orbitlight

Planets, their atmospheres and the companions of their stars, read out of archived high-resolution spectra of the host stars. A transiting planet's atmosphere, a dayside's thermal emission, a reflected copy of the star and a companion star's own lines all move on an orbit, while the host's lines stand still once every spectrum is placed in the host's frame. orbitlight puts every archived spectrum of a host on one velocity grid in that frame, divides out the host, cross-correlates what remains with a template of lines, and sums the cross-correlations along an orbit: the planet's, from its catalogue ephemeris, or every orbit of a grid when the orbit is unknown. Each sum is read against nulls built from the same spectra.

The spectra come from the ESO Science Archive Facility (HARPS, ESPRESSO, NIRPS, CRIRES, UVES), the Canadian Astronomy Data Centre (CFHT SPIRou, the telluric-corrected products) and the Keck Observatory Archive (NIRSPEC, listed but not fetched). Planets and ephemerides come from the NASA Exoplanet Archive, and stellar velocities from SIMBAD where a product carries none. This repository holds the code, the archive plans behind each search, and the results: limits, detection maps, and the cross-correlation rows from which the companion search runs again without a spectrum being downloaded.

Contents

Path	Contents
orbitlight/, orbitlight.py	the package and its command line
results/transits/	the record of the rocky planets' line tests
results/giants/	one file per giant-planet transit set, the K_p and rest-velocity maps of every set but the two NIRPS sets, the reflected-light results, and the record, a row per planet and instrument
results/companions/rows/	per star, the cross-correlation rows with their epochs (and, where kept, barycentric corrections and velocities) and a note naming the star
results/companions/validation.json	the companion search run blind and with a planted companion on six stars
plans/	the archive's answers behind each search: the rocky planets with their transits matched, the giant planets with their spectra placed on their orbits, and the series of a hundred or more spectra of one star over a year or more
templates/GJ367_optical.npy	the M-dwarf template of the companion search, built from GJ 367's archived HARPS spectra

Method

Every spectrum is read as air wavelengths in the barycentric frame, the SPIRou products being moved there with the barycentric correction their image extensions carry. Each band has one grid uniform in velocity at 1 km/s (3800-7880 Å in the optical, 9500-25000 Å in the near-infrared). A spectrum is shifted into the host's frame, interpolated onto the grid, divided by its running median over 501 samples and by the median of its series' reference spectra, a cubic in wavelength is taken out, the telluric bands are set aside, and the sodium nightglow is masked at its telescope wavelength. Each wavelength's scatter across the reference spectra whitens the cross-correlation, which is normalised so that a row of unit noise gives unit scatter.

Rocky planets in transit (transits). At He I 10830, H alpha and Na D, each night's out-of-transit median is the star and a line in time per velocity bin takes out the night's drift. Each in-transit residual is shifted along the planet's orbital velocity and weighted by the transit's light curve, and the absorption is the mean within 12 km/s of the line against the side bands 30-80 km/s away. The null is the same statistic from every run of consecutive out-of-transit spectra in the transit's number and from 200 random draws, and the trail is also run reversed, which keeps the star's own transit distortion and loses the planet. A claim needs 3 sigma against the null and the forward trail above the reversed one by the same margin; the limit folds the reversed trail into the scatter. The Moon's spectra through the 2012 transit of Venus run as a transiting planet with no exosphere.

Giant planets (giants). search cross-correlates every archived transit set with templates of nine atoms (Kurucz's gfall list, each line weighted by gf exp(-E_low/kT) at the planet's equilibrium temperature, every line within three decades of the strongest carrying its decade's share) and of CO and water (petitRADTRANS line-by-line cross-sections). The out-of-transit median row comes out of every row and, for the atomic templates, a bounded Gaussian Doppler shadow out of each in-transit row. The in-transit rows are summed on a map over K_p and rest velocity, and the orbit's value is read against the map's scatter and a sham of out-of-transit rows given the transit's phases and every step the in-transit rows receive. A detection stands at 4 sigma and a sigma above the sham, with the map's own peak above 30 km/s in K_p, within 10 km/s of rest and on the orbit or its plateau, and with no single line carrying more than half the template's weight. Hosts hotter than 7500 K or rotating faster than 40 km/s are flagged. reflected searches the planets without a transit with the star's own spectrum as the template, five principal components taken out of the rows, the rows weighted by Lambert's phase function, and the gain measured by a 100 ppm copy of the star laid into the same spectra, so that the limit is a contrast. emission searches the dayside of hot giants in SPIRou spectra with SysRem on the residuals and K_p free below the edge-on value.

