astro-datalab / notebooks-latest

Default set of Data Lab notebooks, by DL team and contributed by users
BSD 3-Clause "New" or "Revised" License
60 stars 48 forks source link

HVS WD science example notebook #156

Closed jacquesalice closed 1 year ago

jacquesalice commented 1 year ago

This is a draft version. Still have to rephrase some descriptions and figure out why the second plot does not exactly match the Figure 3 plot in the paper

jacquesalice commented 1 year ago

Thanks for the review @rnikutta ! I've updated the notebook with some of the changes you suggested. See my comments on the other suggestions below:

  • Query to select objects within 100pc: Why not select (in addition to some quality cuts) on: ... WHERE distance_gspphot < 100 (this field is given in parsecs)

There are a good amount of NaNs for the distance_gspphot column. 1,340,950,508 out of the 1,811,709,771 total rows are NaN. When I tried this in the NB, I only got 88,130 rows back (with a limit of 250,000) and the resulting plot lost all of the data points along the white dwarf sequence:

Screen Shot 2023-05-30 at 2 06 51 PM

Comments on motivating the steps of the science NB: some of the steps have either very short or no motivation. For instance:

  • Why are all these constraints applied to the queries?
 parallax_over_error > 4
 AND parallax > 0.25
 AND ruwe < 1.4
 AND ipd_frac_multi_peak <= 2
 AND ipd_gof_harmonic_amplitude < 0.1
 AND astrometric_sigma5d_max < 1.5
 AND phot_g_mean_mag - 5 * log10(1000.0 / parallax) + 5 > 6 + 5 * (bp_rp)

Quality cuts? Selection criteria (motivated how?) Something else?

I mention that "The following constraints are applied in the query in order to exclude objects with large uncertainties", and that's really all they give in the paper. From the paper:

"In order to exclude objects with large uncertainties, we included in our selection only objects which satisfy the following conditions: (1) ๐œ›'/๐œŽ๐œ› > 4, i.e. having a relative error of parallax measurement below 0.25; and (2) ๐œ›' > 0.25 mas, i.e. objects with nominal distances that are smaller than 4 kpc. We check the quality of the astrometric solution following the criteria of Fabricius et al. (2021), selecting only objects with (1) renormalised unit weight error (RUWE) < 1.4, (2) IPD_FRAC_MULTI_PEAK < 2, (3) IPD_GOF_HARMONIC_AMPLITUDE < 0.1 and (4) ASTROMETRIC_SIGMA5D_MAX < 1.5. We then apply the colour-magnitude cut suggested by Gentile Fusillo et al. (2021): ๐บ_abs > 6 + 5(๐บ_BP โˆ’ ๐บ_RP)."

  • The Section "Heights above the Galactic plane of the observed HVS WD candidates" also needs a bit more motivation. Why this plot? What do we expect to see? What doe we see? There's some of that in the text, but does not come across well-motivated. Things should follow from the previous steps.

In the paper, this plot is used to potentially provide an alternative age estimate for the candidates whose WD cooling ages and total ages (since zero-age MS) couldn't be estimated because their masses were lower than 0.5 solar masses and "could not form through normal stellar evolution of a single stellar progenitor over a Hubble time". I'll include this in the NB. Other potentially relevant info:

"Monari et al (2018) find the local escape velocity from the Galaxy to be 580ยฑ63 km sโˆ’1 , and suggest it decreases monotonically between 640 km sโˆ’1 at 4 kpc to 550 km sโˆ’1 at 11 kpc (Galactocentric distances). It is therefore likely that our two fastest new HVS WD candidates are unbound hypervelocity WDs kicked following an explosive event, or strong dynamical interaction. The next 8 candidates in Table 1 have tangential velocities ranging between 520 โˆ’ 565 km sโˆ’1 . A non-negligible radial-velocity component could potentially make these WDs be unbound hypervelocity WDs, but overall these might be bound WDs, in which case they are likely to be on highly eccentric orbits. In principle, these just might be the extreme tail of normal halo WDs. Distinguishing between these possibilities require knowledge on the radial-velocity component and/or a good age estimate, given that halo WDs are expected to be old. We have searched for archival data of radial velocities for these objects but found no additional data.

Since halo-formed WDs originate from very old (> 10 Gyr; Jofrรฉ & Weiss 2011; Kilic et al. 2019) populations, identifying younger WDs among these would suggest a disc origin, and hence a large kick, in order to explain their measured velocities. WDs involved in a SN explosion might also have been heated through accretion of material (dynamical detonation in the double-degenerate case; Shen et al. 2018b) or a weak deflagration (for the Iax SNe, as we originally suggested; Jordan et al. 2012), and appear peculiar and/or younger. D6-1โ€“D6-3 for example, have peculiar positions on the HR diagram, suggested to be related to material accretion from the exploding companion (Shen et al. 2018b), while LP 93-21 shows a peculiar composition, suggested to be related to a Iax SN (Ruffini & Casey 2019), and similarly (Vennes et al. 2017; Raddi et al. 2018b) for the LP 40-365 object.

For a sub-sample of our candidates we can estimate the WD cooling ages, as well as the total ages (since zero-age MS), using the WD_models Python package5 . However, this approach is limited to more massive WDs. WDs of masses lower than 0.5 M could not form through normal stellar evolution of single stellar progenitor over a Hubble time. These He or hybrid HeCO WDs (Zenati et al. 2019) have likely undergone a binary evolution stripping process. It is therefore difficult to estimate their true age, in that case, and we cannot exclude a halo origin. To some extent, this could also be the case for slightly more massive WDs which might have been affected by binary evolution, even if their mass is consistent with the age of the Galaxy, in which case their total ages might appear older than they are. The WDs with estimated masses and ages are shown in Table 5. We plot constant age contours for WDs in Figure 3. We also compute the kinematic ages as ๐‘/๐œ‡๐‘ for WDs where the sign of proper motion in latitudinal direction coincides with the sign of Galactic latitude. Typical oscillations in the Galactic gravitational potential occur on timescales comparable to 100 Myr, thus even WDs with cooling ages of 0.2 โˆ’ 0.5 Gyr could have completed a few oscillations if they are bound. The only source with comparable kinematic and cooling age is Gaia DR3 3611573712136684928, but even in this case the cooling age is only one order of magnitude larger than kinematic age.

In order to potentially provide an alternative age estimates for the other candidates, we also provide the height of the observed WDs above the disc, see Figure 4. Casagrande et al. (2016) show that the vast majority of stars residing below 1 kpc from the plane are stars younger than the 10 Gyr age of halo stars. Most of our HVS WD candidates reside below 1 kpc from the plane with the majority below 0.5 kpc (see Table 6). These findings are consistent with a disc origin for the majority of the sample. One should note that due to their low luminosities, one cannot identify very far WDs, which therefore a priori limits the largest distances, and hence also the heights above the disc. We conclude that excluding a halo origin for any individual WD (with no age estimate) in our sample is challenging, but that most of our candidate HVS WDs are likely to have a disc origin, and require a non-trivial velocity kick. Nevertheless, radial-velocity follow-ups are required to better constrain/confirm their origin."