Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

Orbits in Axisymmetric potentials

L. Aguilar 1
Orbits in Axisymmetric potentials
We now tackle the problem of motion in axially Notice that the condition set by the equation we
symmetric potentials. Few galaxies are truly gave in this slide does not imply that the
spherical, but some may be approximated by axisymmetric potential has the same mass within a
figures of revolution. cylinder of radius R (or sphere of radius r = R) as
the spherical potential on the right side.

To begin, we make the observation that a particle


that moves within the equatorial plane of a
potential with cylindrical symmetry, has no way of
knowing that it moves in a non-spherical potential:

So, all the results for


spherical potentials do
apply to these orbits:
So, although the motion is set by an equivalent
they move on planar
spherical potential, we have no way of inferring
rosettes.
the mass profile directly from the rotation curve,
as it is the case for spherical potentials.
We need additional information about the degree
of flattening and shape of the density profile.

L. Aguilar 2
Equations of motion
and types of orbits

L. Aguilar 3
Orbits in Axisymmetric potentials
The equations of motion
The θ component of the equation of motion is then:
It is natural that we use cylindrical coordinates:

which we recognize as the conservation of the z-


component of the angular momentum. Notice that now it
is only one component that is conserved, as opposed to
where we have used the expression for the acceleration the whole vector in the case of spherical potentials.
vector in polar coordinates that we derived in the
section for central forces, and the expression for the In a similar way as the case of spherical potentials, we
gradient in cylindrical coordinates assuming azimuthal can use this conservation property to eliminate the
symmetry. angular equation of motion by defining an effective
potential:
Separating by components:

The last term on the RHS acts as a centrifugal barrier,


similar to the case of spherical potentials.

The equations of motion to solve are


then:
Note that the angular equation can be written as a
total time derivative:
So, motion in an axisymmetric potential can be reduced
to a planar motion in the R-z meridional plane.

L. Aguilar 4
Orbits in Axisymmetric potentials
As Lz increases, the particle can not reach an ever The existence of a centrifugal barrier in spherical and
widening region around the axis of symmetry. axisymmetric potentials makes migration to the center a
difficult problem: it is not sufficient to dissipate energy,
angular momentum must also be lost. This makes the
feeding of supermassive black holes at the centers of

Effective equipotentials for an axisymmetric, singular logarithmic potential


galaxies a difficult affair.

The minimum in the effective potential occurs at the


point where:

If we assume symmetry with respect to the equatorial


plane (φ(-z) = φ(z)), which we will assume from now on,
then the second condition is satisfied at z = 0.

The first condition is satisfied at R = Rc:


with q = 0.6 and vo = 1.

We recognize this relation as the centrifugal


Avoidance
zone

equilibrium condition for a circular orbit of radius Rc.

L. Aguilar 5
Orbits in Axisymmetric potentials
The energy of a particle moving in an axisymmetric
potential is: Since the total energy remains constant, motion is only
possible (K > 0) when Eo > φeff, since it is their difference
that constitutes the kinetic energy in the meridional
plane.
Motion is thus restricted to be within the equipotential:
Then, the total energy is the sum of the kinetic energy
in the meridional plane plus the effective potential.

This equipotential is known as the zero-velocity curve


φeff (ZVC), since the orbit can only reach it with zero
z velocity (in the meridional plane) and can not cross it.
Zero-velocity
curve

Orbit in the potential


of the figure to the
left, but with E and
Lz indicated on top.
The particle is
launched from the
R point (3,1) on the
meridional plane and
with an initial
direction parallel to
E=Eo Zero-velocity the z-axis.
curve

Effective equipotential for an axisymmetric, singular


logarithmic potential with Rc = 0, vo = 1, q = 0.6 and Lz
= 1/4. The Eo = 0.2 surface is shown.
L. Aguilar 6
Orbits in Axisymmetric potentials
Something that we said before, but it is important
to repeat, is that individual orbits may not follow
the symmetry of the potential. But whenever this
occurs, there must be another, complementary
orbit, so that both together satisfy the potential
symmetry. The orbit we showed in the previous
slide (blue in the figure here), clearly does not
have symmetry with respect to the plane z = 0.
However, there is another orbit (brown), so that
both together satisfy the symmetry.

