Potentials and critical points¶
The Jacobi potential¶
Rocheometry's potential is the positive Jacobi potential, normalized by the total mass and kept in the binary's centre-of-mass gauge. In a component's local frame,
where \(r_i^2 = x^2+y^2+z^2\) and \(r_c^2 = (x-1)^2+y^2+z^2\). It is related to the dimensional rotating-frame effective potential by
U is the rotating-frame pseudo-potential of the restricted three-body problem; the Jacobi constant is \(C_J = 2U - v^2\), so a zero-velocity surface has \(C_J = 2U\).
The name U avoids \(\Omega\), which the binary-star literature uses for the differently normalized Kopal potential.
Because the centre-of-mass constant is kept, exchanging the components is an exact identity:
Potential values¶
A potential value carries its convention. A Jacobi value belongs to the binary, so it is the same for both components; a Kopal value is component-centred and must name its component:
import rocheometry
jacobi = rocheometry.Potential.jacobi(1.9)
kopal = rocheometry.Potential.kopal(4.0, component_id=1)
binary = rocheometry.BinaryGeometry(q=0.4)
converted = binary.convert_potential(
kopal, convention=rocheometry.PotentialConvention.JACOBI
)
same = rocheometry.convert_potential(
kopal, q=0.4, convention=rocheometry.PotentialConvention.JACOBI
)
Bare floats are not accepted as potentials.
Every potential Rocheometry returns, such as surface.potential or geometry.potential_at(point), is a Jacobi value rounded once.
A ComponentGeometry takes only Jacobi values, because a Kopal value's component number belongs to the binary; binary views take either.
Kopal potential¶
For component i, with \(q_{K,i} = M_{\mathrm{companion}}/M_i = 1/q_i\), the component-centred Kopal potential in that component's local frame is
normalized by \(GM_i/a\). It converts exactly to
From Wilson–Devinney or PHOEBE¶
Wilson–Devinney (Wilson & Devinney 1971) and PHOEBE (Prša et al. 2016) define the mass ratio as M2/M1 and quote Kopal potentials in the primary's frame, component 1, with \(q_K\) equal to their mass ratio.
So pass the reciprocal mass ratio, and the potential as a Kopal value for component 1; a recipe works through an example.
Component-scaled potential¶
Another common form normalizes by the companion's mass:
Every convention is affine in U, so surfaces, critical points and tangent directions are the same in all of them; gradients and Hessians scale by the same factor, here \(1+q_i\).
Critical points¶
geometry = rocheometry.ComponentGeometry(qi=0.4)
geometry.l1() # the inner Lagrange point
geometry.outer() # the collinear point beyond this component
geometry.triangular() # the triangular point at +y
binary.l1(), binary.l2(), binary.l3(), binary.l4(), binary.l5()
L2 lies beyond the less massive component and L3 beyond the more massive one, so at equal masses L2 is component 2's outer().
On the centreline the three collinear points are the roots of \(\partial U/\partial x\) in \(x < 0\), \(0 < x < 1\) and \(x > 1\).
In a component's local frame, L1 solves
At equal masses \(x_{\mathrm{L1}} = 1/2\) and \(U_{\mathrm{L1}} = 2\); at either extreme of mass ratio \(U_{\mathrm{L1}}\) tends to \(3/2\).
Critical points keep their relative accuracy very close to a component.
When L1 lies so close to x = 1 that its coordinate cannot be represented, l1() raises ValueError.
Each call returns a frozen CriticalPoint(position, potential, hessian).
The potential is a Jacobi value, and the Hessian is the symmetric 3 × 3 Hessian of U in the producer's axes, in units of \(G(M_1+M_2)/a^3\).
Binary views rotate it, \(R H R^\mathsf{T}\), but do not rescale it, so its eigenvalues are the same in every frame.
eigenvalues lists them in ascending order, and morse_index counts the negative ones: 2 for collinear points, 1 for triangular points.
Since U rises towards the components, collinear points rise along the centreline and fall across it, and triangular points rise in the orbital plane and fall out of it.
A recipe uses the Hessian at L1 to find the L1 nozzle factor and the cone angle.