How solmaps works

Deep Space Strategy · solmaps · Technical documentation · Computation status

How solmaps works

For the temporal-graph concept, every formula and the references, see the technical documentation.

solmaps is the open map and route engine of Deep Space Strategy. It computes what it costs to travel between places in the solar system on any date until 2100, and what that means for which places matter. This page lists the data, the models, the checks against real missions and the limits.

Status: experimental. Every model here simplifies the physics. The results are good enough to compare routes, dates and hubs, and to see which way an effect points. Mission design needs higher-fidelity tools such as NASA GMAT or ESA's own trajectory software.

Data

The resource atlas holds 37 entries: the planets, the major moons, Ceres, Vesta, Psyche, Pluto, three named near-Earth asteroids (Eros, Bennu, Ryugu), five asteroid populations and the Sun–Earth and Earth–Moon Lagrange points. It lists 98 resource entries, for water and other volatiles, metals, nitrogen, carbon and phosphorus, hydrogen and helium-3, silicates and sunlight.

The Data page lists every dataset with its address, format and a tested example, from the transfer data release to the daily network snapshots for apps and games.

Each resource entry has two labels. The evidence level says how well science knows it: confirmed (measured in place or in a returned sample), strong (several independent remote-sensing lines), inferred (spectral class or density analogy) or speculative. The verification tag says how solmaps checked it: page-checked (read on the source page), snippet-only (seen only in search results), derived (computed from checked inputs) or unverified. Of the 98 entries, 63 are page-checked. Sources include NASA's NSSDC fact sheets, JPL's Solar System Dynamics pages, mission pages and peer-reviewed papers; every entry links to its source.

Where things are

UseSourceRange and accuracy
Route costs (growth map, 3D routes)JPL DE440 ephemeris via ESA pykep (planet-system barycentres) to 2149, JPL Horizons (DE441) afterwardsAgrees with Horizons within 10 m; agrees with the JPL Table 1 positions to 0.00003 AU for Mars in 2040
Planet positions on the mapsJPL "Approximate Positions of the Planets", Table 1Valid 1800 to 2050; later dates on the maps are approximate and marked
Ceres, Vesta, Psyche, Eros, Bennu, Ryugu and 15 more asteroidsDaily JPL Horizons state vectors (perturbed by the planets)2026 to 2205; Bennu's solution ends in 2135, after its close Earth approach
MoonsJPL mean elements at J2000, period from the parent's massRough; solmaps spreads moon distances out for visibility

Direct transfers

A direct route is one transfer arc from one site to another. For every departure date, every ten days from November 2026 to the end of 2100, solmaps solves Lambert's problem (the arc that connects two positions in a given flight time) for 740 flight times up to 100 years. For each flight time it tries every number of revolutions around the Sun and keeps the cheapest; slow cargo ships can circle the Sun many times while they wait for the right geometry, and such long flights often cost far less than any short one. After a final refinement it keeps the cheapest transfer. solmaps shows 870 routes between 30 bodies over 2,709 departure dates, every tenth day; the growth map keeps the 132 routes between 12 sites. The engine computes every day: up to 18.8 billion Lambert problems. The computation status page lists what solmaps has computed so far.

The cost includes the burn that leaves a parking orbit and the burn that captures into the destination's parking orbit, using the Oberth relation Δv = √(v∞² + 2μ/r) − √(μ/r). Parking orbits: 300 km above Earth, Mars and Venus, 200 km above Mercury, 50 km above Ceres, and circular orbits at the distance of Callisto (Jupiter) and Titan (Saturn). At small asteroids a ship only matches their velocity. The maps use impulsive burns. The routing engine also computes low-thrust transfers for electric ships (see the engine documentation); the maps do not use them yet.

Flyby routes

Flyby routes use planets as slingshots. solmaps searches sequences with up to two flybys of Venus, Earth, Mars or Jupiter, from low Earth orbit and from low Mars orbit, to Jupiter, Saturn, Mercury, Ceres, Vesta and Psyche. Each leg may include one deep-space manoeuvre (pykep's MGA-1DSM model). A global optimiser (pygmo's self-adaptive differential evolution, eight islands, two independent starts per problem) finds the cheapest route in each two-year launch window. The search caps the flight time of each leg at 1.9 times the Hohmann time.