Another interesting feature is that if energy and


angular momentum were the only restrictions for
the orbits to the left, they ought to fill
completely the region inside the zero velocity
curve, as there would be only one orbit for a
particular combination of E and Lz. This is
obviously not the case here, were a given orbit not
only does not fill the allowed region, but there are
other orbits which share the same E and Lz
combination.

So, what’s going on?


L. Aguilar 7
Orbits in Axisymmetric potentials
As we sweep the R-axis at fixed E and Lz, we see
that different orbits appear within the ZVC,
however, not all seem to be part of the same
type.

Ri = 0.4 Ri = 0.6
For instance, both orbits on the upper row are
symmetric with respect to the equatorial plane.
But those in the middle row are not symmetric,
despite sharing the same energy, angular
momentum, direction of initial velocity and being
very close together initially in the R-axis. As we
move further along this axis, another symmetric
Ri = 0.8 Ri = 1.79 orbit appears (lowest panel).

Sequence of orbits, all in As we will see soon, this is the signal of two
the same potential as the things: First, there is another conserved quantity
figure in the previous slide,
as well as same E and Lz,
that restricts the orbits, and as the value of the
but launched from extra conserved quantity changes, we pass through
different points along the different “orbital families”. We will introduce the
line z=0, at 45° with concept of orbital families later on in this chapter.
Ri = 3.1
respect to the R-axis

L. Aguilar 8
Orbits in axisymmetric
potentials and conserved
quantities

L. Aguilar 9
Orbits in Axisymmetric potentials
Revealing the orbital structure:
This constraints
The Poincaré Section orbits of this
energy to be inside
The French mathematician Henri a 3-D volume in the
Poincaré devised a procedure 4-D phase space.
that allows us to reveal the
orbital structure in phase space
of dynamical systems with 2
degrees of freedom, as is the We now imagine that we cut this volume with a plane
case for 3D axisymmetric parallel to a spatial coordinate and its corresponding
.
potentials. velocity, say (x1, x1).
By the way, he is also the
discoverer of chaos in
deterministic systems, but this is Henri Poincaré (1854-1912) .
x1
another story. x1

We begin with a general 2-D potential: φ(x1, x2).


. .
The corresponding phase space is 4-D: (x1, x2, x1, x2) x2
Since we have assumed a time-independent potential,
energy must be conserved:
This plane (or section) is
defined by the condition x2 = 0
.
(but not x. 2 = 0, as we will see.

L. Aguilar 10
Orbits in Axisymmetric potentials
Since energy conservation makes one of the phase space In order to distinguish particles moving along the same
coordinates dependent on the others, we can write: orbit but in opposite sense, we only plot the
intersections where the orbit crosses the plane along a
.
definite direction, say x2 > 0. Then our Poincaré section
This means that setting x2 = 0 constraints each point is given by:
. .
within the (x1, x1) plane to a unique velocity x2.

Similarly, we can define the orthogonal section as:


We now choose a
point on this
plane and
. its
unique x2 velocity Now, if energy is the only constraint, as we said before,
as the starting . the orbit must fill densely the 3D constant energy
x1 volume and its intersections with the section should fill
point of an orbit. x1
We follow the densely an area. But if there is another conserved
orbit through quantity, this additional restriction means that the
the constant intersection will now describe a lower dimensional space:
energy volume a line.
and plot on the
x2
section all points
where the orbit
pierces the
plane.

1 conserved quantity 2 conserved quantities


L. Aguilar 11
Orbits in Axisymmetric potentials
Here are 2 orthogonal Poincaré sections for the singular, axisymmetric logarithmic potential
(the case shown corresponds to qφ = 0.9, Rc = 0.14, vo = 1 and E = 0.5667).
The black dotted lines correspond to lines of constant orthogonal velocity, i.e. the velocity that does not appear
as ordinate in the section. The outermost such contour, which appears with a solid blue line, is the zero-velocity
curve (minimum velocity orthogonal to the section). The colored dotted points correspond to orbit intersections
(different colors, different orbits).
In both sections, we see 2 groups of orbits, each surrounding single points that straddle the section center.
They are the same orbital family but with opposite rotation directions.

L. Aguilar 12
Orbits in Axisymmetric potentials
Here we have three particular orbits in the potential of last slide. The energy and initial point in the y=0
poincaré section are indicated on top of each plot. Notice that all have the same energy. The space shown is (x,
y, vx).
The brown solid is one-half (y < 0) of the constant energy volume (the other half is not shown to reveal the
interior). The yellow lines are the orbits in this 3D projections of phase-space and the green points are the
intersection with the x-vx Poincaré section. The grey shadows at the bottom are the orbits in configuration
space. Notice that all orbits belong to the same family (they are topologically equivalent in phase space).

