""" ################### Timeseries Analysis ################### Alongside the basic **data methods** shown in the :doc:`previous tutorial `, :class:`~ATK.Models.Lightcurve` objects also support Lomb-Scargle timeseries analysis. | Generating Power Spectra ======================== The :meth:`~ATK.Models.Lightcurve.pspec` method can be used to generate power spectra from one or more :class:`Lightcurves `. A minimum and maximum frequency must be provided, along with a number of test frequencies in this range. **If no units are given, values are assumed to be in** :math:`\mathbf{days}^{-1}`. This process transforms the :class:`~ATK.Models.DataSet`'s :class:`~ATK.Models.Lightcurve` objects into one or more :class:`~ATK.Models.Powspec` objects. Multi-band Power Spectra ------------------------ By default, :meth:`~ATK.Models.Lightcurve.pspec` combines all bands for each target before computing a combined power spectrum. """ # sphinx_gallery_start_ignore # fmt: off # isort: skip_file from _utilities import format_plot from bokeh.document import Document # sphinx_gallery_end_ignore from ATK import query asassn_query = query("lightcurve", targets=5346631922949364864, survey="asassn", path="example_lightcurve_5.fits.gz") pspec_data = asassn_query.apply("pspec", fmin=0, fmax=10, samples=10000, inplace=False) # sphinx_gallery_start_ignore pspec_data.store("example_pspec.fits.gz") # sphinx_gallery_end_ignore pspec_data.show(show_types=True) # sphinx_gallery_start_ignore pass # sphinx_gallery_end_ignore # %% # # | # # A :class:`~ATK.Models.Powspec` can be plotted like any other kind of data, with :meth:`~ATK.Models.DataSet.plot` or :meth:`~ATK.Models.DataSet.open`: # sphinx_gallery_start_ignore pspec_data.plot() figure = format_plot(pspec_data.figure, 1.5, 1.5) doc = Document() doc.add_root(figure) # sphinx_gallery_end_ignore pspec_data.open() # sphinx_gallery_start_ignore figure # sphinx_gallery_end_ignore # %% # | # # Single-band Power Spectra # ------------------------- # To instead process each band individually, pass ``multiband = False`` to :meth:`~ATK.Models.Lightcurve.pspec`. This treats each band as being entirely separate: pspec_data = asassn_query.apply("pspec", fmin=0, fmax=10, samples=10000, multiband=False, inplace=False) # sphinx_gallery_start_ignore pspec_data.store("example_pspec.fits.gz") # sphinx_gallery_end_ignore pspec_data.show(show_types=True) # sphinx_gallery_start_ignore pass # sphinx_gallery_end_ignore # %% # | # # This generates one power spectrum per band: # sphinx_gallery_start_ignore pspec_data.plot() figure = format_plot(pspec_data.figure, 1.5, 1.5) doc = Document() doc.add_root(figure) # sphinx_gallery_end_ignore pspec_data.open() # sphinx_gallery_start_ignore figure # sphinx_gallery_end_ignore # %% # # | # | # # Phase-Folding Light Curves # ========================== # :class:`~ATK.Models.Lightcurve` objects can be phase folded with :meth:`~ATK.Models.Lightcurve.fold`. **By default,** :meth:`~ATK.Models.Lightcurve.fold` **first generates a power spectrum from which to derive an optimal frequency.** :meth:`~ATK.Models.Lightcurve.fold` therefore accepts all parameters that can be passed to :meth:`~ATK.Models.Lightcurve.pspec`. # # A multiband phase-folded light curve can threfore be generated and plotted as follows: folded_data = asassn_query.apply("fold", fmin=0, fmax=10, samples=10000, inplace=False) # sphinx_gallery_start_ignore folded_data.store("example_folded_lc.fits.gz") # sphinx_gallery_end_ignore folded_data.show(show_types=True) # sphinx_gallery_start_ignore folded_data.plot() figure = format_plot(folded_data.figure, 3, 1.5) doc = Document() doc.add_root(figure) # sphinx_gallery_end_ignore folded_data.open() # sphinx_gallery_start_ignore figure # sphinx_gallery_end_ignore # %% # | # # Frequency Optimisation # ---------------------- # Despite the underlying power spectrum being the same, the photometry in the above example has been folded on a different peak frequency - in this case the first subharmonic (i.e. ``0.5 * fopt``). # # **By default, a phase-dispersion metric is calculated for a set of harmonics either side of the peak frequency** and the best is chosen. This can help to preserve real periodic structure, especially if the modulation is asymmetric as seen here. # # This behaviour can be disabled by passing ``optimise=False`` to :meth:`~ATK.Models.DataSet.apply` - in this case losing the true orbital frequency in favour of the peak frequency in the power spectrum: folded_data = asassn_query.apply("fold", fmin=0, fmax=10, samples=10000, optimise=False, inplace=False) # sphinx_gallery_start_ignore folded_data.store("example_folded_lc.fits.gz") # sphinx_gallery_end_ignore folded_data.show(show_types=True) # sphinx_gallery_start_ignore folded_data.plot() figure = format_plot(folded_data.figure, 3, 1.5) doc = Document() doc.add_root(figure) # sphinx_gallery_end_ignore folded_data.open() # sphinx_gallery_start_ignore figure # sphinx_gallery_end_ignore # %% # | # # Frequency Overriding # -------------------- # To avoid the derivation of an optimal frequency entirely, a frequency can be passed via the ``freq`` parameter: folded_data = asassn_query.apply("fold", freq=2.773227732277323, inplace=False) # sphinx_gallery_start_ignore folded_data.store("example_folded_lc.fits.gz") # sphinx_gallery_end_ignore folded_data.show(show_types=True) # sphinx_gallery_start_ignore folded_data.plot() figure = format_plot(folded_data.figure, 3, 1.5) doc = Document() doc.add_root(figure) # sphinx_gallery_end_ignore folded_data.open() # sphinx_gallery_start_ignore figure # sphinx_gallery_end_ignore # %% # # | # | # | # # .. rubric:: Download this Tutorial