Data format and SED data¶

Data format for Data object¶

The SED data are internally stored as astropy tables, but it is very easy to import from

1. ascii files
2. numpy array in general

once that is clear the data format. The easiest way to understand the data format is to build an empty table to have a look at the structure of the table:

we can easily access the astropy table

• x column is reserved to frequencies (mandatory)
• y columm is reserved to fluxes (mandatory)
• dx columm is reserved to the error on the frequency,or bin width
• dy columm is reserved to the error on the fluxes
• UL columm is reserved to the flag for Upper Limit
• T_start and T_stop are used to identify the time range to select data using the class ObsData
• data_set

columns with units are implemented using the Units module of astropy (https://docs.astropy.org/en/stable/units/).

and we can easily access the metadata

• z: the redshift of the object
• UL_CL: the CL for the UL
• restframe: possible valuesobs or src, indicating if the data are observed flux, or luminosities, respectively
• data_scale: possible valueslin-lin or log-log, indicating if the data are in linear or logarithmic scale, respectively
• obj_name: the name of the object

Loading from astropy table¶

you can use the default SEDs distributed with the package to get familiar with data handling

# %ECSV 0.9
# ---
# datatype:
# - {name: x, unit: Hz, datatype: float64}
# - {name: dx, unit: Hz, datatype: float64}
# - {name: y, unit: erg / (cm2 s), datatype: float64}
# - {name: dy, unit: erg / (cm2 s), datatype: float64}
# - {name: T_start, unit: MJD, datatype: float64}
# - {name: T_stop, unit: MJD, datatype: float64}
# - {name: UL, datatype: bool}
# - {name: data_set, datatype: string}
# meta: !!omap
# - {z: 0.0308}
# - {restframe: obs}
# - {data_scale: lin-lin}
# - {obj_name: 'J1104+3812,Mrk421'}
# schema: astropy-2.0
x dx y dy T_start T_stop UL data_set
2299540000.0 0.0 1.3409e-14 3.91e-16 0.0 0.0 False campaing-2009
2639697000.0 0.0 1.793088e-14 3.231099e-26 0.0 0.0 False campaing-2009
4799040000.0 0.0 2.3136e-14 2.4e-16 0.0 0.0 False campaing-2009

Saving Data object to a file¶

the data can be loaded from the saved table

Importing from an arbitrary ascii file or numpy array to Data object¶

Assume that your data are stored in an ASCII file named 'test-ascii.txt', with

• x in the first column with frequency in Hz ,
• y in the second column with fluxes in erg cm-2 s-1,
• dy in the third column with the same units as y
• the data are in log-log scale

of course the column number depends on the file that you are using, this is only an example

then you can set the meatdata as follows

of course this method applies if you have a generic 2-dim numpy array.

Importing to Data object from a generic astropy table mapping columns¶

If you want to use a TABLE with arbitrary column names, you can use an import dictionary, mapping the input name to the target. E.g. assume that you column in the input table column named freq that should target the x column, and another named freq err associated to dx you can simply pass the dictionary to the from_file method:

Importing from the ASI ssdc sedtool to Data object¶

To import data from a data file downloaded from the asi ssdc sedtool: https://tools.ssdc.asi.it/SED/

Building the SED the ObsData object¶

As you can see all the meta-data have been properly sourced from the SED file header. You also get information on the length of the data, before and after elimination of duplicated entries, and upper limits

Plotting ObsData¶

or you can create the object to plot on the fly in this way

you can rescale your plot

plotting in the src restframe

to have interactive plot in jupyter

if you want to to have interacitve plot in a jupyter notebook use:

.. code-block:: no

%matplotlib notebook

to have interactive plot in jupyter lab use:

.. code-block:: no

%matplotlib widget

Grouping data¶

As you can see, due to the overlapping of different instruments and to different time snapshots, some points have multiple values. Although this is not a problem for the fit process, you might want to rebin (group) your data. This can be obtained with the following command:

Handling errors and systematics¶

Another important issue when dealing with fitting of data, is the proper handling of errors. Typically one might need to add systematics for different reasons:

• data are not really simultaneous, and you want to add systematics to take this into account
• data (typically IR up to UV), might have very small errors compared to those at higher energies. This might bias the minimizer to accommodate the parameters in order to fit 'better' the low frequencies branch.

For these reasons the package offer the possibility to add systematics

Filtering data sets¶

we use the show_data_sets() method to have know wich data sets are defined in our table

we use show_dataset=True to have the legend of all the datasets

we filter out the data set -1 using the filter_data_set() method. Please not with exclude=True we exclude dataset in filters

we can pass more datasets, comma separated

we can also use filter_data_set to exclude only the datasets in filters with exclude=False

we can revert sed_data to the original state with the reset_data() method

Saving sed_data and loading¶

you can save and relaod you sed_data