The Effective Potential

Comet C/2023 A3 fell toward the Sun for tens of thousands of years, accelerating the whole way, and by perihelion was moving at over 240,000 km/h — yet it never came close to hitting the star pulling it in. It swung around at 58 million kilometres and climbed back out. Nothing pushed it away; gravity attracted it at every instant. What saved the comet was not a force but a bookkeeping identity: as long as it has any angular momentum at all, the energy cost of approaching the center grows like $1/r^2$ and eventually outbids the $1/r$ gravitational reward. There is a closest approach the comet cannot cross, and you can compute it without solving a single differential equation. The tool that makes this one-line reasoning possible is the effective potential, $V_{\mathrm{eff}}(r) = V(r) + L^2/2\mu r^2$ . It takes the genuinely two-dimensional dance of an orbit — radius changing, angle sweeping, both coupled — and compresses it into the motion of a fictitious bead sliding along a one-dimensional wire. The wire's shape is $V_{\mathrm{eff}}$ ; the bead's height ceiling is the conserved energy $E$ . Where the bead stops, the real orbit turns around. Where the wire has a valley, a circular orbit lives. Whether the valley is a true minimum decides whether that orbit survives a nudge. This single curve is the qualitative theory of orbits in its entirety: turning points, circular radii, stability, radial oscillation frequencies — all read off a sketch. It is the working diagram of mission designers, the reason planets do not fall into the Sun, and, suitably modified by general relativity, the reason things do fall into black holes.

Key takeaways

Conservation of angular momentum ( $L = \mu r^2\dot\phi = \mathrm{const}$ ) lets the angular kinetic energy $\tfrac12\mu r^2\dot\phi^2$ be rewritten as a pure function of radius, $L^2/2\mu r^2$ .
Folding that term into the true potential defines the effective potential $V_{\mathrm{eff}}(r) = V(r) + L^2/2\mu r^2$ : the radial motion becomes exact one-dimensional motion, $E = \tfrac12\mu\dot r^2 + V_{\mathrm{eff}}(r)$ .
The centrifugal term $L^2/2\mu r^2$ is repulsive bookkeeping, not a real force — it diverges as $r\to0$ for any $L\neq0$ , walling off the origin unless the true attraction is steeper than $1/r^2$ in the potential.
Extrema of $V_{\mathrm{eff}}$ are circular orbits: minima are stable (small nudges oscillate at $\omega_r = \sqrt{V_{\mathrm{eff}}''(r_0)/\mu}$ ), maxima are unstable.
For a power-law force $f = -kr^n$ , stable circular orbits exist exactly when $n > -3$ : gravity ( $n=-2$ ) passes comfortably; inverse-cube attraction is marginal.
Turning points of the real orbit — perihelion and aphelion — are simply the roots of $E = V_{\mathrm{eff}}(r)$ .
Key idea: conservation of $L$ buries the angular motion inside a one-dimensional potential $V_{\mathrm{eff}}(r) = V(r) + L^2/2\mu r^2$ ; the entire qualitative theory of orbits is then read off the shape of this single curve as if it described a particle on a line.

The core insight

Conservation of $L$ lets you bury the angular motion inside a one-dimensional potential $V_{\mathrm{eff}}(r)=V(r)+L^2/2\mu r^2$ ; the entire qualitative theory of orbits is then read off the shape of this single curve as if it described a particle on a line.

Where this lives today

The most consequential effective-potential diagram in contemporary physics belongs to general relativity. The Schwarzschild geometry modifies the Newtonian curve by one extra attractive term, $-GML^2/c^2r^3$ , and that single change redraws the fate of matter: the centrifugal barrier no longer rises forever but crests and turns over, so the well bottom (stable circles) and the hilltop (unstable circles) approach each other as $L$ decreases and merge at the innermost stable circular orbit, $r_{\mathrm{ISCO}} = 6GM/c^2$ . Inside it, no amount of angular momentum can hold an orbit up. That radius sets the inner edge of black-hole accretion disks — hence the X-ray spectra used to measure black-hole spin — and it is where binary black holes transition from inspiral to plunge in the waveforms LIGO and Virgo record. The "effective-one-body" models used to generate those waveform templates are, quite literally, an industrial-strength refinement of $V_{\mathrm{eff}}(r)$ . The same construction runs quantum mechanics: the radial Schrödinger equation for hydrogen contains the centrifugal term $\hbar^2\ell(\ell+1)/2m_er^2$ added to the Coulomb potential — the identical barrier with $L^2$ quantized — and it is why only $s$ -states ( $\ell=0$ ) have electron density at the nucleus, which in turn is why electron capture and hyperfine structure care about orbital angular momentum. From planetary stability through Rutherford scattering's distance of closest approach to the shape of every atom: one curve, drawn once per problem, before any equation is solved.

The physical picture

Take the orbit of a planet — an ellipse being traced in a plane — and stand a piece of graph paper next to it. On the paper, plot only the distance $r$ between planet and Sun against the potential-energy landscape it experiences. The two-dimensional orbit collapses onto a bead rolling in a one-dimensional valley. The bead speeds up rolling downhill (the planet falls inward, gaining radial speed), slows climbing uphill, and turns around exactly where its energy line meets the valley wall. Those two wall-touches are perihelion and aphelion. The angular motion has not disappeared — the planet is silently sweeping around the Sun the whole time — but every bit of its angular kinetic energy has been repackaged as a function of $r$ and drawn into the landscape itself.

