmethod from_voltages(times, voltages, zero_fill_amount, time_scale=1e-06, **kwargs)

Constructs a spectrum from a raw signal.

Parameters:

times (Union[ndarray, Sequence[float]]) – Time points of the signal.
voltages (Union[ndarray, Sequence[float]]) – Signal amplitudes at the corresponding time points.
zero_fill_amount (int) – How many zeros to add at the end. Rounds to the nearest power of two if <= 0.
time_scale (float) – Scaling factor of the time points to convert to seconds. Defaults to millis (1e-6).

Return type:

Self

Returns:

The spectrum constructed.

shift_freqs(amount)

Return type:: Self
Parameters:: amount (float)

smooth(sigma=4)

Return type:: Self
Parameters:: sigma (float)

find_peaks(dist_hz=None, min_height=None, prominence=None)

Finds the frequencies of peaks matching the parameters. See scipy.signal.find_peaks(...).

Return type:: list[float]

filter_peaks(peaks, max_hz_diff, max_amp_diff, min_amp, roi=(-20, 20))

Filters the peaks by ROI, symmetricity, amps difference.

Parameters:

peaks (ndarray) – List of peaks to filter (Hz).
max_hz_diff (float) – Max allowed abs difference between positions of the peaks.
max_amp_diff (float) – Max allowed abs difference between amplitudes of the peaks.
min_amp (float) – Minium allowed amplitude of a peak.
roi (tuple[float, float]) – Frequency range to search in. Defaults to (-20, 20).

Returns:

Pairs of peaks found. List[float]: List of mean amplitudes of the peaks pairs. List[float]: List of symmetricity scores of the peak pairs (more is better).

Return type:

List[Tuple[float, float]]

apply_harmonics_mask(freq=50, k=0.99)

Masks out harmonics by a cosine wave.

Parameters:

freq (float) – Frequency of the harmonics to mask.
k (float) – Cutoff threshold of the cosine wave (higher = thinner mask). Defaults to 0.99.

Return type:

Self

cut_freqs(start=-inf, end=inf)

Cuts the the spectrum to the frequency range provided (inclusive).

Parameters:

start (float) – Left-most side of the frequencies range. Defaults to -inf (no cut).
end (float) – Right-most side of the frequencies range. Defaults to +inf (no cut).

Return type:

Self

Returns:

Modified version of the object.

normalizes_frequencies(fast=False, peaks_prominence=0.15, return_peaks=False)

Finds the central peak and shifts it to 0Hz.

Parameters:

fast (bool) – Uses argmax(abs(spec)) if True. Tries to find the anchor peaks and align them symmetrically if False. Falls back to fast=True if unable to find any anchor peaks.
peaks_prominence (float) – Used for finding anchor peaks when fast=False. See scipy.signal.find_peaks.
return_peaks (bool) – Returns the peaks found (after normalization) if True.

Return type:

Union[Self, tuple[Self, ndarray]]

correct_phase(method='acme', baseline_correct=False, **kwargs)

Uses the nmrglue’s autops method to correct the phase.

Optionally applies the median base-line correction component-wise.

Parameters:

method – Phase correction method to use. See nmrglue.proc_autophase.autops. Defaults to “acme”.
baseline_correct (bool) – Whether to apply the median base-line correction. Defaults to False.

Return type:

Self

Returns:

Modified version of the object.

kivi_baseline_correction()

Applies the @kivicode’s version of baseline correction.

Return type:: Self

remove_central_peak(base_freq=0)

Finds a pair of downward peaks closest to the base frequency and nullifies the values between them.

Parameters:: base_freq (float) – Frequency of the central peak. Defaults to 0.
Return type:: Self

plot(ax=None, comp_getter=<function real>, **kwargs)

Plots the spectrum with matplotlib.

Parameters:

ax – Axis to plot on. Uses the default one is not provided.
comp_getter (Callable[[ndarray], ndarray]) – Callable to convert the complex amplitudes to real numbers. Defaults to np.real.
kwargs – Additional arguments to be provided to ax.plot(...).

Returns:

The axis on which the spectrum is drawn.

halves_symmetry_score(base_frequency=0)

Calculates the inverse cosine distance between the left and right halves of the spectrum.

Parameters:: base_frequency (float) – Central frequency to split the spectrum with. Defaults to 0.
Returns:: Inverse cosine distance. The lower the score the higher the spectrum’s symmetry rate.
Return type:: float

as_plotly(plot_type='line', comp_getter=<function real>, **kwargs)

Parameters:: comp_getter (Callable[[ndarray], ndarray])

copy()

Return type:: Self

save(save_dir)

Saves the spectrum into frequencies.npy and amplitudes.npy NumPy arrays in the folder provided.

Return type:: None
Parameters:: save_dir (Path)

at(freq)

Return type:: float
Parameters:: freq (float)

classmethod from_file(spectra_dir)

Reconstructs the spectrum object from the files saved by the NMRSpectrum.save(…) method.

Raises:: OSError – If couldn’t find the save files.
Return type:: Self
Parameters:: spectra_dir (Path)

Module contents

use_experimental(flag=True)

Return type:: None
Parameters:: flag (bool)