iiinova — where your digital self keeps evolving

William Thompson @wrtastro.bsky.social · 09/12/2025

SPIDERS is installed on the Subaru Naysmith platform and we got first fringes on from the upstream AO system. Next step, a real star! Huge thanks to Subaru, NAOJ, and the SCExAO teams for accommodating us. #exoplanets #instrumentation

0201

William Thompson @wrtastro.bsky.social · 23/10/2025

(most of) The local team together to see SPIDERS off! Fingers crossed the ride over wont be too bumpy 🫨

group photo around an astronomical instrument, inside a lab with a green wall in the background.

angle view of an astronomical instrument on a pallet

140

William Thompson @wrtastro.bsky.social · 18/10/2025

SPIDERS is out of the clean room and almost ready to ship — it’s really happening! #instrumentation #exoplanets

1413

William Thompson @wrtastro.bsky.social · 15/07/2025

Octofitter v7, now released! This release overhauls the modelling syntax and add some powerful new capabilities for advanced users. [1/4] (Octofitter is a package for Bayesian modelling of exoplanet orbits)

1141

William Thompson @wrtastro.bsky.social · 10/07/2025

Position of Epsilon Eridani b in-front/behind the star, and phase angle over time for those interested. Updated pre-print should be ready soon! 🔭 #exoplanets

a plot of a planet's position vs the star in 2D indicated as coloured lines. The line colour goes towards red or blue as the planet is in front of the star or behind. It is blue on the left, red on the right, and green, indicating 0, at the top and bottom. The separation is about 1000 milliarcseconds left or right increasing to about 1300 at top or bottom, creating an oval.

0433

William Thompson @wrtastro.bsky.social · 06/07/2025

There’s something very special about the gulf islands 🏝️

130

William Thompson @wrtastro.bsky.social · 03/06/2025

I was honoured today to give the Plaskett medal award talk to open this years Canadian Astronomical Society meeting in Halifax! Great to share our work on #exoplanets imaging and modelling 🔭

1231

William Thompson @wrtastro.bsky.social · 22/05/2025

Octofitter v6 just released! 🐙 * Overhauled Hipparcos / Gaia modelling * Multi-planet epicycle approximation * Include astrometric jitter, north angle, and plate-scale uncertainties * Dynamical priors (prevent crossing) * Celerite gaussian processes Much more! (fig. credit in reply) #exoplanets 🔭

1245

William Thompson @wrtastro.bsky.social · 29/04/2025

Here's a bonus video of SPIDERS maintaining a dark hole in real time, despite simulated turbulence!

0112

William Thompson @wrtastro.bsky.social · 29/04/2025

Recent images from our SPIDERS pathfinder instrument demonstrating the self-coherent camera--next generation tech for direct imaging of exoplanets! We use a special coronagraph to interfere the starlight with itself, like a classic double slit experiment... [1/3] 🔭 #Instrumentation #exoplanets

fringed image of star behind coronagraph

unfringed image of star behind coronagraph

4512

William Thompson @wrtastro.bsky.social · 25/03/2025

I learned today that am being awarded the J.S. Plaskett medal for most outstanding doctoral thesis in astronomy in Canada. Thank you @casca.ca.web.brid.gy, and my committee for the nomination! 🔭

CASCA is pleased to announce Dr. William Thompson as the 2025 recipient of the Plaskett Medal for the most outstanding doctoral thesis in astronomy or astrophysics.

Dr. Thompson’s thesis, defended at the University of Victoria in 2023 under the direction of Profs. Falk Herwig and Christian Marois, is entitled, “Lights in Motion: Observing Nearby Planets with Imaging, Wavefront Sensing, Orbital Detection, and Spectroscopy”. The title hints at the breadth of the research described within Dr. Thompson’s thesis work, including developing new computational methods to analyze angular differential imaging observations and employing these methods to discover a new candidate exoplanet around HR 8799; developing the first multi-technique exoplanet finding code that can incorporate radial velocity, astrometry and high-contrast imaging data; carrying out both simulation and laboratory tests of two new techniques for exoplanet imaging; and more. The thesis external examiner notes that, “it is the combination of innovative thinking, rigorous simulation and laboratory testbed results which make the work so impressive.” The software developed as part of Dr. Thompson’s thesis research is publicly available and is already being used by other exoplanet researchers, and his instrumentation development work provides first steps toward possible hardware contributions to the Habitable Worlds Observatory. CASCA is delighted to recognize Dr. Thompson for these outstanding contributions.

