Stäckel Potential

A Stäckel potential is a class of integrable, three-dimensional gravitational potentials—discovered by German mathematician Paul Stäckel in his 1890 habilitation thesis and introduced into galactic dynamics by Grigori Kuzmin (1956), Donald Lynden-Bell (1962), and Tim de Zeeuw (1985)—in which the Hamilton–Jacobi equation of classical mechanics separates additively in confocal ellipsoidal coordinates (λ, μ, ν). In a general axisymmetric or triaxial galaxy potential, only one or two integrals of motion (the specific energy E and, in axisymmetric systems, the vertical angular momentum component L_z) follow from spatial symmetry alone, while numerical orbit integrations show that stars in the Milky Way and elliptical galaxies obey a "third integral" I₃ that prevents their vertical and radial velocity dispersions from equalizing (as observed in the solar-neighborhood velocity ellipsoid and vertex deviation). Stäckel potentials are the most general non-spherical potentials in three dimensions that possess three globally analytic isolating integrals of motion (E, I₂, I₃) quadratic in the momenta, providing the mathematical blueprint for the orbital backbone of triaxial elliptical galaxies and action–angle modeling of the Galactic disk.

To define a triaxial Stäckel potential in three-dimensional Cartesian space (x, y, z), one introduces confocal ellipsoidal coordinates τ = (λ, μ, ν) defined as the three real roots of the cubic equation for τ at any point (x, y, z):

x²/(τ + α) + y²/(τ + β) + z²/(τ + γ) = 1 with −γ ≤ ν ≤ −β ≤ μ ≤ −α ≤ λ

where α, β, γ are three fixed negative constants (−γ < −β < −α < 0) whose differences Δ₁ = (β − α)^1/2 and Δ₂ = (γ − β)^1/2 specify the focal distances along the z- and y-axes. The coordinate surfaces of constant λ, μ, and ν form three orthogonal quadric surfaces at every point:

• Surfaces of constant λ: Confocal ellipsoids enclosing the origin, becoming spherical as λ → ∞ (where λ ≈ r²).
• Surfaces of constant μ: Confocal hyperboloids of one sheet pierced along the intermediate axis.
• Surfaces of constant ν: Confocal hyperboloids of two sheets separating the polar caps.

In the axisymmetric oblate limit where the two equatorial focal constants coincide (α = β), confocal ellipsoidal coordinates reduce to prolate spheroidal coordinates (λ, ν, ϕ), where ϕ = arctan(y/x) is the cylindrical azimuth around the symmetry axis z and the coordinate curves in the meridional (R, z) plane are confocal ellipses and hyperbolas sharing foci at R = 0, z = ±(γ − α)^1/2.

Paul Stäckel (1890) proved that the Hamilton–Jacobi partial differential equation for the classical action S(λ, μ, ν) separates additively into three one-dimensional ordinary differential equations—S(λ, μ, ν) = S_λ(λ) + S_μ(μ) + S_ν(ν)—if and only if the gravitational potential Φ(λ, μ, ν) has the Stäckel form (Weinacht, 1924; Lynden-Bell, 1962; de Zeeuw, 1985):

Φ(λ, μ, ν) = −F(λ)/[(λ − μ)(λ − ν)] − F(μ)/[(μ − ν)(μ − λ)] − F(ν)/[(ν − λ)(ν − μ)]

where F(τ) is an arbitrary smooth single-variable function defined over the three coordinate intervals τ ∈ {λ, μ, ν}.

Because the Hamilton–Jacobi equation separates in a Stäckel potential, every stellar orbit conserves three isolating integrals of motion: the total specific energy H = E, and two non-classical integrals I₂ and I₃ quadratic in the canonical momenta p_τ = ∂S/∂τ = P_τ²τ̇. For each coordinate τ ∈ {λ, μ, ν}, the separated one-dimensional momentum equation reads:

p_τ² = [1 / (2(τ + α)(τ + β)(τ + γ))] [Eτ² − I₂τ + I₃ + F(τ)]

In an axisymmetric oblate Stäckel potential (α = β), the second integral reduces to the azimuthal angular momentum squared, I₂ = ½L_z², while the third integral I₃ takes the explicit Kuzmin–de Zeeuw form:

I₃ = ½(L_x² + L_y²) + ½(γ − α)v_z² + (γ − α)z²U(λ, ν)

Integrating the canonical momentum p_τ around a full oscillation cycle in each coordinate τ ∈ {λ, μ, ν} yields the three conserved orbital action variables J = (J_λ, J_μ, J_ν), J_τ = (1/2π) ∮ p_τ dτ. By Jeans's theorem for the collisionless Boltzmann equation, every orbit in a Stäckel potential is regular and confined to a three-dimensional invariant torus in six-dimensional phase space parameterized by f(E, I₂, I₃) = f(J_λ, J_μ, J_ν).

