Download - Mean Motion Resonances
![Page 1: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/1.jpg)
Mean Motion Resonances
Hamilton-Jacobi equation and Hamiltonian formulation for 2-body problem in 2 dimensions
Cannonical transformation to heliocentric coordinates for N-body problem
Symplectic integratorsCanonical transformation with
resonant angleMean motion resonances
Orbital trajectories of Jupiter's four largest moons over a period of 10 Earth days, illustrating the 4:2:1 resonance among the orbits of the inner three
![Page 2: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/2.jpg)
Integrable motion• Integrable: n degree of freedom Hamiltonian has n
conserved quantities• Examples:
– Hamiltonian is a function of coordinates only– Hamiltonian is a function of momenta only – in this case we can
call the momenta action variables and we can say we have transformed to action angle variables
– Hamiltonian has 1 degree of freedom and is time independent. H(x,v) gives level contours and motion is along level contours.
• Arnold-Louiville theorem - integrable implies that the Hamiltonian can be transformed to depend only on actions
![Page 3: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/3.jpg)
Hamilton Jacobi equation• If the coordinate q does not appear in the Hamiltonian
then the corresponding momentum p is constant• Try to find a Hamiltonian that vanishes altogether, then
everything is conserved• Generating function S2(q1,q2,…;P1, P2, …,t) function of old
coordinates and new momenta which we would like conserved
• New Hamiltonian
![Page 4: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/4.jpg)
Finding conserved momenta 2-body problem
• Our new momenta are conserved, constants of integration, Pi
• 2 body in 2 dimensions
• Hamilton Jacobi equationnew momenta
old coordinates
![Page 5: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/5.jpg)
• Assume separable
• Substitute momenta in to Hamilton Jacobi equation
• Since separable
• Hamilton Jacobi equation gives
Because we choose S separable these derivatives must be constant,as they are constant we also make them be our new momenta
![Page 6: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/6.jpg)
S depends on new monenta so α1α2 are new momenta
New coordinates
using expression for L
is energy
![Page 7: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/7.jpg)
• Sub in for constants and use
Q1 is time of perihelion
![Page 8: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/8.jpg)
• With a similar integral we can show that Q2 is angle of perihelion
• 3D problem done similarly• These coordinates not necessarily ideal. Add and subtract
them to find the Delaunay, modified Delauney and Poincaré coordinates
(Hamilton Jacobi equation and 2 body problem continued)
![Page 9: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/9.jpg)
Cannonical transformations
Different approaches 1. Choose desirable generating functions2. Solve integrals resulting from Hamilton-
Jacobi equation 3. Choose new coordinates and momenta and
show they satisfy Poisson brackets4. Use expansions (e.g., Birkhoff normal form)
(can lead to problems with small divisors)
![Page 10: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/10.jpg)
Example(continuing two body problem)
Desirable to have a third angle as a coordinateThe mean anomaly M = n(t-τ)
as expected
![Page 11: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/11.jpg)
Harmonic oscillator
• Use Poisson bracket to check that these variables are cannonical
you don’t get 1 here unless there is the factor of 2 in the variable
![Page 12: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/12.jpg)
Harmonic oscillator
Using a generating function
![Page 13: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/13.jpg)
Heliocentric coordinatesN-bodies in inertial frame
generating function
new Hamiltonian
![Page 14: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/14.jpg)
Democratic heliocentric coordinate system
• If P0=0 Hamiltonian becomes
barycenter so can be set to zeroThis is equivalent to using momenta that are in center of mass coordinate system
Keplerian termInteraction termDrift term
![Page 15: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/15.jpg)
Democratic heliocentric coordinates
• Hamiltonian is nicely separable in – heliocentric coordinates and – barycentric momenta
![Page 16: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/16.jpg)
Jacobi Coordinates
To add a body: work with respect to center of mass of all previous bodies. Coordinate system requires a tree to define.
![Page 17: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/17.jpg)
Symplectic Integrators• Wisdom and Holman used a Hamiltonian in Jacobi
coordinates for the N-body system that also separated into Keplerian and Interaction terms.
• Hamiltonian approximated by
– Integrated all bodies with f,g functions for dt = 1/Ω (that’s Hkep)– Velocities given a kick caused by Interactions– Integrate bodies again with f,g functions dt = 1/Ω
• Integrator is symplectic but integrates a Hamiltonian that approximates the real one.