From low Earth orbit, the best slingshot route costs less than the direct route in every two-year window from 2027 to 2098 for Mercury, Saturn, Vesta, Ceres and Psyche, and in 34 of 36 windows for Jupiter. The median saving ranges from 1.0 km/s (Psyche) to 4.0 km/s (Mercury, with two Venus flybys); the technical documentation lists every target. From low Mars orbit, slingshots save less: the median saving ranges from 0.5 km/s (Psyche) to 1.9 km/s (Saturn), and for Ceres and Psyche the direct route costs less in about half of the windows.

Checks against reality

CheckReferencesolmaps
Planet positions from DE440 through pykepJPL Horizons state vectorsWithin 10 m
Round trips to near-Earth asteroids (NASA NHATS, seven published trajectories)NHATS speeds at departure, at the asteroid and at Earth entryWithin 0.001 km/s at the asteroid; within 0.012 km/s at Earth departure, because NHATS starts at Earth's centre and solmaps at the Earth–Moon barycentre
Transfers to ten asteroids, launches 2027–2035 (JPL Small-Body Mission-Design maps, 305,000 cells)JPL's excess speeds at departure and arrivalJPL's solution is among solmaps' solutions in 97.7 to 99.98 percent of the cells; all lie within 0.07 km/s
Leaving a 300 km Earth orbit for Mars3.59 km/s for the ideal Hohmann transfer between circular orbits in one plane3.61 km/s in the best window to 2100
Leaving a 300 km Earth orbit for Jupiter6.30 km/s for the ideal Hohmann transfer, computed the same way6.44 km/s in the best window to 2100
Earth to Mars transfer, 30 Oct 2026, 308 dayspykep's Lambert solver: v∞ 3.03 and 2.58 km/sSame values
Two independent network implementations, 2026–2048JavaScript on JPL Table 1 against Python on DE440Median difference 0.000 km/s, 95th percentile 0.02 km/s
Cassini (Earth–Venus–Venus–Earth–Jupiter–Saturn)Flown launch and flyby dates, 1997–1999Launch and inner flybys within 9 to 34 days
Juno, Galileo, DawnFlown sequences and datesSame sequences; inner encounters within one to three months

The ideal transfer assumes circular orbits in one plane. The real orbits are elliptical and inclined, so the best real window costs 0.6 percent more for Mars and 2.2 percent more for Jupiter. The optimiser only minimises Δv. Real missions also had to respect launch energy, arrival dates and, for Dawn, ion propulsion, so solmaps often picks a slower, cheaper arrival (Cassini at Saturn, Galileo at Jupiter).

Strategy measures

Propellant leverage

A hub with local water can refuel ships, so the propellant for the onward leg never leaves Earth. For each destination, solmaps compares the Earth-launched mass for a direct flight, exp(Δv_direct / vₑ), with the mass needed to reach the hub, exp(Δv_hub / vₑ). A destination counts only when the hub is cheaper to reach than the destination and lies closer to it than low Earth orbit does. The measure ignores dry mass, tankage, time at the hub and the cost of the plant. In the current results an Earth–Moon L1 depot fed by lunar ice saves 52 to 84 percent of Earth-launched mass over 2026–2100 (63 to 84 percent with direct routes only); the onward leg leaves the hub only after the ship has arrived there.

Depot orbits

solmaps scored 771 candidate solar orbits by two costs: the average cheapest transfer to eight destinations over 20 years, and the cost of supplying the depot from low Earth orbit. No orbit is best on both costs; the growth map lists the best orbit at each supply cost.

Cyclers

A cycler metric for Earth–Mars cyclers (maintenance burns at the flybys and the speeds a taxi must match) is in development; solmaps computes the cycler set for 2027–2097 (see the computation status).

solmaps

solmaps compresses distances from the Sun logarithmically and enlarges sizes, so Mercury and the Kuiper belt fit in one view. Directions and inclinations are true. It draws asteroid populations as clouds of illustrative particles inside their published ranges. It solves route arcs again between the displayed positions, so they meet the bodies on screen; their costs come from the DE440 network.

