Hybrasil Star and Exoplanet Map

Open the map at hybrasil.app

An interactive 3D map of nearby star systems. Every star, planet and moon is drawn on its published orbit and animated in the browser with three.js. Each star also gets a set of solvent bands, the ranges of orbital distance where water, ammonia, methane and five other candidate solvents could stay liquid.

• Seven featured systems, built from the literature: the Solar System, TRAPPIST-1, Alpha Centauri, Proxima Centauri, Upsilon Andromedae, 55 Cancri and Castor.
• About 420 planetary systems within 100 light years, and 68 systems of four or more stars with no known planet, from the Hybrasil Star and Exoplanet Dataset.
• Two views of stellar motion: every nearby star with its velocity, and the S-stars on their orbits round Sgr A*, the black hole at the centre of the Milky Way.

This repository, https://github.com/CadeSummers/hybrasil, is the official upstream, and https://hybrasil.app is the official site. Forks are welcome under the licences below; please do not present one as the original.

Goals

1. From raw data to simulation. Take the raw material astronomers publish, such as orbital solutions and parameter tables from papers, and telescope data like radial velocities, transit timings, astrometry and survey catalogs. Build each simulation directly from that data, with every value traceable to the publication or observation it came from.
2. A public resource for nearby stars. Publish the simulations as a website: an open, browsable collection of models for the star systems around us, for anyone who wants an astronomical model of a nearby star or a single place where the simulations are gathered.

Running it

The site needs nothing installed. To run it yourself, from a clone of this repository (Python 3.10 or later):

pip install -e .                    # or: pip install -r requirements.txt

python -m hybrasil.cli --serve      # the whole app, at http://127.0.0.1:8765/
python -m hybrasil.cli --system solar          # one system, opens in the browser
python -m hybrasil.cli --system trappist-1
python -m hybrasil.cli --source catalog --system "HD 18143"
python -m hybrasil.cli --list-solvents
python -m hybrasil.cli --system solar --renderer plotly  # the original Plotly viewer
python -m hybrasil.cli --help                  # every option

Optional extras: pip install -e ".[horizons]" for live elements from NASA JPL Horizons (--source horizons), ".[png]" for PNG export, and ".[dev]" for the test suite. Install from a clone rather than from a wheel: the catalog systems are read from the dataset snapshot in data/, which is not packaged.

What opening the simulation gives you

These are the baseline. Tests pin both of them (tests/test_featured_baseline.py, tests/test_catalog_loader.py::TestWholeSystems).

• Every featured system is always there. Opening the viewer in any way puts all seven built-in systems in the switcher along the top. --system chooses which one opens first. A system from another source (--source catalog, --source file, --source horizons) is added after the seven as an eighth button, and opens first. A catalog system that a featured system already draws opens as the featured one. For example, --source catalog --system "55 Cnc g" opens the featured 55 Cancri.
• A multi-star system is drawn whole. Every star in the system is drawn, and each planet orbits its own star. 55 Cancri shows both 55 Cnc A and 55 Cnc B, with B's two planets circling B. Alpha Centauri shows A, B and Proxima.
  • Built-in systems place each companion from its published orbit.
  • Catalog systems place it from the dataset's Separation columns on the multi-stars tab. A solved orbit is drawn as its ellipse. A projected separation, a lower bound on the real distance, is drawn as a circle at that distance and marked hypothetical. Every companion in the dataset has a sourced separation (23 September 2026 entry on Sources & Provenance).

The web app

python -m hybrasil.cli --serve

This opens the three.js viewer from a small local server at http://127.0.0.1:8765/. The bar along the top has two ways in:

• ★ Featured systems: one button for each built-in system, whose orbits come from the literature, and one for the Nearby Stellar Motion view.
• More Modeled Systems ▾: a dropdown that opens on a System Type: row: Planetary (the default), High Star Count (Planetless) (the systems of four or more stars within 100 ly that have no planet; the ones with a brown dwarf -- a planet in this model -- and Castor are under Planetary) and Stellar Motion (Nearby Stellar Motion, and the Sgr A* Cluster: the S-stars on their orbits round the Milky Way's supermassive black hole). The search box searches the chosen type. Planetary lists every planetary system in the dataset, about 420 of them. It lists the systems with the most planets first, and nearest first among those with the same count. The list is twelve systems tall and scrolls for the rest; click one to draw it. Candidate planets count, so the number is how many planets the viewer draws. The search box at the top of the dropdown finds any system by name, and tolerates typos ("trapist" finds TRAPPIST-1). It also matches catalog-style names typed the long way ("tau ceti" finds tau Cet) and nicknames ("Copernicus"). A planet's or a companion star's name finds the system it belongs to.