L. Aguilar 13
Orbits in Axisymmetric potentials
Here we continue the sequence of orbits of the previous slide. The potential and energy are the same, as well as
the initial x-coordinate. The only change is in the initial velocity.
We can see that we have crosses the boundary that separates two orbital families and we now have box orbits
that do not have a fixed sense of rotation. The first orbit envelope is very similar to that of the last in the
previous slide. In the second orbit, two avoidance zones appear and the orbit resembles a bowtie. In the last
orbit, we approach the axial orbit. The corresponding points in the Poincaré section approach the zvc.

L. Aguilar 14
Orbits in Axisymmetric potentials
Poincaré sections of Milky Way Galaxy models

In the chapter on potentials we saw three potential models used to describe our Galaxy: the Bahcall & Soneira,
. the
Ostriker and Caldwell and the Allen and Santillán models. Here we present a series of Poincaré sections (R, R) that
reveal the orbital structure of these three models.

Bahcall & Soneira model

Each column is at a fixed radial position


indicated in the upper edge (in kpc).
Each row is at fixed angular momentum
indicated in the right edge (in units of the
corresponding value for the local circular
orbit).

Notice how for low angular momentum,


almost all orbits have only 2 conserved
quantities (energy and angular
momentum), while for high angular
momentum, all orbits seem to have an
extra conserved quantity.

L. Aguilar 15
Orbits in Axisymmetric potentials
Poincaré sections of Milky Way Galaxy models

In the chapter on potentials we saw three potential models used to describe our Galaxy: the Bahcall & Soneira,
. the
Ostriker and Caldwell and the Allen and Santillán models. Here we present a series of Poincaré sections (R, R) that
reveal the orbital structure of these three models.

Ostriker & Caldwell model

Same as before, but now for this model.

Although orbits with only 2 conserved


quantities are less prevalent than in the
previous model, the same trend with
angular momentum is apparent.
We also notice that around 1 kpc the
sections seem more complex, which is the
region where we the potential goes from
bulge-dominated to disk-dominated.

L. Aguilar 16
Orbits in Axisymmetric potentials
Poincaré sections of Milky Way Galaxy models

In the chapter on potentials we saw three potential models used to describe our Galaxy: the Bahcall & Soneira,
. the
Ostriker and Caldwell and the Allen and Santillán models. Here we present a series of Poincaré sections (R, R) that
reveal the orbital structure of these three models.

Allen & Santillán model

Same as before, but now for this model.

Here the prevalence of orbits with only 2


conserved quantities at low angular
momentum is larger. Close to the circular
orbit the 1 kpc section seems to be
transitional, with all sections further out
pretty much the same (lower row).

Source for all sections:


Valera, Aguilar & Schuster (1994)
in “Numerical Simulations in Astrophysics”
Proc. of the 1st UNAM-CRAY Supercomputing Workshop.
Ed. J. Franco, S. Lizano, L. Aguilar, E. Daltabuit. p. 111
(Cambridge Univ. Press)

L. Aguilar 17
Orbits in Axisymmetric potentials

The question of the “Third integral”

To finish this section we will mention a famous problem


in Galactic Structure: the question of the “third
integral”.

As it is obvious from the Poincaré sections of the three


galactic models shown, most orbits seem to have a third
conserved quantity. The Poincaré section allows us to
identify its presence. Unfortunately it doesn’t tells us
what this extra conserved quantity, or third integral, is.

L. Aguilar 18
The Epicyclic
approximation:
Life near the circular
orbit

L. Aguilar 19
The epicyclic approximation
Since the majority of stars near the Sun (thin disk) The constant term can be renormalized away. Then, to
move very close to circular orbits, it is useful to quadratic order, the effective potential around the
develop an approximation that describes these orbits. minimum (see figure) can be approximated as:
This is the so called epicyclic approximation.

We recall that the circular orbit is given by the


minimum of the effective potential on the z = 0 plane.
where we have shifted the coordinate origin: x ≡ R-Rc.
Also, the equations of motion in the meridional plane
are given by:

~φeff

For R ≈ Rc, where the latter is the radius of the circular x R


orbit, we can approximate the effective potential as:

where we have dropped the linear terms, since the 1st


derivatives are zero at the minimum. The quadratic
cross term has also been omitted by symmetry with
respect to the galactic plane.