Figure

The effective potential well for an attractive inverse-square force — one curve carries the whole qualitative theory of orbits

Three features of the curve do all the work. First, the inner wall: for any nonzero $L$ the centrifugal term $L^2/2\mu r^2$ grows faster as $r\to 0$ than the gravitational $-k/r$ falls, so $V_{\mathrm{eff}}\to+\infty$ at the origin. Every energy line hits that wall at a finite radius — an innermost turning point. This is the precise, quantitative reason an orbiting body with angular momentum cannot reach the center no matter how much energy it carries. Physically: shrinking $r$ at fixed $L$ forces $\dot\phi = L/\mu r^2$ to skyrocket, and the angular kinetic energy bill $L^2/2\mu r^2$ must be paid out of radial motion, which stalls. Second, the valley: where the attractive tail and the repulsive barrier balance, $V_{\mathrm{eff}}$ has a minimum. A bead resting at the bottom is a body whose radius never changes — a circular orbit. The valley's depth is the most negative energy any orbit of that angular momentum can have; its curvature sets how fast a slightly-nudged orbit breathes in and out around the circle. Third, the water line: draw the conserved energy $E$ as a horizontal line. The bead lives where the curve is below the line, stalls where they cross, and is forbidden above. Two crossings mean a bound orbit rocking between perihelion and aphelion; one crossing means an unbound visit with a single closest approach; a line tangent at the bottom is the circle. That reading — energies against a curve — is developed fully in Classification of orbits by energy; this article builds the curve itself and extracts what its shape says: where circular orbits sit and whether they survive perturbation.

Building the effective potential

Everything starts from the two conservation laws established earlier in the central-force chain. After the two-body reduction, the relative motion is a single fictitious particle of reduced mass $\mu$ moving in the plane with polar coordinates $(r,\phi)$ , and its conserved energy has three parts — radial kinetic, angular kinetic, and potential:

E = \tfrac12\mu\dot r^2 + \tfrac12\mu r^2\dot\phi^2 + V(r) \qquad (1)

where: $\mu = m_1 m_2/(m_1+m_2)$ — the reduced mass; $r$ — the separation of the two bodies; $V(r)$ — the true central potential. Nothing has been simplified yet: this is just kinetic energy written in polar coordinates. The middle term is the one we are about to eliminate. The elimination is bought entirely by angular momentum conservation. For any central force the angular momentum of the relative motion is a constant of the motion:

L = \mu r^2\dot\phi = \mathrm{const} \qquad (2)

where: $L$ — the magnitude of the conserved angular momentum, fixed once and for all by the initial conditions. This is the lever: it ties $\dot\phi$ rigidly to $r$ , so the angular velocity is not an independent variable at all — a body at half the radius must sweep four times as fast.

Derivation AFolding the angular motion into a one-dimensional potential

We are deriving the reduced energy equation (3) and with it the definition of the effective potential (4), starting from the energy (1) and angular momentum conservation (2). The strategy is pure substitution: solve (2) for $\dot\phi$ , insert it into the angular kinetic term, and watch that term turn into a function of $r$ alone — at which point it is indistinguishable, mathematically, from a potential energy, and can be absorbed into one. The point to watch is conceptual rather than algebraic: the new term is still kinetic energy. Absorbing it into the potential is a choice of bookkeeping, and forgetting that leads directly to the double-counting trap flagged later.

Step 1. Solve the angular momentum relation (2) for the angular velocity: $\dot\phi = L/\mu r^2$ . Why this is progress: the two-dimensional problem had two velocities, $\dot r$ and $\dot\phi$ ; conservation of $L$ has just expressed one of them entirely in terms of the coordinate $r$ .

Step 2. Substitute into the angular kinetic energy:

\tfrac12\mu r^2\dot\phi^2 = \tfrac12\mu r^2\left(\frac{L}{\mu r^2}\right)^{2} = \frac{L^2}{2\mu r^2}

Why the $r$ -dependence flips sign of behavior: kinetic energy that used to grow with $\dot\phi$ now grows as $r$ shrinks — at fixed $L$ , closer means faster means energetically costlier. The repulsive character of the barrier is conservation of angular momentum wearing a potential-energy costume.

Step 3. Insert this into (1) and group everything that depends only on $r$ :

E = \tfrac12\mu\dot r^2 + \underbrace{V(r) + \frac{L^2}{2\mu r^2}}_{V_{\mathrm{eff}}(r)} \qquad (3)

where: the braced combination defines the effective potential. Equation (3) has exactly the form of a one-dimensional energy law for a particle of mass $\mu$ with position $r$ : the two-dimensional orbit problem has become a bead-on-a-wire problem.