The hosted site is the same app written out as static files: python -m hybrasil.cli --build-site site pre-builds every system's scene into site/ (about two minutes, about 420 MB), and any static host can serve it.

What is stored and what is built on lookup

The star and planet data is stored. Each system's 3-D scene is built the first time you open it.

• The data:
  • The seven featured systems are written into the code (hybrasil/data/builtin_systems.py), with values taken from published papers.
  • Every other system comes from the dataset snapshot committed in data/. The dataset is maintained in a private spreadsheet, and each new snapshot of it is committed here, so every version of the site can be rebuilt from this repository alone.
• The scenes:
  • When the page loads, the server sends only the list of systems (/api/systems). Nothing is drawn yet.
  • When you pick a system, the server works out its orbits, animation frames and solvent bands, and converts them into a three.js scene. It then sends the scene to the page (/api/systems/<id>).
  • Each built scene is kept in memory until the server stops. A second visit to a system is instant; the first visit to a large one takes a moment.
• On disk: nothing is written, apart from any Suggest Changes submissions, which --serve appends to a local suggestions.jsonl. --output html is the separate way to save a system as a file, and --build-site the way to save them all.

The viewers

The default viewer is three.js. Its logarithmic depth buffer lets one camera cover everything from a moon's orbit to the whole system, so zooming is smooth at every scale. --renderer plotly opens the original Plotly viewer. The two share their band sliders, search and info panel from one file (hybrasil/viz/band_kit.js), so they can't drift apart. PNG export (--output png) always uses Plotly.

tests/test_browser.py drives both viewers in headless Chromium through Playwright. After pip install -r requirements-dev.txt, run python -m playwright install chromium once. The browser tests take about three minutes; pytest -m "not browser" skips them.

In the viewer, the system buttons along the top switch between systems. Behind every system is the Milky Way, NASA's Deep Star Maps 2020 (from Gaia DR2), dimmed so the orbits stay readable. It is turned into each system's own frame, so the galactic-centre arrow points at the galactic centre in the picture. The Sky button turns it off, and the viewer remembers that choice. Only the three.js viewer draws the sky. The Bands button shows the solvent bands, and the panel in the top-left moves them: pressure, albedo and the greenhouse effect. python -m hybrasil.cli --help lists every option. CLAUDE.md has the full command list and the architecture, and docs/architecture/ a deep dive on each part.

How the solvent bands are computed

A band is the range of orbits where a solvent's surface temperature sits between its melting and boiling points. It is a first-order screen, not a habitable zone. The difference is explained below.

1. Blackbody temperature

A planet at d AU from a star of luminosity L (in suns), with Bond albedo A, has an equilibrium temperature of

T_eq = 278.33 K · L^¼ · (1 − A)^¼ / √d

278.33 K is a perfect absorber at 1 AU from the Sun. Turned around for distance, each edge of a band is

d = 278.33² · √L · √(1 − A) / T²

with T the boiling point for the inner (hot) edge and the melting point for the outer (cold) edge.

2. Pressure and albedo

Melting and boiling points come from the dataset, transcribed at seven pressures from 0.1 to 100 bar. Between those columns the boiling point is interpolated with Clausius–Clapeyron (1/T linear in ln P), and the melting point linearly in ln P. Outside a solvent's own transcribed range there is no band at all, because extrapolating a vapour-pressure curve is how you draw a band where nothing is liquid. Albedo defaults to Earth's 0.3. Both are sliders.

3. Greenhouse effect (beta)

A flat offset G, in kelvin, added to the blackbody temperature. The default of 33 K is Earth's: a 288 K mean surface over a 255 K blackbody. Each edge becomes

d = 278.33² · √L · √(1 − A) / (T − G)²

It is marked beta because a real greenhouse depends on the atmosphere and the flux it traps, not on a constant. Past a solvent's boiling point there is no band anywhere. Past its melting point the band runs out to where the star's warmth falls to the 2.725 K cosmic microwave background.