L. Aguilar 20
The epicyclic approximation
The general solution is:
We now define the constants:

The constant κ is the epicyclic frequency, while ν is called


the vertical frequency. Notice that, since the curvature
z z of the potential for points in the galactic plane is larger
These constants, which have
units of time frequency, along the vertical direction (the density gradient is
measure the curvature of the x R larger), the vertical frequency is bigger and the
effective potential along the R corresponding period shorter:
and z directions, at the
minimum.
(Rc,0)

Let us look a bit more in detail these frequencies:


The approximate potential can now be written simply
as:

and the corresponding equations of motion are:

which is just a couple of independent harmonic oscillators.

This should not surprise us, as the epicyclic approximation


is just a linearization of our original problem.

L. Aguilar 21
The epicyclic approximation
We then have:

The epicyclic frequency is very important in


characterizing the kinematics of near-circular orbits.
However, it is not an observable variable. So, we are
now going to establish a link between it and
observable quantities, in particular, the rotation
curve. where in the second row we used the first expression
for Ωc squared and in the third row the second
We write the circular orbit velocity as:
expression.

Comparing now with the expression for κ2:


where Ωc is the angular velocity of the circular orbit at
radius R and Lc is its corresponding z-component of
angular momentum.

From here, we get two expressions for Ωc squared: So, the epicyclic frequency is determined
by the rotation curve!

L. Aguilar 22
The epicyclic approximation
Let’s examine three cases:

I. Solid body rotation

II. Constant circular velocity

III. Keplerian fall


off

So, in general:
Solid Constant
Keplerian
body velocity
fall off The more homogeneous a mass distribution, the closer
rot. curve
to the upper limit.
The more centrally concentrated, the closer to the
lower limit.

L. Aguilar 23
The epicyclic approximation
Limits of the epicyclic Motion in the galactic plane
approximation
As we saw, the solution for motion in x is just:
For vertical motion, the epicycle approximation is valid
as long as:
where xo > 0 and α are integration constants set by the
initial conditions.

Consider now a star moving under the epicycle


From Poisson’s equation in cylindrical coordinates: approximation around the circular orbit with the same
angular momentum.
We have:
Orbit Radial pos. Ang. Vel. Ang. Mom.
If we fix R, we see that:
Circular Rc Ωc R c 2 Ωc
. .
Epicyclic Rc + x(t) θ R2 θ

The angular velocity of the epicyclic particle


so, the epicyclic approximation is valid for vertical is:
Then,
motion only as long as ρ(z) ≈ constant, which limits
ourselves to a very thin disk (z ≤ 50 pc).

which is a good approximation since

L. Aguilar 24
The epicyclic approximation
x Epicyclic
particle
Integrating this expression: If we take the y axis
along the tangential
direction, we can write:
particle in y
circular
orbit
We now use the solution for x(t): θ-θc
Substituting the
solution for θ(t): Rc

The constant: So the motion on the plane of x


symmetry is an ellipse whose
center is given by the guiding
is intrinsic to the potential and it is adimensional. center provided by the y
circular orbit of the same
angular momentum.
The azimuthal solution is then: The axis ratio of the ellipse is
given by:

Rc
and the sense of
where θc(t) = θo + Ωc t is the azimuthal position of a rotation on the
particle in the corresponding circular orbit. elliptical epicycle is
contrary to the galactic
rotation.
L. Aguilar 25
The epicyclic approximation
For solid body rotation, we recall that: κ = 2Ωc ⟹ γ = 1
For a constant rotation curve: κ = √2 Ωc ⟹ γ = √2
For Keplerian fall off: κ = Ωc ⟹ γ = 2.

Solid
body
rotation

κ = 2Ωc
Flat
rotation
curve
κ = √2 Ωc
Keplerian
The epicycle period is: PR = 2π /κ, while the guiding fall off
center period is Pθ = 2π / Ωc .
κ = Ωc
Whenever the ratio (κ/Ωc) is commensurate (it is
rational), the epicycle motion will close on itself in the
non-rotating inertial frame. In the solar neighborhood
we have (xo/yo) ≈ 0.7
L. Aguilar 26

You might also like