Companions with the orbit left free (companions). A template star's absorption pattern is cross-correlated with every spectrum of a star in the star's own frame. Out of each row come the star's own pattern at its own strength, the sky's pattern moved into the sky's frame, and five principal components of the rows. The trail is summed along every orbit of a grid in period, phase and amplitude: amplitudes from 40 to 150 km/s; periods at a periodogram's resolution from half a day up to the longest period at which a system of two solar masses allows each amplitude (302 days at 40 km/s, 5.7 days at 150 km/s), with periods near 0.5, 1, 2 and 4 days, the sidereal day and the half year left out; and readings within 30 km/s of the star or within 10 km/s of a row's telluric velocity left out of the sum. The summit is read against three nulls from the same rows: the epochs shuffled, whole nights shuffled, and nights shuffled only among nights at the same time of year, so that what follows the hour of observing or the season stays in the null as it stays in the data. The lowest of the three significances is kept. survey keeps each star's rows, and reanalyse searches kept rows.

Results

Rocky planets. Of the 114 planets below 1.8 Earth radii of stars cooler than 4000 K within 50 pc, 42 have archived spectra taken during a transit, and 13 give four or more spectra on each side of one. Their 25 planet, instrument and line tests, and the 2 of the Moon through the transit of Venus, show no planetary absorption. The 3 sigma limits on helium absorption in the planet's frame are:

Planet	Instrument	In-transit spectra	He I 10830 limit (per cent)
GJ 486 b	SPIRou	86	0.12
HD 260655 b	NIRPS	32	0.16
HD 260655 c	NIRPS	28	0.30
L 98-59 d	NIRPS	40	0.50
L 98-59 c	NIRPS	125	0.57
TOI-270 b	NIRPS	7	0.75
LP 791-18 d	SPIRou	8	2.1
TRAPPIST-1 d	SPIRou	4	9.4
TRAPPIST-1 b	SPIRou	18	15.7

The M dwarfs' H alpha and Na D cores move by 1 to 13 per cent between spectra with the stars' flares and the sky's sodium, and the reversed trail matches every excursion. The Moon through the transit of Venus shows a core residual of -2.9 per cent in H alpha and -2.7 in Na D from the night's airmass, in the forward and reversed trails alike. A 0.5 per cent line injected in the planet's frame returns 0.12 to 0.20 per cent in the core mean, which the limits do not correct for.

Giant planets. 47 transit sets of 42 planets were searched. Vanadium stands in WASP-121 b at 6.2 sigma on the orbit from 105 HARPS in-transit spectra, with the map's peak at K_p 210 km/s and rest velocity -3 km/s against the orbit's K_p of 219, and the sham at 2.8; Ben-Yami et al. (2020) reported it at 4.5 sigma from the same spectra. Water in WASP-76 b reads 4.7 sigma at the orbit's K_p from its 87 archived SPIRou in-transit spectra with the sham at 1.9, but the map's own peak lies at K_p 50 km/s, so the rule holds it. HD 209458 b is empty. Every other species falls short of the rule, below 4 sigma, within a sigma of its sham, off the map's own peak or resting on one line; the strongest of them, WASP-101 b's Fe II at 5.7 sigma against a sham of 0.0, has the map's peak at K_p 30 km/s and -29 km/s from rest. Reflected light from 51 Peg b is absent at 43 ppm (3 sigma) from 300 HARPS and ESPRESSO spectra, the probe copy returned at 0.90 of its strength, which puts the contrast of 6e-5 reported by Martins et al. (2015) at 4.2 sigma in this map. The 299 HARPS spectra of tau Boo give 546 ppm, since they fall near superior conjunction, where the principal components take most of the probe.