• The integrator has bounded energy error and allows very large step sizes
![Page 18: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/18.jpg)
Second order Symplectic integrator
Poisson bracket with H gives evolution
Find coefficients so this is true to whatever order you desire (see Yoshida review)
![Page 19: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/19.jpg)
Second order Symplectic integrator
Expand to second order
=
~
![Page 20: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/20.jpg)
Second order for N-body
• Evolve Kepsteps via f,g functions• Drift step lets positions change as the Hamiltonian term only depends
on momenta• Interaction steps only vary velocities as they only depend on
coordinates• Reverse order.• Central step chosen to be most computationally intensive
– This is known as the democratic heliocentric second order integrator (Duncan, Levison & Lee 1998)
![Page 21: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/21.jpg)
Symplectic integrators- Harmonic oscillator
• The exact solution is
• To first order in τ
• However and energy increases with time
• A symplectic scheme can be constructed with
![Page 22: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/22.jpg)
More on symplectic integrators
• This is symplectic because the det is 1 and so volume is conserved• There is a conserved quantity but it’s not the original Hamiltonian
• You can show this by computing this quantity for p’,q’• The difference between new and old Hamiltonian depends on timestep!• Integrator is no longer symplectic if timestep varied.
![Page 23: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/23.jpg)
Approximating interactions with Periodic delta functions
• Interaction terms are used to perturb the system every time step
• Class of integrators where fast moving terms are replaced with periodic delta function terms.
• Note interaction term only depends on coordinates. • Here is an example:• We add extra terms to interaction term
periodic function
![Page 24: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/24.jpg)
Integrating the approximate Hamiltonian
• For t≠2πn
• For t=0,2π,4π…
• Integrate over the delta function
• Procedure: integrate unperturbed Hamiltonian between delta function spikes. At each t=0,2π,4π… update momenta. These are the velocity kicks.
![Page 25: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/25.jpg)
Justifications• Often the Hamiltonian has many cosine terms. As long as frequencies
are not commensurate, a perturbative transformation can remove non-resonant terms. Dynamics is only weakly sinusoidally varied by these terms. Cosine terms that are not commensurate are ignored.
• We can add in cosine terms without significantly changing the dynamics.
• The approximate Hamiltonian is conserved exactly.• Sizes of errors can be quantified. Errors are bounded as there is a
conserved quantity.• You can’t change the step size as this would change the approximate
Hamiltonian integrated. If you change the step size, the bounded property of errors is lost.
• Wisdom, Holman, and collaborators
![Page 26: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/26.jpg)
Similarity to Standard map
• I, θ mod 2π • If constant K is small there is no
chaos. • Width of chaotic zone depends on K• Above a certain value of K -> global
chaos
![Page 27: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/27.jpg)
RegularizationGiven • Hamiltonian H(p,q)• Initial conditions H(p0,q0,t=0)=H0
Let
We have a new Hamiltonian system but time can be re-scaled
Symplectic integrators that require equal timesteps can be constructed with the new time variable
`Extended phase space’ using time to deal with initial conditions issue
![Page 28: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/28.jpg)
General view of resonance
vector of integers k such that
Contrast with Periodic orbitsa period T such that for every frequency ωi
is close to an integer Zi
T is a multiple of the period of oscillation for every angle
![Page 29: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/29.jpg)
Small divisor problem
We would like to find new variables similar to the old variables
try this and try to find nice values for ck
Canonical transformationInsert this back into Hamiltonian
frequencies
expand perturbation in Fourier series
Choose These can be small!
![Page 30: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/30.jpg)
Small divisor problem continued
Nearing a resonance, the momentum goes to infinityWe will see that the infinity is not real, but a result of our assumption that new and old variables are similar
If there are no small divisors when removing first order perturbations, there may be small divisors when attempting to remove higher order perturbations from Hamiltonian via canonical transformation
first order perturbations removed
![Page 31: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/31.jpg)
Using the resonant angle
n-1 conserved quantities Ji because Hamiltonian lacks associated anglesCan expand H0 in orders of J0
resonant angle is a new coordinate
generating function
![Page 32: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/32.jpg)
Resonant angle
Simple 2D Hamiltonian. There is no infinity in the problem. Dynamics is similar to that of the pendulum.We did not assume that all new angles were similar to old angles in the transformation. The infinite response previously seen near resonance was caused by the choice of coordinate system
Above we considered only a single cosine termWhen there is more than one cosine term, then dynamics can be more complicated
Adopt a condition that the system is always sufficiently far from resonance, allowing perturbation theory to be done at all orders (Kolmogorov approach)Consider proximity of resonances (Chirikov approach)
![Page 33: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/33.jpg)
Expanding about a mean motion resonance (Celestial mechanics)
• j:j-k resonance exterior to a planet
• generating function
• new momenta
mean longitude of planet
unperturbed Hamiltonian units μ=1
ψ is resonant angle