The atlas opens with the twelve sites and loads all 30 bodies of the transfer data a moment later (about 5 MB); the Bodies switch returns to the twelve. Trips lists every launch window from a place (or to it) within the waiting time, to all destinations or to one you pick, coloured from cheap (gold) to costly (red). "Launch a ship" puts the camera on board and follows the ship, past its flyby planets on slingshot routes. The map draws planets far larger than their true size, so during a ride a flyby planet shrinks to at most half its size and the ship passes just outside it. The network view places every body by its earliest arrival from a starting place; the whole-network view arranges all bodies in 3D by the travel time between each pair. The Display menu hides single bodies or whole groups from both views and shows, for each body, the share of launch dates from 2026 to 2100 with a direct route within a chosen budget. The table lists every trip option for filtering, and the trade-offs view shows all stored options of one route, with a 3D cost surface of waiting time, flight time and cost. Links open a route or a body directly, for instance solar-system-3d.html#route=earth-mars or #body=2 for Pallas.

Guided tour

On the first visit solmaps runs a short tour in six steps: the solar system as a network, one route from Earth to Mars, a ride past Mars on a slingshot, the network view, the live missions, and a choice of where to go next. The "?" button in the header starts it again.

Live missions

The Live missions view shows spacecraft in flight on today's date: Psyche, Lucy, Hera, Juice, Europa Clipper, BepiColombo, OSIRIS-APEX, Hayabusa2, Juno, New Horizons, the two Voyagers, Parker Solar Probe, Solar Orbiter, the two ESCAPADE spacecraft and MMX, with Tianwen-2 in the list although JPL Horizons holds no trajectory for it. Paths come from JPL Horizons, heliocentric and in the same frame as the bodies (flyby/fetch_missions.py); the page shows half a year behind and ahead of each probe, and the whole path, flown and planned, for the probe you select. Where Horizons ends a trajectory, for instance Psyche's in February 2029 before its arrival, the map ends it too. The facts on each card (launch, destinations, events, status) come from the agencies' pages, NASA's NSSDCA catalogue and Horizons, with a link for each; where two sources differ by a day, the card shows one and the registry keeps both (atlas3d/missions_registry.json, checked on 11 November 2026). A probe can be followed, ridden along, or compared with what a direct trip between the same bodies costs in solmaps' data from the date shown; the probes themselves fly with electric engines and flybys, which that comparison does not model. Links open a mission at a date, for instance #mission=lucy&date=2027-08-12.

Mission planner

The Missions tab plans a mission for one ship: a preset or your own dry mass, payload, tank, thrust and exhaust speed, a start, any number of stops with a stay, cargo to deliver or take on and refuelling, and a return. The rocket can bring the ship to low orbit, to escape, or supply the whole departure. The planner loads the legs it needs from the planner files on Hugging Face (every stored option of a route for every tenth departure day, as flight time and the two excess speeds) and prices both ends of every leg itself: an Oberth burn of √(v∞² + 2μ/r) − √(μ/r) into or out of low orbit, √(v∞² + 2μ/r) − √(2μ/r) for a loose orbit that just stays bound, braking in the atmosphere at Earth (up to 6 km/s) and Mars (up to 7 km/s) with a 0.1 km/s trim burn, and the excess speed itself at small bodies. The rocket equation carries the ship's mass through every burn. The search keeps, for every stop and arrival date, the most propellant the ship can have left; waiting at a stop costs nothing. The result is a mission sheet with every phase, a list of later returns that leave more propellant, the best mission for each launch year, a ride along in 3D and an export as CSV or JSON. Dates follow the 10-day grid of the planner files. This first version flies impulsive legs only; slingshot routes and electric ships follow, and arcs that circle the Sun more than once appear in the sheet but not in the ride.

Build it yourself

The computations run in a container with ESA pykep and pygmo from conda-forge, so nothing needs installing on the host apart from Docker, Python 3 and Node.

make image      # build the container
make ephemeris  # fetch the daily JPL Horizons states (about 200 MB)
make data       # transfer network and alternatives
make test       # test suite, including the NHATS check
make site       # pages into docs/

The README lists every step, the flyby search and the daily data for all 30 bodies. Further validation scripts: network/validate.js, flyby/validate_cassini.py, flyby/validate_missions.py and flyby/compare_networks.py.

Limits and next steps

Deep Space Strategy is a research programme by Björn Hanneke; solmaps is its open engine. Code under the MIT licence, data and documentation under CC BY 4.0; source at github.com/bhanneke/solmaps. It builds on JPL's ephemerides and Small-Body Database, ESA's pykep, pygmo and three.js.