In his 1985 study, Tim de Zeeuw demonstrated that the signs of (I₂, I₃) partition phase space in any triaxial Stäckel galaxy into four distinct families of regular 3D orbits: box orbits (I₂ < 0, I₃ > 0, which oscillate across all three axes with ⟨L⟩ = 0 and pass through the harmonic core, but are scattered if a steep Dehnen profile cusp or central black hole is added), short-axis tube orbits (I₂ > 0, I₃ > 0, circulating around the minor z-axis), and inner and outer long-axis tube orbits (I₃ < 0, circulating around the major x-axis).

Furthermore, choosing the stratified ellipsoidal density profile

ρ(x, y, z) = ρ₀(1 + m²)⁻² where m² = (x/a)² + (y/b)² + (z/c)² (a ≥ b ≥ c > 0)

yields the Perfect Ellipsoid (de Zeeuw & Lynden-Bell, 1985)—the unique triaxial mass distribution stratified on concentric similar ellipsoids (a triaxial analog of the Plummer model and modified Hubble King profile) whose gravitational potential is a globally separable Stäckel potential.

Although composite galaxies containing cuspy Einasto profile halos, Hernquist profile bulges, and Miyamoto–Nagai potential disks are not globally separable Stäckel potentials, regular stellar orbits in the Milky Way behave locally as though they were moving in a fitted Stäckel potential. James Binney (2012) introduced the Stäckel Fudge algorithm (implemented in galactic dynamics libraries such as galpy and AGAMA), which approximates any axisymmetric galactic potential Φ(R, z) locally along a star's orbit as a prolate spheroidal Stäckel potential with focal distance Δ(E, L_z). This enables astronomers to compute the three canonical actions (J_R, J_ϕ = L_z, J_z) for millions of stars in the Gaia catalog in milliseconds per star without numerical orbit integration, powering modern equilibrium distribution-function models f(J), Eddington inversion formula generalizations, and Jeans equations inversions of the Galactic disk and halo.

Why do astronomers need a third integral of motion I_3 in the Milky Way?

If the phase-space distribution function of the Milky Way disk depended only on the two classical axisymmetric integrals—energy E and vertical angular momentum L_z, so f = f(E, L_z)—then v_R and v_z would enter f symmetrically through v_R² + v_z², forcing the radial and vertical velocity dispersions to be equal everywhere (σ_R = σ_z). In reality, solar-neighborhood stars have σ_R ≈ 35 km s⁻¹ ≠ σ_z ≈ 18 km s⁻¹, demonstrating that Galactic orbits conserve an effective third isolating integral I₃ (or vertical action J_z).

Why are there no chaotic orbits in a Stäckel potential?

By the Liouville–Arnold theorem of Hamiltonian mechanics, any three-degree-of-freedom Hamiltonian system that possesses three independent, mutually commuting isolating integrals of motion (H = E, I₂, I₃) is integrable. Every trajectory is quasi-periodic on an invariant 3-torus in 6D phase space, leaving zero phase-space volume for chaotic diffusion.

How does the Kuzmin–Toomre disk relate to Stäckel potentials?

Grigori Kuzmin (1956) showed that Φ(R, z) = −GM/[R² + (a + |z|)²]^1/2 is an axisymmetric Stäckel potential above and below the midplane z = 0. Thickening this potential by replacing |z| with (z² + b²)^1/2 yields the Miyamoto–Nagai potential.

1. Weinacht, J. (1924). "Über die bedingt-periodische Bewegung eines Massenpunktes." Mathematische Annalen, 91(3–4), 279–299. https://doi.org/10.1007/BF01556084
2. Lynden-Bell, D. (1962). "Stellar dynamics: Potentials with isolating integrals." Monthly Notices of the Royal Astronomical Society, 124(2), 95–123. https://doi.org/10.1093/mnras/124.2.95
3. de Zeeuw, T. (1985). "Elliptical galaxies with separable potentials." Monthly Notices of the Royal Astronomical Society, 216(2), 273–334. https://doi.org/10.1093/mnras/216.2.273
4. de Zeeuw, P. T., & Lynden-Bell, D. (1985). "Best approximate quadratic integrals in stellar dynamics." Monthly Notices of the Royal Astronomical Society, 215(4), 713–730. https://doi.org/10.1093/mnras/215.4.713
5. Binney, J. (2012). "Actions for axisymmetric potentials." Monthly Notices of the Royal Astronomical Society, 426(2), 1324–1327. https://doi.org/10.1111/j.1365-2966.2012.21757.x