Step 4. Read off the definition and its force:

V_{\mathrm{eff}}(r) = V(r) + \frac{L^2}{2\mu r^2}, \qquad \mu\ddot r = -\frac{\mathrm{d}V_{\mathrm{eff}}}{\mathrm{d}r} = f(r) + \frac{L^2}{\mu r^3} \qquad (4)

Why the radial equation follows: differentiating (3) with respect to time and cancelling $\dot r$ gives $\mu\ddot r = -V_{\mathrm{eff}}'(r)$ — the fictitious bead obeys Newton's law in the effective landscape. The outward piece $L^2/\mu r^3$ is the "centrifugal force"; its potential parent is the barrier term. *Verified: differentiating (3) reproduces the radial equation of motion derived independently in Reduction to the Radial Equation; matches Goldstein §3.3 and Taylor ch. 8.*

The one-dimensional analogy is exact, not approximate — equation (3) is an identity for any central potential and any orbit. The radial speed at any radius follows immediately by solving for $\dot r$ :

\dot r = \pm\sqrt{\frac{2}{\mu}\Big[E - V_{\mathrm{eff}}(r)\Big]} \qquad (5)

where: the $\pm$ tracks whether the body is currently moving outward or inward, flipping at each turning point. The square root is real only where $E \ge V_{\mathrm{eff}}$ — the algebraic form of "the bead stays below the water line" — and the turning points $\dot r = 0$ are exactly the roots of $E = V_{\mathrm{eff}}(r)$ .

Circular orbits: the extrema of the curve

A circular orbit is a solution with $r$ constant, so both $\dot r$ and $\ddot r$ vanish. By the radial equation in (4) that requires the slope of the effective potential to vanish — the bead must sit at a flat point of the wire:

\left.\frac{\mathrm{d}V_{\mathrm{eff}}}{\mathrm{d}r}\right|_{r_0} = 0 \quad\Longleftrightarrow\quad V'(r_0) = \frac{L^2}{\mu r_0^3} \qquad (6)

where: $r_0$ — the circular-orbit radius. The right-hand form says something familiar in disguise: $V'(r_0)$ is the inward true force magnitude, and $L^2/\mu r_0^3 = \mu r_0\dot\phi^2$ is exactly the centripetal requirement $\mu v^2/r$ . The extremum condition of the abstract curve is the freshman "gravity supplies the centripetal force" balance — now derived rather than assumed, and valid for any central force law. Whether that orbit persists is a question about the second derivative, and it splits cleanly:

V_{\mathrm{eff}}''(r_0) > 0 \;\Rightarrow\; \text{stable}, \qquad V_{\mathrm{eff}}''(r_0) < 0 \;\Rightarrow\; \text{unstable} \qquad (7)

where: stability means a slightly disturbed orbit stays near circular, oscillating in radius; instability means the disturbance grows exponentially and the body either spirals away or plunges. A valley bottom traps; a hilltop expels.

Go deeper · Derivation B · 1 minStability of circular orbits and the radial oscillation frequency

Derivation BStability of circular orbits and the radial oscillation frequency

The goal is the stability criterion (7) together with a bonus the criterion alone does not give: the frequency $\omega_r$ at which a stable near-circular orbit breathes. The strategy is the universal one for motion near equilibrium — write $r(t) = r_0 + \eta(t)$ with $|\eta| \ll r_0$ , Taylor-expand the effective force to first order in $\eta$ , and recognize the resulting equation as either a simple harmonic oscillator or an exponential runaway depending on one sign. This is the same linearization that governs every equilibrium in physics; the only central-force-specific input is the shape of $V_{\mathrm{eff}}$ .

Step 1. Start from the exact radial equation from (4), $\mu\ddot r = -V_{\mathrm{eff}}'(r)$ , and substitute $r = r_0 + \eta$ : $\mu\ddot\eta = -V_{\mathrm{eff}}'(r_0+\eta)$ . Why $\ddot r = \ddot\eta$ : $r_0$ is a constant, so all the time dependence lives in the perturbation.

Step 2. Taylor-expand the slope about $r_0$ :

V_{\mathrm{eff}}'(r_0+\eta) = \underbrace{V_{\mathrm{eff}}'(r_0)}_{=\,0} + V_{\mathrm{eff}}''(r_0)\,\eta + \mathcal{O}(\eta^2)

Why the first term dies: $r_0$ was chosen as an extremum — the circular-orbit condition (6) is precisely the statement $V_{\mathrm{eff}}'(r_0)=0$ . The leading survivor is the curvature term.

Step 3. Keep the linear term and obtain the perturbation equation:

\mu\ddot\eta = -V_{\mathrm{eff}}''(r_0)\,\eta

where: everything now hinges on the sign of $V_{\mathrm{eff}}''(r_0)$ . Positive curvature makes this a restoring force; negative curvature makes it anti-restoring.

