Detailed walkthrough

Let's look in more detail at what arvi can do.

We will use HD69830 as an example, since this target has several public datasets in DACE.

In [2]:

from arvi import RV

from arvi import RV

In [3]:

s = RV('HD69830')

s = RV('HD69830')

querying Simbad...

querying Gaia...

querying DACE for HD69830...

requesting DACE data as public

RVs available from

      HAMILTON ├ 2021ApJS..255....8R -  (43 observations)

               └ 2014ApJS..210....5F -  (32 observations)

         HARPS ├ 2006Natur.441..305L -  (74 observations)

       HARPS03 ├ 3.2.5 - HARPS (199 observations)

  HIRES-post04 ├ 2021ApJS..255....8R -  (529 observations)

         HIRES ├ 2017arXiv170203571B -  (439 observations)

           APF ├ 2021ApJS..255....8R -  (273 observations)

selecting latest pipeline (2021ApJS..255....8R) for HAMILTON

using Gaia information to remove secular acceleration

removing secular acceleration from RVs

subtracted weighted average from HAMILTON  : (-0.864 m/s)

subtracted weighted average from HARPS     : (30289.137 m/s)

subtracted weighted average from HARPS03   : (30138.473 m/s)

subtracted weighted average from HIRES_post04: (-1.141 m/s)

subtracted weighted average from HIRES     : (-0.184 m/s)

subtracted weighted average from APF       : (-1.165 m/s)

subtracted weighted averages from ('fwhm', 'bispan')

The DACE query returned observations from several instruments. Since these observations are public and may have been published in a refereed paper, some are identified by the publication id (e.g. 2021ApJS..255....8R).
Inside arvi, these are called pipelines. Note that arvi automatically selects the latest pipeline (in this case, the most recent publication) for HAMILTON.

By default, the secular acceleration and a weighted average per instrument are subtracted from the RVs. This puts all RVs around zero. Let's plot the data:

In [4]:

fig, axs = s.plot()
fig

fig, axs = s.plot() fig

Out[4]:

Note that there is data for both HIRES and HIRES_post04 as well as HARPS and HARPS03. This corresponds to the same instrument but different reductions (and different publications). Some of the observations will be in common, but not all. We can easily remove the HIRES and HARPS instruments, assuming that HIRES_post04 and HARPS03 have the most up-to-date data.

In [5]:

s.remove_instrument('HIRES', strict=True)
s.remove_instrument('HARPS', strict=True)

s.remove_instrument('HIRES', strict=True) s.remove_instrument('HARPS', strict=True)

Removed observations from 'HIRES'

Removed observations from 'HARPS'

Note

We used the strict=True option above to select only HIRES and HARPS. By default, the match with the instrument names is done recursively as 'HIRES*', so without strict=True we would have removed both HIRES and HIRES_post04 as well as HARPS and HARPS03.

Let's see the update datasets

In [6]:

fig, axs = s.plot()
fig

fig, axs = s.plot() fig

Out[6]:

Attributes¶

The s object holds several useful attributes related to the observations and to the target.

For example, you may notice above that arvi queried Simbad and Gaia for the target name. The results are stored as attributes

In [7]:

s.simbad

s.simbad

Out[7]:

HD69830 (V=5.95, sp_type='G8:V')

In [8]:

s.gaia

s.gaia

Out[8]:

HD69830 (DR3 id=5726982995343100928)

which themselves have interesting attributes

In [9]:

s.simbad.B, s.simbad.V, s.simbad.bmv
s.simbad.sp_type
s.simbad.coords
s.simbad.plx
...;

s.simbad.B, s.simbad.V, s.simbad.bmv s.simbad.sp_type s.simbad.coords s.simbad.plx ...;

In [10]:

s.gaia.radial_velocity, s.gaia.radial_velocity_error
...;

s.gaia.radial_velocity, s.gaia.radial_velocity_error ...;

Each instrument is also stored as an attribute of s and is itself an RV object:

In [11]:

s.HARPS03

s.HARPS03

Out[11]:

RV(star='HD69830', N=199)

The data themselves are stored as numpy arrays

In [12]:

s.time
s.vrad
s.svrad
...;

s.time s.vrad s.svrad ...;

Note

The names vrad and svrad have become somewhat standard in the community to refer to the radial velocities and their uncertainties.

Masking¶

A central concept in arvi is masking of observations. The s object holds a boolean mask that indicates which observations are considered

In [13]:

s.mask

s.mask

Out[13]:

array([ True,  True,  True, ...,  True,  True,  True], shape=(1044,))

and several methods can modify this mask. For example

In [14]:

s.remove_point(1)

s.remove_point(1)

removing point 1 from HAMILTON

In [15]:

s.mask

s.mask

Out[15]:

array([ True, False,  True, ...,  True,  True,  True], shape=(1044,))

Other useful methods are

In [16]:

s.remove_before_bjd
s.remove_after_bjd
s.remove_before_bjd
s.remove_prog_id
s.remove_non_public
s.remove_condition
...;

s.remove_before_bjd s.remove_after_bjd s.remove_before_bjd s.remove_prog_id s.remove_non_public s.remove_condition ...;

The s object itself reflects the number of masked points

In [17]:

s

s

Out[17]:

RV(star='HD69830', N=1044, masked=1, 4 instruments)

and the data without the masked points can be accessed with

In [18]:

s.mtime
s.mvrad
s.msvrad
...;

s.mtime s.mvrad s.msvrad ...;

or

In [19]:

s.time[s.mask]
...;

s.time[s.mask] ...;

Warning

The remove_instrument method we used above does not modify s.mask but instead completely removes the data from that instrument.

Plots and analysis¶

We already saw the s.plot() function, but it's easy to plot other quantities than the RVs. For example

In [20]:

fig, axs = s.plot_rhk()
fig

fig, axs = s.plot_rhk() fig

43 NaN values for HAMILTON

529 NaN values for HIRES_post04

273 NaN values for APF

Out[20]:

Note

Only the HARPS03 data has log R'HK values, which is why the function warns us about NaN values for the other instruments.