--- title: "Building Two-Body Orbits From Scratch" output: rmarkdown::html_vignette vignette: > %\VignetteIndexEntry{Building Two-Body Orbits From Scratch} %\VignetteEngine{knitr::rmarkdown} %\VignetteEncoding{UTF-8} --- ```{r, include = FALSE} knitr::opts_chunk$set( collapse = TRUE, comment = "#>", fig.width = 7, fig.height = 5 ) ``` ```{r setup, message = FALSE} library(orbitr) library(ggplot2) library(dplyr) ``` Most tutorials hand you a finished orbit and ask you to trust that the numbers are right. This guide goes the other way: it walks through the decisions that shape a two-body orbit so that by the end you can dial in any stable configuration you want from scratch. The three things you need to decide are: 1. Where each body starts (its **position**) 2. How fast the orbiting body is moving and in what direction (its **velocity**) 3. Which body is heavy enough to sit still at the center (its **mass**) Get any of these wrong and the orbit either crashes, escapes, or looks bizarrely eccentric. Get them right and the result is a clean closed loop that repeats forever. ## The Convention: Central Body at the Origin For any two-body problem, the easiest mental model is to put the heavier body at the origin with zero velocity and give the lighter body all the interesting initial conditions. This works as long as the central body is *much* more massive than the orbiter — good for Sun/planet and Earth/Moon, not so good for binary stars of comparable mass (more on that at the end). ```{r eval = FALSE} system <- create_system() |> add_body("Star", mass = 1e30) |> # heavy; sits at origin add_body("Planet", mass = 1e24, x = 1e11, vy = ?) # light; on the +x axis ``` That `?` in the velocity is the whole ballgame. Everything else follows from how much sideways motion we give the planet at that starting distance. ## Why Velocity Must Be Perpendicular to Position Gravity always pulls the planet *radially* — straight toward the star along the line connecting them. If the planet's velocity has a component pointing along that same radial line, the orbit won't be symmetric: the planet will either be falling inward or climbing outward at its starting position, not at the closest or farthest point of its path. For a clean circle or a symmetric ellipse with its major axis on a coordinate plane, you want the starting velocity to be **perpendicular** to the starting position vector. Put the planet on the +x axis (`x = 1e11, y = 0`) and give it a purely vertical velocity (`vx = 0, vy = v`). The starting position is then either the closest point (perihelion) or the farthest point (aphelion) of the ellipse, depending on whether `v` is above or below the circular speed. This isn't a physics requirement — any velocity produces a valid orbit — but it makes the geometry much easier to reason about. ## The Circular Orbit Velocity At distance $r$ from a central mass $M$, the speed that produces a perfect circle is: $$v_{\text{circ}} = \sqrt{\frac{G \, M}{r}}$$ In code: ```{r} M <- 1e30 # central mass (kg) r <- 1e11 # starting distance (m) v_circ <- sqrt(gravitational_constant * M / r) v_circ ``` Plug that into the planet's `vy` and you get a circle: ```{r circle-orbit} create_system() |> add_body("Star", mass = M) |> add_body("Planet", mass = 1e24, x = r, vy = v_circ) |> simulate_system(time_step = seconds_per_hour, duration = seconds_per_year) |> plot_orbits() ``` That's it. One line of math, one call to `add_body()`, and you have a stable circular orbit. ## What Happens If You Get the Speed Wrong? Real orbits are almost never perfect circles. The more interesting question is: what does the orbit look like when `vy` *isn't* exactly `v_circ`? Let's sweep through a few values and find out. ```{r velocity-sweep} make_sim <- function(v, label) { create_system() |> add_body("Star", mass = M) |> add_body("Planet", mass = 1e24, x = r, vy = v) |> simulate_system(time_step = seconds_per_hour, duration = seconds_per_year * 2) |> filter(id == "Planet") |> mutate(case = label) } bind_rows( make_sim(0.7 * v_circ, "0.7 v_circ (too slow)"), make_sim(1.0 * v_circ, "1.0 v_circ (circle)"), make_sim(1.2 * v_circ, "1.2 v_circ (ellipse)"), make_sim(1.4 * v_circ, "1.4 v_circ (wider ellipse)") ) |> ggplot(aes(x = x, y = y, color = case)) + geom_path(linewidth = 0.8) + geom_point(x = 0, y = 0, color = "gold", size = 4, inherit.aes = FALSE) + coord_equal() + theme_minimal() + labs(title = "Same starting point, different starting speeds", color = NULL) ``` A few things to notice: - **At exactly `v_circ`** you get a circle. - **Below `v_circ`**, the orbit is an ellipse whose *far* side is still at the starting distance `r`, and whose *near* side dips closer to the star. The starting point is aphelion (farthest point). - **Above `v_circ`** (but