Step 4. Read off the two behaviors. For $V_{\mathrm{eff}}''(r_0) > 0$ the equation is simple harmonic motion with

\omega_r = \sqrt{\frac{V_{\mathrm{eff}}''(r_0)}{\mu}} \qquad (8)

so the radius oscillates: the orbit is a slowly breathing near-circle, stable. For $V_{\mathrm{eff}}''(r_0) < 0$ the solutions are $e^{\pm|\omega_r| t}$ and the perturbation grows without bound: unstable. Why this frequency matters beyond stability: comparing $\omega_r$ with the angular sweep rate $\omega_\phi = L/\mu r_0^2$ decides whether the perturbed orbit closes — equal frequencies give a closed, non-precessing ellipse; unequal ones give a precessing rosette. For Kepler, $\omega_r = \omega_\phi$ exactly. *Verified: reduces to the standard SHM result; matches Goldstein §3.6 and Taylor §8.8; for the Kepler potential the formula gives $\omega_r = \mu k^2/L^3$ , equal to the orbital frequency, consistent with closed ellipses.*

Which force laws allow stable circles at all

The stability test becomes strikingly clean for a whole family of forces at once. Suppose the attraction is a power law, $f(r) = -kr^n$ with $k>0$ — this covers gravity ( $n=-2$ ), the isotropic spring ( $n=+1$ ), and every hypothetical in between and beyond. One short calculation answers the stability question for the entire family.

Go deeper · Derivation C · 1 minStability condition for a power-law force

Derivation CStability condition for a power-law force

We are deriving the celebrated criterion that circular orbits in an attractive power-law force $f(r) = -kr^n$ are stable if and only if $n > -3$ . The strategy: compute $V_{\mathrm{eff}}''$ in general, then use the circular-orbit condition (6) to eliminate the angular momentum $L$ in favour of the force constant — the trick that turns a two-parameter question (which $L$ ? which $r_0$ ?) into a statement about the exponent alone. Watch how completely $L$ and $r_0$ drop out of the final sign: stability is a property of the force law, not of the particular orbit.

Step 1. Differentiate $V_{\mathrm{eff}} = V(r) + L^2/2\mu r^2$ twice:

V_{\mathrm{eff}}''(r) = V''(r) + \frac{3L^2}{\mu r^4}

Why the barrier contributes $+3L^2/\mu r^4$ : each derivative of $r^{-2}$ pulls down one negative power and one sign; two of them give $(-2)(-3)r^{-4}/2\mu \cdot L^2$ .

Step 2. Express the potential derivatives through the force. From $f = -V'$ and $f(r) = -kr^n$ : $V'(r) = kr^n$ and $V''(r) = nkr^{n-1}$ .

Step 3. Use the circular-orbit condition (6) to eliminate $L$ : $L^2/\mu r_0^3 = V'(r_0) = kr_0^n$ , hence

\frac{3L^2}{\mu r_0^4} = \frac{3}{r_0}\cdot\frac{L^2}{\mu r_0^3} = 3kr_0^{\,n-1}

Why eliminate $L$ : the curvature at $r_0$ then depends only on the force law and the radius, and the radius will factor out.

Step 4. Assemble the curvature at the circular radius:

V_{\mathrm{eff}}''(r_0) = nkr_0^{\,n-1} + 3kr_0^{\,n-1} = (n+3)\,k\,r_0^{\,n-1}

where: $k > 0$ and $r_0^{\,n-1} > 0$ always, so the sign of the curvature — and with it stability — is the sign of $(n+3)$ .

Step 5. Conclude: stable circular orbits exist exactly for $n > -3$ . Gravity ( $n = -2$ ) gives curvature $\propto +1$ : stable. The inverse-cube force ( $n = -3$ ) is marginal — the effective potential's leading terms both scale as $1/r^2$ and the curvature vanishes identically. Anything steeper has only unstable circular orbits: the slightest inward nudge wins the tug-of-war for good and the body spirals into the center. Why this connects to Bertrand's theorem: stability of circles is necessary but not sufficient for closed orbits; demanding closure for all bound orbits narrows $n > -3$ down to just $n = -2$ and $n = +1$ . *Verified: recovers the classical criterion; matches Landau and Lifshitz §14 and Goldstein §3.6; the marginal case $n=-3$ agrees with the exact solvability of the inverse-cube orbit (Cotes spirals).*

Figure

Stable valley versus unstable hilltop — the curvature of the effective potential at the circular radius decides the orbit's fate

Which equation answers which question

you want	use	notes
the effective landscape	$V_{\mathrm{eff}} = V(r) + L^2/2\mu r^2$ , eq. (4)	one curve per value of $L$
radial speed anywhere	$\dot r = \pm\sqrt{\tfrac{2}{\mu}(E - V_{\mathrm{eff}})}$ , eq. (5)	turning points at $E = V_{\mathrm{eff}}$
circular-orbit radius	$V_{\mathrm{eff}}'(r_0) = 0$ , eq. (6)	equals the centripetal balance
is the circle stable	sign of $V_{\mathrm{eff}}''(r_0)$ , eq. (7)	valley yes, hilltop no
breathing frequency	$\omega_r = \sqrt{V_{\mathrm{eff}}''(r_0)/\mu}$ , eq. (8)	compare to $\omega_\phi$ for closure
power-law verdict	stable iff $n > -3$	gravity safe; inverse-cube marginal