La CASCA a le plaisir d’annoncer que le M William Thompson, Ph. D., est le lauréat de la médaille J.S. Plaskett 2025 pour la thèse de doctorat la plus remarquable en astronomie ou en astrophysique.

La thèse du M Thompson, Ph. D., soutenue à l'Université de Victoria en 2023 sous la direction des Profs. Falk Herwig et Christian Marois, est intitulée « Lumières en mouvement : observation de planètes proches à l'aide de l'imagerie, de la détection de front d'onde, de la détection orbitale et de la spectroscopie ». Le titre fait allusion à l'ampleur de la recherche décrite dans la thèse. Les travaux de thèse du M Thompson, Ph. D., comprenaient le développement de nouvelles méthodes de calcul pour analyser les observations d'imagerie différentielle angulaire et l'utilisation de ces méthodes pour découvrir une nouvelle exoplanète candidate autour de HR 8799 ; le développement du premier code multitechnique de recherche d'exoplanètes capable d'intégrer la vitesse radiale, l'astrométrie et les données d'imagerie à contraste élevé ; la réalisation de simulations et de tests en laboratoire de deux nouvelles techniques d'imagerie d'exoplanètes, et plus encore. L'examinateur externe de la thèse note que « c'est la combinaison d'une pensée innovante, d'une simulation rigoureuse et de résultats de banc d'essai en laboratoire qui rend le travail si impressionnant ». Le logiciel mis au point dans le cadre de la thèse du M Thompson, PhD, est accessible à tous et est déjà utilisé par d'autres chercheurs sur les exoplanètes, et son travail de développement d'instrumentation constitue un premier pas vers d'éventuelles contributions matérielles à l'Observatoire des mondes habitables. La CASCA est ravie de souligner les réalisations de M Thompson, Ph. D., en lui décernant ce prix.

6260

William Thompson @wrtastro.bsky.social · 04/03/2025

Here are some individual draws from the posterior. I think any individual row should work. The angular parameters are in units of radians, period in years, tp in MJD (ps: plain text copyable version in the alt-text). We'll post the full posteriors and reproducible scripts once accepted.

A plain text data table that reads:

M_pri b_e b_ω b_τ b_P b_i b_Ω b_a b_tp b_mass
Float64 Float64 Float64 Float64 Float64 Float64 Float64 Float64 Float64 Float64
──────────────────────────────────────────────────────────────────────────────────────────────────
0.831638 0.168298 3.89686 0.858561 7.21718 0.640533 3.29034 3.51345 59651.7 1.06503
0.795307 0.0308685 5.94108 0.169789 7.28537 0.874402 2.96544 3.48309 57840.3 0.879026
0.849784 0.133792 4.07826 0.885673 7.3039 0.7676 3.17299 3.56694 59751.3 0.962535
0.828249 0.0122253 2.14998 0.58055 7.33537 0.621978 3.60262 3.54694 58943.9 1.11392
0.830251 0.120784 5.83218 0.16658 7.32244 0.914066 3.4074 3.54528 57834.0 0.866626
0.843352 0.105946 4.02053 0.889047 7.3851 0.739503 3.33878 3.58424 59786.6 0.955619
0.843297 0.0298729 0.202907 0.251375 7.37376 0.68507 3.18727 3.58067 58065.5 1.08521
0.842145 0.0804329 5.19621 0.0807889 7.38184 0.878696 3.38147 3.58141 57606.3 0.903373
0.831515 0.0393159 5.18298 0.0626671 7.43359 0.73388 3.33258 3.5833 57558.6 1.16991
0.802696 0.0307339 0.643902 0.324703 7.22671 0.642977 3.42096 3.47524 58245.6 1.01233

311

William Thompson @wrtastro.bsky.social · 03/03/2025

We are now able to predict the planet's location to a narrow region on any given date. This figure shows how these predictions depend on the different data sources included. [5/7]

predictions of the planet's location on 2025-01-01 shown as 1-sigma contours, broken down by data source. A green ring is shown for the RV data, while a black oval (much more concentrated) shows the planet is likely to be South of the star, about 1-arsecond separation.

1120

William Thompson @wrtastro.bsky.social · 03/03/2025

There's been lots of interest in this planet in the intervening years, of course. Now with JWST, we hope that an imaging detection is near! In anticipation, we've updated the orbit based on a full 40 years of RV data, and astrometry from Hipparcos, Hubble, Gaia DR2, and DR3. [3/7]

Figure showing the RV orbit fit to data staring in 1981 and going until mid 2020. Data from 13 different RV instrument groups are coloured differently.

Orbit posterior draws compared to RV and the various absolute astrometry sources.