below escape velocity), the orbit is an ellipse whose *near* side is the starting distance, and the *far* side swings out much further. The starting point is perihelion (closest point). - **At or above $\sqrt{2} \cdot v_{\text{circ}}$**, the orbit is no longer bound — the planet escapes on a parabolic or hyperbolic trajectory and never comes back. We'll see this in the next section. The "too slow" case will actually crash into the star in this simulation because with `softening = 0` the acceleration diverges at close approach. That's one reason the `unstable-orbits` vignette recommends adding a softening length for anything that passes close to the center. ## Escape Velocity If you push `vy` hard enough, the planet's kinetic energy exceeds the gravitational binding energy and it leaves forever. The threshold is: $$v_{\text{esc}} = \sqrt{\frac{2 \, G \, M}{r}} = \sqrt{2} \cdot v_{\text{circ}}$$ ```{r escape-velocity} v_esc <- sqrt(2) * v_circ bind_rows( make_sim(0.99 * v_esc, "0.99 v_esc (very elongated ellipse)"), make_sim(1.00 * v_esc, "1.00 v_esc (parabolic escape)"), make_sim(1.10 * v_esc, "1.10 v_esc (hyperbolic escape)") ) |> ggplot(aes(x = x, y = y, color = case)) + geom_path(linewidth = 0.8) + geom_point(x = 0, y = 0, color = "gold", size = 4, inherit.aes = FALSE) + coord_equal() + theme_minimal() + labs(title = "At escape velocity and above, the orbit opens up", color = NULL) ``` Just below escape velocity you still have a (very stretched) closed ellipse. At and above it, the trajectory is an open curve and the planet never returns. ## Which Direction Does the Orbit Go? The sign of `vy` sets the direction of travel. Starting on the +x axis: - `vy > 0` → the planet moves in the +y direction initially → **counterclockwise** orbit (viewed from +z). - `vy < 0` → the planet moves in the $-$y direction initially → **clockwise** orbit. ```{r direction} bind_rows( make_sim( v_circ, "vy = +v_circ (counterclockwise)"), make_sim(- v_circ, "vy = -v_circ (clockwise)") ) |> ggplot(aes(x = x, y = y, color = case)) + geom_path(linewidth = 0.8) + geom_point(x = 0, y = 0, color = "gold", size = 4, inherit.aes = FALSE) + coord_equal() + theme_minimal() + labs(color = NULL) ``` Both traces the same circle — the only difference is the direction of travel, which only becomes visually obvious in an animation. ## Rotating the Whole Setup There's nothing magical about the +x axis. If you'd rather place the planet on the +y axis, you just swap: position goes in `y`, velocity goes in `vx` (with the opposite sign convention for direction). ```{r rotated} create_system() |> add_body("Star", mass = M) |> add_body("Planet", mass = 1e24, y = r, vx = -v_circ) |> simulate_system(time_step = seconds_per_hour, duration = seconds_per_year) |> plot_orbits() ``` Same circle, rotated 90 degrees. You can place the planet anywhere in the plane, as long as the velocity vector is perpendicular to the position vector and has magnitude `v_circ` (for a circle) or some deliberate deviation from it (for an ellipse). ## When the Central Body Isn't Much Heavier Everything above assumed the star is so heavy that it doesn't move appreciably. That's a fine approximation when the mass ratio is $\gtrsim 10^4$ (Sun/Earth is $\sim 3 \times 10^5$), but it breaks down for binary stars, Pluto-Charon, or any system where the two masses are comparable. For those, both bodies orbit their common center of mass (the **barycenter**), and you need to give *both* bodies initial velocities that satisfy conservation of momentum. The [Kepler-16 example](examples.html) shows how to do this for a realistic binary-star-plus-planet system. The short version: - Put the two bodies on opposite sides of the origin at distances inversely proportional to their masses. - Give each an initial velocity in opposite directions, again inversely proportional to mass. - The orbital speed of each body around the barycenter uses the *other* body's mass and the total separation. For anything resembling a planet around a much heavier star, the single-body approach in this guide is all you need. ## Recap To build a stable two-body orbit from scratch: 1. Pick a central mass $M$ and place it at the origin with zero velocity. 2. Pick a starting distance $r$ for the orbiter and place it on one axis (e.g. `x = r, y = 0, z = 0`). 3. Compute $v_{\text{circ}} = \sqrt{GM/r}$. 4. Give the orbiter a velocity perpendicular to its position vector (e.g. `vy = v_circ, vx = 0`). Use exactly `v_circ` for a circle, less for an inward ellipse, more for an outward ellipse, `sqrt(2) * v_circ` or above for an escape trajectory. That's the entire recipe. Every built-in example in `orbitr` — and every custom system you build — boils down to applying these four steps to one pair of bodies at a time.