The Kepler well, quantitatively

For gravity, $V(r) = -k/r$ with $k = Gm_1m_2$ , every feature of the well has a clean closed form, and it is worth committing two of them to memory. Setting $V_{\mathrm{eff}}'= 0$ with $V' = k/r^2$ gives $k/r_0^2 = L^2/\mu r_0^3$ , so the well bottom sits at

r_0 = \frac{L^2}{\mu k} \qquad (9)

where: $r_0$ — the unique circular-orbit radius for angular momentum $L$ ; larger $L$ pushes the circle outward quadratically. This same combination reappears throughout the chain as the semi-latus rectum of the conic. Substituting back gives the depth:

V_{\mathrm{eff}}(r_0) = -\frac{k}{r_0} + \frac{L^2}{2\mu r_0^2} = -\frac{\mu k^2}{2L^2} \qquad (10)

where: the depth is half the value of $-k/r_0$ — the barrier refunds exactly half the gravitational debt, a $1/r$ -specific coincidence that is the virial theorem in miniature. No orbit with angular momentum $L$ can have energy below this; the energy ladder upward from the well bottom — ellipses, the parabola, hyperbolas — is climbed in Classification of orbits by energy, and the shapes themselves are proven conics via Binet's orbit equation and the Kepler problem.

Limiting cases

Case	Result	Meaning
Kepler attraction $V = -k/r$	single minimum at $r_0 = L^2/\mu k$ , depth $-\mu k^2/2L^2$	one stable circle per $L$ ; bound orbits for $E<0$ ; barrier always wins at small $r$
isotropic spring $V = \tfrac12\mu\omega^2 r^2$	well rises at both ends	every $L\neq0$ orbit is bound; closed centered ellipses (Bertrand's other survivor)
steep attraction, $V = -c/r^{m}$ , $m \ge 2$	$V_{\mathrm{eff}} \to -\infty$ as $r\to0$	attraction beats the barrier; capture and spiral-in become possible; circles unstable or absent
$L = 0$	$V_{\mathrm{eff}} = V(r)$ exactly	barrier gone; purely radial plunge through the center line
$r \to \infty$	$V_{\mathrm{eff}} \to V(r) \to 0^-$ (Kepler)	barrier is irrelevant far out; escape decided by the sign of $E$
$L$ very large	$r_0 = L^2/\mu k$ grows, well depth $\propto 1/L^2$ shrinks	fast-spinning orbits sit far out in an ever-shallower well — barely bound

Common trapThe centrifugal barrier is a real force acting on the orbiting body

It is not — and the misreading causes two distinct errors. The term $L^2/2\mu r^2$ is the angular kinetic energy rewritten as a function of $r$ using $L = \mu r^2\dot\phi$ ; no new interaction exists in the problem, and in the inertial frame the only force on the body is the true attraction $f(r)$ , always pointing inward. Error one is double-counting: writing the energy as $\tfrac12\mu\dot r^2 + \tfrac12\mu r^2\dot\phi^2 + V_{\mathrm{eff}}$ counts the angular energy twice — use the angular term or the barrier, never both. Error two is misattributing the outward tendency: a comet climbing away from perihelion is not being pushed by a centrifugal force; it is coasting on angular speed that conservation of $L$ will not let it shed fast enough. Also mind the fine print: the barrier exists only for $L \neq 0$ (head-on motion feels the bare $V(r)$ ), and it guards the origin only when $V$ diverges more slowly than $1/r^2$ — inverse-cube and steeper attractions crash through it.

Worked examples

1 · Locating the ISS in its effective-potential well (confidence builder)

The ISS ( $m = 4.2\times10^5\ \mathrm{kg}$ ) orbits at altitude $420\ \mathrm{km}$ , so $r = 6.371\times10^6 + 4.2\times10^5 = 6.79\times10^6\ \mathrm{m}$ , with $GM_\oplus = 3.986\times10^{14}\ \mathrm{m^3/s^2}$ . Find its angular momentum, verify $r$ is the well bottom for that $L$ , find the well depth, and the radial oscillation frequency. (Earth is so much heavier that $\mu \approx m$ and $k = GM_\oplus m$ .)

Step 1. Circular speed from the centripetal balance (6): $v = \sqrt{GM_\oplus/r} = \sqrt{3.986\times10^{14}/6.79\times10^{6}} = \sqrt{5.87\times10^7} = 7.66\times10^3\ \mathrm{m/s}$ .

Step 2. Angular momentum: $L = mvr = 4.2\times10^5 \times 7661 \times 6.79\times10^6 = \boxed{2.19\times10^{16}\ \mathrm{kg\,m^2/s}}$ .

Step 3. Well-bottom radius from (9): $r_0 = L^2/\mu k$ with $k = GM_\oplus m = 1.674\times10^{20}$ : $r_0 = (2.185\times10^{16})^2/(4.2\times10^5 \times 1.674\times10^{20}) = 4.77\times10^{32}/7.03\times10^{25} = \boxed{6.79\times10^6\ \mathrm{m}}$ — exactly the actual radius. A circular orbit is the statement that the body sits at the bottom of its own $V_{\mathrm{eff}}$ well.