4. Star colour

Two stars of equal luminosity do not warm a planet equally. A red dwarf's light is mostly near-infrared: air scatters less of it and water vapour and CO₂ absorb more of it, so the planet reflects less and the same flux keeps it warmer. Each star therefore gets a factor f on the flux its planets absorb:

f(Teff) = S_eff(5772 K) / S_eff(Teff)
S_eff(Teff) = 1.0512 + 1.3242e-4·t + 1.5418e-8·t² − 7.9895e-12·t³ − 1.8328e-15·t⁴
              t = Teff − 5780 K

d = 278.33² · √L · √(f · (1 − A)) / (T − G)²

• Source: S_eff is the runaway-greenhouse limit from Kopparapu et al. 2013 (ApJ 765, 131, Table 3). It uses a thin, water-vapour-rich atmosphere, the closest of their climate limits to the fixed Earth-like atmosphere the bands assume.
• The Sun: the factor is normalised at the Sun's 5772 K, so the Sun's f is exactly 1 and the Solar System does not move.
• Examples: Proxima Centauri (2980 K) gets 1.215, which moves every band out by 1.10×. Upsilon Andromedae (6212 K) gets 0.946.
• Temperature range: the fit covers 2600–7200 K. A star outside that range is taken at the nearer end rather than extrapolated. TRAPPIST-1, at 2516 K, is taken at 2600 K.
• Missing temperature: a star with no known temperature gets no correction. Its description says so.
• Every solvent moves together: the factor scales every solvent's band around a star by the same amount, and never decides whether a band exists. The caveat is that the fit was made for water-vapour atmospheres. Applying it to a methane or sulfur band assumes the same scattering and absorption behaviour.
• Switching it off: --star-colour off removes the correction.

Bands versus the habitable zone

A habitable zone (HZ) is where climate feedbacks can keep water liquid, not where a fixed atmosphere happens to be the right temperature. Its inner edge is the runaway greenhouse: water vapour traps heat until the oceans are lost, long before the surface reaches 100 °C. Its outer edge assumes a cooling planet builds up a thick CO₂ atmosphere. The water band has neither, so it sits closer to its star than the HZ. Around the Sun the band is 0.56–1.12 AU and the conservative HZ is about 0.98–1.71 AU.

Proxima b is the worked example. It receives 0.64× Earth's flux, inside Proxima's HZ (0.042–0.081 AU). The water band, at 1 bar, albedo 0.3 and Earth's 33 K greenhouse:

	Water band	Proxima b (0.0486 AU)
Blackbody only	0.0218–0.0437 AU	outside
With star colour	0.0241–0.0481 AU	outside by 0.0004 AU
With star colour and a 34.05 K greenhouse	reaches 0.0486 AU	at the edge; inside above that

Star colour closes most of the gap, but not all of it. The rest is the feedbacks above, which a single greenhouse number cannot represent.

Where the numbers come from

• Stellar luminosities and temperatures: the dataset, whose Sources & Provenance tab records a source for each. The star-colour model is written up there as well, in the entry dated 22 September 2026.
• Proxima Centauri: L = 0.00151 L☉ and Teff = 2980 K, both from Ribas et al. 2017 (A&A 603, A58).
• Rules for changing band code: docs/architecture/solvent-bands.md (the physics, and the rules for changing it) and docs/architecture/viewer-bands.md (how the viewer recomputes bands as the sliders move).

Contributing

Corrections to the data are the most useful thing anyone can send, and they need no account and no code: open the system on hybrasil.app and press Suggest Changes. Name the value, what it should be, and the paper or catalogue it comes from. Every suggestion is checked by hand.

Issues (Data correction, New system, Bug) and pull requests are welcome too. CONTRIBUTING.md explains how a data change reaches the site, what makes a source easy to accept, and the code rules; commits need a DCO sign-off (git commit -s). Everyone taking part agrees to the Code of Conduct. Security problems go privately, as SECURITY.md describes.

Citing

If you use the map or its code, please cite the software; if you use the data, please cite the dataset. GitHub's Cite this repository button gives both, from CITATION.cff.

Summers, C. (2026). Hybrasil Star and Exoplanet Map [Computer software]. https://hybrasil.app

Summers, C. (2026). Hybrasil Star and Exoplanet Dataset [Data set]. https://hybrasil.app

Licence

• Code: GNU Affero General Public License v3.0 or later (LICENSE). If you run a modified version as a network service, you must offer its users the modified source.
• Data: the Hybrasil Star and Exoplanet Dataset, the snapshots and the research data in data/, are licensed under Creative Commons Attribution-ShareAlike 4.0 (LICENSE-DATA). Credit it as NOTICE shows. The scripts in data/ are code, under AGPL; every source file says which licence it is under in its SPDX header.
• Third-party material keeps its own terms: the Milky Way background from NASA's Deep Star Maps, Gaia data under CC BY-SA 3.0 IGO, and the catalogues, archives and libraries listed in THIRD_PARTY_NOTICES.md.

Copyright (c) 2026 Cade Summers.