Data

kima can load different kinds of data, either from column-based text files or Python arrays. The different data classes have some useful attributes and methods.

Radial velocity time series (RVData)

The simplest dataset is a time series of radial velocity measurements from one single instrument. The RVData class is used to load these data from a file, which must contain at least three columns:

data.txt

time   RV   RVerror
...    ...  ...

and can be loaded as

from kima import RVData

data = RVData('data.txt', skip=1)

The columns in the file can be space (␣), TAB (⇥), or comma (,) separated.

Note

The file can have more columns; by default, kima will read the first three and ignore any remaining ones.
Aso, you can actually mix separators in the same file! But please don't...

Regarding units, the first column is assumed to be in days (although this is not strictly enforced), and the RVs and uncertainties are by default assumed to be given in m/s. Most importantly, they both have to be in the same units!

The skip argument can be used to skip the first line(s) of the file.

If you want to load RV data from multiple instruments, simply provide a list of filenames:

instrument 1instrument 2

data_inst1.txt

time   RV   RVerror
...    ...  ...

data_inst2.txt

time   RV   RVerror
...    ...  ...

data = RVData(['data_inst1.txt', 'data_inst2.txt'], skip=1)

Useful methods

A few methods of the RVData class can be useful, e.g., when setting priors:

data.get_timespan()
data.get_RV_span()

The full documentation is available here.

Quick plot

The RVData class has a plot method that allows you to quickly check the data:

data.plot() #(1)!

1. This plot show the RVs from the 51 Peg example.

Activity indicators

Sometimes it's useful to use activity indicators simultaneous to the RVs. These can be considered as basis vectors for decorrelation or for joint GP fits, for example. The RVData class can also load these data if they are stored in the subsequent columns of a file:

data_with_indicators.txt

time   RV   RVerror   ind1    ind2
...    ...  ...       ...     ...

which can be loaded as

data = RVData('data.txt', skip=1, indicators=['ind1', 'ind2'])

The data is stored in the actind (read-only) attribute of the RVData object, which appears as a list of lists.

Note

A particular column of the file can easily be skipped by providing an empty string as the indicator name:

data = RVData('data.txt', skip=1, indicators=['', 'ind2'])

Provided the indicators were loaded, the RVmodel.indicator_correlations option can be used to model linear correlations. On the other hand, some models, like the RVFWHMmodel for example, assume that the provided RVData object has two activity indicators corresponding to the FWHM and its uncertainty.

Gaia astrometric time series (GAIAdata)

This should only ever be a single datafile with 5 columns at least

gaia_data.txt

time   along-scan position   along-scan position error   scan angle   parallax factor along-scan
...             ...                     ...                 ...                  ...

These should all be provided within the Gaia data. These 5 columns in this order are required, after which you can add other columns useful to you in analysis if you like.

This data can be loaded as

from kima import GAIAdata

data = GAIAdata('gaia_data.txt', skip=1)

Hipparcos-Gaia proper motions (HGPMdata)

TODO