Step 4. Well depth from (10): $V_{\mathrm{eff}}(r_0) = -\mu k^2/2L^2 = -(4.2\times10^5)(1.674\times10^{20})^2/(2\times4.77\times10^{32}) = \boxed{-1.23\times10^{13}\ \mathrm{J}}$ . Per kilogram that is $-29.3\ \mathrm{MJ/kg}$ — the energy that must be supplied, per kilogram, to lift the station from this orbit to rest at infinity.

Step 5. Radial frequency from (8). For Kepler, $V_{\mathrm{eff}}''(r_0) = -2k/r_0^3 + 3L^2/\mu r_0^4 = \mu^3k^4/L^6$ , so $\omega_r = \mu k^2/L^3 = (4.2\times10^5)(1.674\times10^{20})^2/(2.185\times10^{16})^3 = 1.177\times10^{46}/1.043\times10^{49} = \boxed{1.13\times10^{-3}\ \mathrm{rad/s}}$ . That is a breathing period of $2\pi/\omega_r = 5.57\times10^3\ \mathrm{s} = 92.8$ minutes — identical to the orbital period $2\pi r/v$ . Radial and angular clocks tick together: a nudged ISS orbit becomes a small closed ellipse, not a precessing rosette.

2 · Reading a transfer orbit off the well (exam level)

A satellite is in a geostationary transfer orbit with perigee $r_p = 7.0\times10^6\ \mathrm{m}$ and apogee $r_a = 4.2\times10^7\ \mathrm{m}$ . Using only the effective-potential formalism (per unit mass: $\varepsilon = \tfrac12\dot r^2 + v_{\mathrm{eff}}$ , $\ell = r^2\dot\phi$ ), find $\ell$ , $\varepsilon$ , and the speeds at perigee and apogee. $GM_\oplus = 3.986\times10^{14}\ \mathrm{m^3/s^2}$ .

Step 1. Perigee and apogee are the two turning points: $\varepsilon = -GM_\oplus/r + \ell^2/2r^2$ at both radii. Writing this once for $r_p$ and once for $r_a$ gives two equations in the two unknowns $(\varepsilon, \ell)$ .

Step 2. Subtract the two turning-point equations to eliminate $\varepsilon$ : $GM_\oplus\left(\tfrac{1}{r_p} - \tfrac{1}{r_a}\right) = \tfrac{\ell^2}{2}\left(\tfrac{1}{r_p^2} - \tfrac{1}{r_a^2}\right)$ . Dividing by $\left(\tfrac{1}{r_p} - \tfrac{1}{r_a}\right)$ and using $a^2-b^2 = (a-b)(a+b)$ : $GM_\oplus = \tfrac{\ell^2}{2}\left(\tfrac{1}{r_p} + \tfrac{1}{r_a}\right)$ , so $\ell^2 = \dfrac{2GM_\oplus\, r_p r_a}{r_p + r_a}$ .

Step 3. Numbers: $\ell^2 = 2 \times 3.986\times10^{14} \times \dfrac{7.0\times10^6 \times 4.2\times10^7}{4.9\times10^7} = 7.972\times10^{14} \times 6.0\times10^6 = 4.78\times10^{21}$ , giving $\ell = \boxed{6.92\times10^{10}\ \mathrm{m^2/s}}$ .

Step 4. Energy from the perigee turning point: $\varepsilon = -\dfrac{3.986\times10^{14}}{7.0\times10^6} + \dfrac{4.78\times10^{21}}{2(7.0\times10^6)^2} = -5.69\times10^7 + 4.88\times10^7 = \boxed{-8.13\times10^6\ \mathrm{J/kg}}$ . Negative, as it must be for an orbit trapped between two turning points. (Cross-check: $\varepsilon = -GM_\oplus/(r_p+r_a) = -3.986\times10^{14}/4.9\times10^7 = -8.13\times10^6$ ✓.)

Step 5. Speeds. At a turning point $\dot r = 0$ , so all motion is angular and $v = \ell/r$ : $v_p = 6.92\times10^{10}/7.0\times10^6 = \boxed{9.88\ \mathrm{km/s}}$ and $v_a = 6.92\times10^{10}/4.2\times10^7 = \boxed{1.65\ \mathrm{km/s}}$ . The perigee speed is what a real GTO injection burn delivers — and the 6-fold speed ratio is just $r_a/r_p$ , angular momentum conservation made audible.

Figure

Same physics, two views — the annulus a bound orbit fills in space, and the well the same orbit rocks in

Origins · 1 minIsaac Newton and how this was discovered

Isaac Newton

1643–1727 · Trinity College, Cambridge

The effective potential as a named object is a nineteenth-century convenience, but its physics is in the Principia. In Book I, Propositions 43–45, Newton proved his "theorem of revolving orbits": adding a force proportional to $1/r^3$ to any central attraction leaves the radial motion completely unchanged and merely makes the orbit's apsides revolve. That inverse-cube force is precisely the form of the centrifugal term — Newton had discovered that angular momentum's contribution to radial dynamics behaves as an inverse-cube ingredient that can be traded against real forces. He then wielded the result as a precision instrument: because the Moon's apogee advances only slowly (about 3° per orbit), the true force law had to sit extremely close to inverse-square — turning apsidal precession into the first quantitative test of universal gravitation. The modern habit of reading orbital stability off a one-dimensional curve is the same argument, refactored.