1120

William Thompson @wrtastro.bsky.social · 03/03/2025

The first hints of ε Eri b, also known as "AEgir," were reported in 1995 by Campbell, Walker and Yang based on RV data from CFHT in the 80s, then confirmed with RV by Hatzes+ in 2000, and with the addition of Hubble absolute astrometry in 2006 by Benedict+ (see fig.) [2/7]

1100

William Thompson @wrtastro.bsky.social · 03/03/2025

Time for the promised thread on ε Eridani b! I should emphasize this manuscript is a submitted draft—comments from the community are very welcome. Bottom line: ε Eri b is likely a 1.0 Jupiter-mass planet, only 3.2pc away, on a 3.55 AU near-circular orbit, ~aligned with its debris disk. 🔭🪐🧪 [1/7]

Figure showing a the plausible orbits for Eps Eri b, and a panel showing the scale of the orbits relative to the outer debris disk model of Booth et al.

A three panel figure showing the statistical distribution of inclination versus mass for the planet, broken down by seven different combinations of data. The scatter points follow an m*sin(i) curve, except for a series that does not include any RV data. A small black concentrated region shows the full model, with 1 and 3 sigma contours. The planet's mass is quoted at 0.98 +- 0.09 Mjup, and inclination as 41 +- 7 degrees.

98223

William Thompson @wrtastro.bsky.social · 27/02/2025

Yes indeed! We find a mutual inclination of 10 ± 7 °. My money is on them being close to co-planar, and this small signal comes from systematics in the astrometry or disk models. But only time will tell for sure.

130

William Thompson @wrtastro.bsky.social · 27/02/2025

We've been cooking up some fun new results for the planet Epsilon Eridani b! Submitted and should be on the arxiv by Monday. In the meantime, here's a couple teaser figures :) 🪐🔭🧪

3305

William Thompson @wrtastro.bsky.social · 27/02/2025

Stabilized variational non-reversible parallel tempered sampling—this MCMC can handle just about anything you throw at it! pigeons.run 🧪

1123

William Thompson @wrtastro.bsky.social · 11/02/2025

Lastly we pin down the mutual inclination of the orbit of the Ba/Bb pair versus their orbit around the primary, A, at 31.0±2.5°

060

William Thompson @wrtastro.bsky.social · 11/02/2025

Ultimately, we find that we can determine the dynamical mass of the binary pair independently from the inner orbit—and it agrees to better than 1% with the mass we find from the outer orbit... a great test of our ability to measure the masses of directly imaged planets! [3/4]

A plot that shows the mass of the binary on the horizontal axis, probability density on the vertical axis, with two curves labelled "inner model" and "outer model" along with the text "agree within: 0.7%, 0.9 sigma, 1 Mjup".

170

William Thompson @wrtastro.bsky.social · 11/02/2025

First, we show that we can find a low SNR companion in the GRAVITY data despite the rapid orbital motion, and that orbit agrees near-perfectly with RV data. Very promising for planning future observations of faint, rapidly orbiting planets! [2/4]

160

William Thompson @wrtastro.bsky.social · 11/02/2025

Perhaps a little premature since it's still under review, but I'd like to share a couple results from our preprint now up on the arXiv: arxiv.org/abs/2502.05359 🔭 We dive deep into the orbits of the binary brown dwarf companion 🪐 GL229 Ba / Bb newly discovered this fall by Xuan et al. [1/4]

A two-panel figure, showing plausible orbits of the outer companion, then a zoom in (labelled "1000x zoom") to show the inner orbit of the two brown dwarfs.

1205

William Thompson @wrtastro.bsky.social · 19/12/2024

Very minor update to PairPlots that addressing some vexing typographical details in the margin credible limits. They now exactly match the font of the rest of the plot. Before VS after, plus multiple series now have limits printed. Yes, I’m probably a little obsessive…

070

William Thompson @wrtastro.bsky.social · 13/12/2024

So I know the central limit theorem is a thing, but can anyone explain why my daily # of GitHub commits follow a normal distribution centred on Wednesday?? #GitHubWrapped

A graph of GitHub commits by day of the week, with a peak on Wednesday and tails reaching Saturday/Sunday.

170

William Thompson @wrtastro.bsky.social · 12/12/2024

🧪🔭 Attending AAS in DC this January? Join us at our introduction to Julia programming for astronomy workshop! #julialang

Poster with the following text:

Julia programming for Astronomy
Two day workshop @ AAS 245 Winter Meeting

January 11th and 12th

- introduction to Julia
- data frames
- plotting
- modelling
- high performance
- multithreading
-create your first package!

