tau_generator() data

[nikkujensen]

What is the data that is to be used in the tau_generator() function? Is this phase data? If so, why is it needed to generate a list of tau values? Isn't only the lower and upper bounds, the number desired, and the sampling time needed?

[fmeynadier]

tau_generator does more than the name suggests...

https://github.com/aewallin/allantools/blob/master/allantools/allantools.py#L1780

""" pre-processing of the tau-list given by the user (Helper function)

    Does sanity checks, sorts data, removes duplicates and invalid values.
    Generates a tau-list based on keywords 'all', 'decade', 'octave'.
    Uses 'octave' by default if no taus= argument is given.

    Parameters
    ----------
    data: np.array
        data array
    rate: float
        Sample rate of data in Hz. Time interval between measurements
        is 1/rate seconds.
    taus: np.array
        Array of tau values for which to compute measurement.
        Alternatively one of the keywords: "all", "octave", "decade".
        Defaults to "octave" if omitted.
        
        +----------+--------------------------------+
        | keyword  |   averaging-factors            |
        +==========+================================+
        | "all"    |  1, 2, 3, 4, ..., len(data)    |
        +----------+--------------------------------+
        | "octave" |  1, 2, 4, 8, 16, 32, ...       |
        +----------+--------------------------------+
        | "decade" |  1, 2, 4, 10, 20, 40, 100, ... |
        +----------+--------------------------------+
        | "log10"  |  approx. 10 points per decade  |
        +----------+--------------------------------+
    v: bool
        verbose output if True
    even: bool
        require even m, where tau=m*tau0, for Theo1 statistic
    maximum_m: int
        limit m, where tau=m*tau0, to this value.
        used by mtotdev() and htotdev() to limit maximum tau.

    Returns
    -------
    (data, m, taus): tuple
        List of computed values
    data: np.array
        Data
    m: np.array
        Tau in units of data points
    taus: np.array
        Cleaned up list of tau values
    """

(I am surprised this is not collected, or made accessible in the sphinx documentation, I think it could be... I'll have a look !)

[nikkujensen]

This is what I get

>>> import allantools
>>> dir(allantools)
['Dataset', 'Noise', 'Plot', '__all__', '__builtins__', '__cached__', 
'__doc__', '__file__', '__loader__', '__name__', '__package__', 
'__path__', '__spec__', '__version__', 'adev', 'allantools', 
'autocorr_noise_id', 'calc_hdev_phase', 'calc_htotdev_freq', 
'calc_mtotdev_phase', 'ci', 'confidence_interval', 
'confidence_interval_noiseID', 'dataset', 'edf_greenhall', 
'edf_mtotdev', 'edf_simple', 'edf_totdev', 'frequency2fractional', 
'frequency2phase', 'gradev', 'hdev', 'htotdev', 'mdev', 'mtie', 
'mtie_phase_fast', 'mtotdev', 'noise', 'noise_kasdin', 'oadev', 
'oadev_realtime', 'ohdev', 'ohdev_realtime', 'phase2frequency', 
'phase2radians', 'plot', 'realtime', 'tdev', 'tdev_realtime', 'theo1', 
'three_cornered_hat_phase', 'tierms', 'totdev', 'trim_data', 'ttotdev']
>>> 

[fmeynadier]

Oh, sorry for not noticing in the first place : you are using the conda-packaged version. I managed to replicate your issue with the PyPi package 2019.9 too.

Please try with the current "master", it should expose the tau_generator function correctly.

[nikkujensen]

No worries. So the current master is not the same as the current PyPi and conda package?

[fmeynadier]

Correct. I think the version number could be changed to reflect that (but to be honest I have yet to find an intuitive and simple workflow for python package numbering, in spite of the enhancements brought by the pyproject.toml scheme, my git logs are still full of spurious "version bump"... @aewallin I could suggest an update of the packaging based on pyproject.toml, if you wish. I also began to use https://python-poetry.org for my own projects, I think it's fine but maybe not fitting everyone's taste, and does not completely solve the version numbering issue - as far as I understand !).