History

1687

— Newton's Principia, Book I Props. 43–45: the theorem of revolving orbits shows an added inverse-cube force only precesses the apsides — the centrifugal term's dynamics in embryonic form, and apsidal motion becomes a test of the force law.

1747

— Clairaut announces the lunar apsidal precession disagrees with inverse-square gravity by a factor of two, briefly proposing to modify the force law; by 1749 he resolves it with a more careful perturbation expansion — a triumph for

1/r^2

1873

— Joseph Bertrand proves that among all central forces only

-k/r

and the linear spring force close every bound orbit — the sharpest statement ever extracted from the shape of the effective potential.

1916

— Schwarzschild's solution of Einstein's equations adds a

-GML^2/c^2r^3

term to the effective potential; the barrier now has a top, and sufficiently close orbits plunge.

1972

— Bardeen, Press and Teukolsky systematize effective potentials for orbits around rotating black holes, the working framework of modern accretion-disk and gravitational-wave astrophysics.

Going further

Texts. Goldstein, Poole & Safko, Classical Mechanics, 3rd ed., §3.3 and §3.6 · Landau & Lifshitz, Mechanics, 3rd ed., §14 · Morin, Introduction to Classical Mechanics, ch. 7 · Taylor, Classical Mechanics, ch. 8.

Provenance. Verifier should confirm: (1) the algebra eliminating $\dot\phi$ to give $V_{\mathrm{eff}}=V+L^2/2\mu r^2$ ; (2) the Kepler results $r_0=L^2/\mu k$ , $V_{\mathrm{eff,min}}=-\mu k^2/2L^2$ , and the eccentricity $e=\sqrt{1+2EL^2/\mu k^2}$ (Goldstein §3.7, Morin Ch. 7); (3) the power-law stability bound $n>-3$ via $V_{\mathrm{eff}}''=(n+3)kr^{n-1}$ (Landau-Lifshitz §14); (4) the radial oscillation frequency $\omega_r=\sqrt{V_{\mathrm{eff}}''/\mu}$ . All formulas are standard textbook results. Worked-example numbers on this page (ISS: $v = 7.66\ \mathrm{km/s}$ , $L = 2.19\times10^{16}\ \mathrm{kg\,m^2/s}$ , well depth $-1.23\times10^{13}\ \mathrm{J}$ , $\omega_r = 1.13\times10^{-3}\ \mathrm{rad/s}$ ; GTO: $\ell = 6.92\times10^{10}\ \mathrm{m^2/s}$ , $\varepsilon = -8.13\times10^6\ \mathrm{J/kg}$ , $v_p = 9.88\ \mathrm{km/s}$ , $v_a = 1.65\ \mathrm{km/s}$ ) are illustrative arithmetic recomputed from the boxed formulas.

Builds on: Central force fundamentals · The two-body problem and reduced mass · Angular momentum conservation in central fields · Reduction to the radial equation. Leads to: Classification of orbits by energy · The Kepler problem · Conic-section orbits · Kepler's third law.

Is the centrifugal potential the same thing as the centrifugal force I met in rotating frames?

Close kin, not the same object. The rotating-frame centrifugal force $-m\boldsymbol{\omega}\times(\boldsymbol{\omega}\times\mathbf{r})$ appears when you bolt your coordinates to a frame rotating at fixed $\boldsymbol{\omega}$ . The centrifugal term in $V_{\mathrm{eff}}$ appears in an inertial frame, from eliminating $\dot\phi$ with conservation of $L$ ; the "frame" it secretly matches is one co-rotating with the body itself, whose $\dot\phi$ varies along the orbit. The gradient of $L^2/2\mu r^2$ is $L^2/\mu r^3 = \mu r\dot\phi^2$ — exactly the instantaneous centrifugal force of that co-moving frame. So the barrier is the centrifugal effect correctly integrated into energy bookkeeping, with $L$ held constant instead of $\omega$ . Confusing the two constants ( $L$ fixed vs $\omega$ fixed) gives wrong signs in stability arguments — a classic gradebook error.

If the barrier isn't a real force, what physically stops the comet at perihelion?

Gravity does — by way of geometry. In the inertial frame the only force is the inward pull. But as the comet closes in, conservation of angular momentum makes its transverse speed grow as $1/r$ , so its kinetic energy grows as $1/r^2$ while gravity's energy subsidy only grows as $1/r$ . At some radius all the inward radial motion has been converted into transverse motion: $\dot r = 0$ , the comet is moving purely sideways, and gravity — still pulling inward — now curves the path back outward in $r$ . Nothing pushed outward at any moment. The effective potential simply prices this conversion in advance: the barrier is the transverse kinetic energy the comet is contractually obliged (by $L$ conservation) to carry at each radius.