Companions. The six-star test of results/companions/validation.json lays a companion into each star's rows at the rows' own noise level on a 7.3-day orbit at 50 km/s:

Star	Blind summit	Planted companion
HD 85512	3.87 d, 40 km/s, -0.6 sigma	7.30 d, 50 km/s, 14.6 sigma
tau Ceti	2.59 d, 70 km/s, 1.5 sigma	7.30 d, 50 km/s, 19.9 sigma
Gl 876	0.55 d, 40 km/s, -1.3 sigma	7.30 d, 50 km/s, -1.1 sigma
HD 114076	112.58 d, 50 km/s, -1.2 sigma	not recovered
alpha Cen A	86.67 d, 45 km/s, 3.7 sigma	not recovered
alpha Cen B	166.76 d, 40 km/s, 6.2 sigma	not recovered

The planted companion is found at its period and amplitude in the two stars with 300 spectra spread over many nights. In Gl 876, with 152 spectra, the summit sits on it but does not stand above the nulls, and HD 114076 does not recover it. alpha Cen A and B, 4 to 10 arcseconds apart, each carry the other's light in the fibre, and their summits are that light, not companions.

Coverage

This release holds part of what the plans define. The companion rows cover 138 of the 372 HARPS series of a hundred or more spectra over a year or more in plans/companions_plan.json, which are 130 stars: the archive files some stars under two or three names, and the two series each of HD 161098 and HD 199288 hold the same spectra. The plan's ESPRESSO, NIRPS, UVES and CRIRES series are not searched. Per-star results of the companion search are not included; reanalyse produces them from the rows. The giant-planet search covers 43 HARPS and ESPRESSO transit sets with four or more spectra on each side of the transit, the SPIRou sets of WASP-76 b and HD 209458 b, and the NIRPS sets of WASP-76 b and WASP-121 b. WASP-76 b's NIRPS set is marked not measured, its telluric residuals driving the sham map to -243 sigma, and WASP-121 b's holds no species (water at -0.4 sigma, CO at -2.7); the CRIRES, UVES and NIRSPEC sets are not searched. Reflected light has been searched for 51 Peg b and tau Boo b only. The emission mode has no result here: its run on tau Boo b's dayside SPIRou spectra does not recover the published CO at the planet's known K_p.

Running

pip install numpy scipy astropy torch      # the companion trail search needs torch, on a GPU or the CPU; the cross-correlations fall back to numpy's FFT without it
export ORBITLIGHT_DATA=/path/to/data       # spectra and plans; results under ORBITLIGHT_RESULTS, default <data>/results

python orbitlight.py transits plan          # archives queried, transits matched
python orbitlight.py transits fetch --streams 2
python orbitlight.py transits record

python orbitlight.py giants plan --vmax 11
python orbitlight.py giants fetch --instruments HARPS ESPRESSO --min-in 4
python orbitlight.py giants search --planets "WASP-121 b"
python orbitlight.py giants reflected --planets "51 Peg b"

python orbitlight.py companions plan
python orbitlight.py companions survey --inst HARPS --per-star 300 --ccf-only --template templates/GJ367_optical.npy
python orbitlight.py companions reanalyse --inst HARPS --part 0/2   # trail searches, in parallel shares

The giant-planet templates read their line data from the data root: Kurucz's gfall08oct17.dat in $ORBITLIGHT_DATA/giants/lines/ for the atoms, with NIST's lines standing in where it is absent, and petitRADTRANS's line-by-line cross-sections (wlen.dat and the sigma_* files) in $ORBITLIGHT_DATA/giants/opacities/CO_main_iso/ and H2O_main_iso/ for the molecules, with HITRAN's lines standing in where they are absent.

To search the kept rows, copy results/companions/rows/ to $ORBITLIGHT_DATA/companions/ccf/ and plans/companions_plan.json to $ORBITLIGHT_DATA/companions/plan.json, then run companions reanalyse. Each mode's --help lists its options.

Acknowledgements

Based on data obtained from the ESO Science Archive Facility. This research used the facilities of the Canadian Astronomy Data Centre, operated by the National Research Council of Canada with the support of the Canadian Space Agency, and observations obtained at the Canada-France-Hawaii Telescope. It has made use of the Keck Observatory Archive, operated by the W. M. Keck Observatory and the NASA Exoplanet Science Institute under contract with NASA; of the NASA Exoplanet Archive, operated by the California Institute of Technology under contract with NASA under the Exoplanet Exploration Program; and of the SIMBAD database, operated at CDS, Strasbourg, France. The atomic lines are Kurucz's gfall list, and the molecular cross-sections are petitRADTRANS's (Mollière et al. 2019).

Licence

Code is under the MIT License and everything else under CC BY 4.0; see LICENSE.