![Page 34: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/34.jpg)
Expand around resonance
with coefficients
![Page 35: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/35.jpg)
Adding in resonant terms from the disturbing function
secular term
have opposite signs
![Page 36: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/36.jpg)
First order resonances
corotation term proportional to planet eccentricity usually dropped but can be a source of resonance overlap + chaos (Holman, Murray papers in 96)
sets distance to resonance
![Page 37: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/37.jpg)
One last C-transformation and going down a dimension
New Hamiltonian has no second angle so J2 is conserved We can ignore it in dynamicsIt may be useful to remember J2 later to relate changes in eccentricity to changes in semi-major axis in the resonance
![Page 38: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/38.jpg)
Distance to Resonance, Pendulum
Hamiltonian transformed to a pendulum Hamiltonian but with momentum shifted
fixed point is shifted
![Page 39: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/39.jpg)
Dimensional analysis
• Units of H cm2/s2 Units of Γ cm2/s • Units of a cm-2 Units of δ cm s-3/2
• δ2 × a units s-3 – Unit of time a-1/3 δ-2/3 sets typical libration period
• δ2/3 units cm2/3 s-1
– Unit of momentum δ2/3a-2/3 sets resonant widths• δ is proportional to Planet mass
– Resonant widths depend on planet mass to the 2/3 power.• Migration is a variation in mean motion so has units s-2
– Relevant to figure out adiabatic limit
![Page 40: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/40.jpg)
As distance to resonance varied
• Only one side has a separatrix• On one side it looks like a
harmonic oscillator• 3 fixed points but only 2 stable• Without circulating about origin
means in resonance• Can think of drifting problems as
having time dependent b
coordinate system
![Page 41: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/41.jpg)
Drifting systems
• Migrating planets• Dust under radiation forces
(though dissipative dynamics is NOT Hamiltonian)
• Satellite systems with tidal dissipation
![Page 42: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/42.jpg)
Drifting systems
Particle must jump to other side, no capture possible
Particle can be pumped to high eccentricity and remain in librating island, resonance capture possible
![Page 43: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/43.jpg)
Production of Star Grazing and Star-Impacting Planetesimals via Planetary Orbital Migration
Quillen & Holman 2000, but also see Quillen 2002 on the Hyades metallicity scatterFEB
Impact
Eject
Numerical integration of a planet migrating inwards
![Page 44: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/44.jpg)
Integration with toy model
2~ e
resonant angle fixed -- Capture
Escape
![Page 45: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/45.jpg)
Failure to capture
• Volume in resonance shrinking rather than growing (particle separating from planet)
• Non-adiabatic limit. Rapid drifting• Initial particle eccentricity is high.• Prevented by strong subresonances
![Page 46: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/46.jpg)
Adiabatic Capture Theoryfor Integrable Drifting Resonances
Rate of Volume swept by upper separatrixRate of Volume swept by lower separatrix
Rate of Growth of Volume in resonance
Probability of Capturec
VV
VP
V
VVV
Theory introduced by Yoder and Henrard, applied toward mean motion resonances by Borderies and Goldreich
![Page 47: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/47.jpg)
Critical eccentricity• Below a critical eccentricity probability
of capture is 100%• This eccentricity can be estimated via
dimensional analysis from the typical momentum size scale in the resonance
• It depends on order of resonance and a power of planet mass. Hence weak resonances require extremely low eccentricity for capture.
• Exact probability above ecrit in adiabatic limit can be computed via area integrals
![Page 48: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/48.jpg)
2
( ) / 2,
0
( , ; , )
cos( ( ) )k
k pk p p
p
K a b c
k p p
Limitations of Adiabatic theory
• At fast drift rates resonances can fail to capture -- the non-adiabatic regime.
• Subterms in resonances can cause chaotic motion.
temporary capture in a chaotic system
pe
![Page 49: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/49.jpg)
Coupled pendulum
• plot points every t=1/ν to make a surface of section
• 2D system (4D phase space) can be plotted in 2D with area preserving map
• Nice work relating timescales for evolution to overlap parameter by Holman and Murray
forced pendulum
analogous first order resonance system
![Page 50: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/50.jpg)
Chaos first develops near separatrix
chaotic orbits more likely on this side
![Page 51: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/51.jpg)
Non adiabatic limit• If drift is too fast capture is unlikely. Drifting
past resonance must be slow compared to libration timescale.
• Can be estimated via dimensional analysis• Use the square of the timescale of the
resonance. • This follows as drift rates are in units of t-2
![Page 52: Mean Motion Resonances](https://reader030.vdocument.in/reader030/viewer/2022012908/568165fe550346895dd92df1/html5/thumbnails/52.jpg)
Reading• Morbidelli’s book Chap 1• Murray and Dermott (solar system dynamics) Chap 8• Yoshida, H. 1993, Celestial Mechanics and Dynamical Astronomy, 56,
27, “Recent Progress in the Theory and Application of Symplectic Integrators”
• Duncan, Levison & Lee 1998, AJ, 116, 2067, “A multiple timestep symplectic algorithm for integrating close encounters”
• Holman, Matthew J., Murray, N. W. 1996, AJ, 112, 1278, “Chaos in High-Order Mean Resonances in the Outer Asteroid Belt”
• Quillen, A. C. 2006, MNRAS, 365, 1367, `Reducing the probability of capture into resonance’
• Mustill & Wyatt 2011, MNRAS, 413, 554, `A general model of resonance capture in planetary systems: first- and second-order resonances’