Why does a minimum of a potential mean stability? I want the physical mechanism, not the math

At the well bottom the net radial "force" $-V_{\mathrm{eff}}'$ is zero — attraction and the angular-momentum cost balance. Displace the body slightly inward: the barrier's push (steepening as $1/r^3$ ) now exceeds the attraction's pull, so the net effect drives it back out. Displace it outward: attraction wins and pulls it back in. Every excursion generates a response pointing home — the definition of a restoring force, and small restoring forces proportional to displacement are simple harmonic motion, giving the breathing frequency $\omega_r = \sqrt{V_{\mathrm{eff}}''/\mu}$ . At a maximum the response points away from home in both directions: any nudge is amplified. The curvature $V_{\mathrm{eff}}''$ is just the per-unit-displacement strength of that response, sign included.

Where does the factor arrangement in

L^2/2\mu r^2

come from? I keep writing

L^2/2\mu r

L/2\mu r^2

on exams

Reconstruct it in five seconds rather than memorizing: angular kinetic energy is $\tfrac12\mu v_\perp^2$ , and the transverse speed is $v_\perp = r\dot\phi = L/\mu r$ (from $L = \mu r v_\perp$ ). Square it: $\tfrac12\mu (L/\mu r)^2 = L^2/2\mu r^2$ . The $r^2$ downstairs is forced by $v_\perp \propto 1/r$ ; the single $\mu$ downstairs survives because one of the two $\mu$ 's from squaring cancels against the $\tfrac12\mu$ upstairs. Dimension check: $[L^2] = \mathrm{kg^2\,m^4\,s^{-2}}$ , divide by $\mathrm{kg\,m^2}$ and you get $\mathrm{kg\,m^2\,s^{-2}}$ — joules. Any misplacement fails this check immediately.

Does the effective-potential trick work for forces that are not central?

Not directly — it stands on two legs that non-central forces kick away. First, the orbit must stay in a plane (needs $\mathbf{L}$ conserved in direction); second, $L$ 's magnitude must be constant so $\dot\phi$ can be eliminated in favour of $r$ alone. A non-central force exerts torque, $L$ drifts, and the "barrier" $L(t)^2/2\mu r^2$ stops being a function of $r$ only. The rescue in practice: when the non-central part is weak (Earth's equatorial bulge acting on a satellite, a third body perturbing a planet), one treats the motion as an effective-potential orbit whose parameters $L$ and $E$ drift slowly — the method of osculating elements. And a time-dependent central force keeps the planarity but breaks energy conservation, so the water-line picture fails while the barrier survives.

Black holes have an innermost stable circular orbit but Newtonian gravity doesn't. What changes in the effective potential?

General relativity (Schwarzschild) adds one term: $V_{\mathrm{eff}} = -\tfrac{GM}{r} + \tfrac{L^2}{2r^2} - \tfrac{GML^2}{c^2r^3}$ (per unit mass). That last piece is attractive and scales as $1/r^3$ — steeper than the barrier's $1/r^2$ — so at small $r$ it always wins. The Newtonian wall becomes a crest: a maximum (unstable circle) appears inside the minimum (stable circle). As you reduce $L$ , the pair slide together and annihilate at $r = 6GM/c^2$ , the ISCO. Below that, no circular orbit is stable at any $L$ , which is why accretion disks have inner edges and why inspiralling black-hole binaries end in a plunge. Note the Newtonian stability analysis of this article still runs verbatim — only the input curve changed.

Is the well depth

-\mu k^2/2L^2

the same as the orbit's total energy?

Only for the circular orbit. The well depth is the minimum possible energy at the given $L$ ; a body actually orbiting at the well bottom has exactly that energy, $E = -\mu k^2/2L^2$ . Every other orbit with the same $L$ has $E$ above the bottom: ellipses fill the range $-\mu k^2/2L^2 < E < 0$ , and unbound orbits have $E \ge 0$ . Conversely, at fixed energy $E<0$ , orbits of different $L$ sit in different wells — the semimajor axis (hence period) depends only on $E$ , while $L$ selects the eccentricity. Keeping "depth = floor, not the energy itself" straight prevents the common exam slip of equating $-\mu k^2/2L^2$ to the energy of a general ellipse.

Why do the radial and angular frequencies coming out equal for gravity matter so much?

Because their ratio decides what the orbit looks like. In one radial breathing period the body sweeps an angle $\omega_\phi/\omega_r \times 2\pi$ . If that is exactly $2\pi$ — as for Kepler, where both frequencies equal $\mu k^2/L^3$ — perihelion returns to the same direction every cycle and the orbit is a closed, frozen ellipse. If the ratio is irrational, perihelion advances each cycle and the orbit is a rosette that never repeats. Bertrand's theorem says the closed-for-all-orbits property is a razor's edge: only $-k/r$ and $\tfrac12\mu\omega^2r^2$ have it. That fragility is a gift to astronomers — any small extra physics (planetary perturbations, solar oblateness, general relativity) betrays itself as a slow apsidal precession, which is how Mercury's famous 43 arcseconds per century became the first test of Einstein.

The Effective Potential

Where this lives today

The physical picture

Building the effective potential

Circular orbits: the extrema of the curve

Which force laws allow stable circles at all

Which equation answers which question

The Kepler well, quantitatively

Limiting cases

Worked examples

1 · Locating the ISS in its effective-potential well (confidence builder)

2 · Reading a transfer orbit off the well (exam level)

Going further

Before this

Where this leads