Register through AAS portal!

Questions. Contact pebarrett@gwu.edu

1214

William Thompson @wrtastro.bsky.social · 03/12/2024

New version PairPlots.jl out now! We now calculate the effective sample size of your chains to determine the histogram binning and contour smoothing much more robustly. Before vs After:

A corner plot showing an MCMC chain - the contours look very blocky.

Same as before, but with larger histogram bins and smooth contours.

081

William Thompson @wrtastro.bsky.social · 13/11/2024

I got to meet real live astronaut Dr. Jenni Gibbons yesterday and give her a demonstration of our exo-🪐 imaging instrument SPIDERS! 🔭 #instrumentation

0191

William Thompson @wrtastro.bsky.social · 10/11/2024

Oh and by the way, here’s new a tutorial on how to compare 1 vs. 2 planet models by calculating the full Bayesian evidence, in just five minutes flat: sefffal.github.io/Octofitter.j...

072

William Thompson @wrtastro.bsky.social · 08/11/2024

Besides RV, nearly a dozen other types of data are supported (everything from Gaia to GRAVITY) PS: can anyone guess the exoplanet below? NIRSpec data coming this January!! 🔭🪐

050

William Thompson @wrtastro.bsky.social · 08/11/2024

New release of Octofitter, the ultra fast 🪐 orbit modelling code for Python and Julia. Version 5 with some nice speed improvements for RV modelling large datasets. Check it out here: - sefffal.github.io/Octofitter.j... - github.com/sefffal/octo... 🔭

3215

William Thompson @wrtastro.bsky.social · 06/02/2024

This speed also unlocks new capabilities: you can tractably answer chained questions like, how is an orbit likely to be constrained if we add more data at the currently predicted location? [6/7]

110

William Thompson @wrtastro.bsky.social · 06/02/2024

We bundle state of the art MCMC samplers: Hamiltonian Monte Carlo, and non-reversible parallel tempering. HMC is fast, while PT can explore crazy multimodal posteriors like this one. [6/7]

100

William Thompson @wrtastro.bsky.social · 06/02/2024

🐙 tries to have it all: it’s super fast, accurate, and guides users towards robust Bayesian workflows. Most RV or astrometry fits take *less than* 30s! For large RV datasets, we support linear-time approximations to Gaussian process fitting for stellar noise. [3/7]

100

William Thompson @wrtastro.bsky.social · 06/02/2024

As the sensitivities of radial velocity, direct imaging, and astrometric motion are starting to overlap, we need ways to find planets using combinations of data, and even stack images across epochs. [2/7]

120

William Thompson @wrtastro.bsky.social · 06/02/2024

Announcing Octofitter (🐙), an ambitious tool for all kinds of exoplanet orbit modeling workflows: fit planet orbits to relative astrometry, proper motion anomaly, radial velocity, images, and more! [1/7] sefffal.github.io/Octofitter.jl/ 🧪🔭🪐

2152

William Thompson @wrtastro.bsky.social · 30/11/2023

PairPlots.jl, the Julia library for making corner plots, now supports units and missing values! Table columns containing units are automatically pulled into the plot labels. Check out the docs at sefffal.github.io/PairPlots.jl...

030

William Thompson @wrtastro.bsky.social · 06/11/2023

The tourists are back again 🦭

020

William Thompson @wrtastro.bsky.social · 03/11/2023

In exoplanet direct imaging though, we’re not usually limited by photon noise. So it’s pretty much just a win! Each of our 640x512 pixels can record a spectrum from 0.85 to 1.65 micron at R of almost 20,000! 3/n

010

William Thompson @wrtastro.bsky.social · 03/11/2023

What’s an IFTS? It’s a type of imaging spectrograph that works through interferometry. While a bit finicky to get right, they can provide way higher spectral resolution and spaxel counts, at the cost of different (often but not always worse) photon noise. 2/n

120

William Thompson @wrtastro.bsky.social · 14/10/2023

We got up to 80% eclipse here in Victoria! Very odd how all the birds stop chirping and go quiet. It’s interesting how we’re always surrounded by images of the sun projected through trees etc but never notice it until it “changes”

0111

William Thompson @wrtastro.bsky.social · 11/10/2023

Well this was kind of a fun notification to receive from Google scholar!

010

William Thompson @wrtastro.bsky.social · 03/09/2023

Our hotel in Montreal was across from a three storey tall orange. I’ve been coming here my whole life and apparently missed that this is a local landmark.

A three storey tall orange serving food. Looks kind of like an old diner.

020