Advanced Beam Dynamics & Hamiltonian Formalism: Rick Baartman, TRIUMF February 3, 2015
Advanced Beam Dynamics & Hamiltonian Formalism: Rick Baartman, TRIUMF February 3, 2015
Advanced Beam Dynamics & Hamiltonian Formalism: Rick Baartman, TRIUMF February 3, 2015
Formalism
D
2. Introduction
February 3, 2015
dpx
= evy B (1)
dt
dpy
= −evxB (2)
dt
dpz
= 0 (3)
dt
We set p~ = m~v , (come back to this later). Then we find vz is constant and
vx2 + vy2 is constant, and each varies harmonically. Thus particles execute
helical orbits.
For a finite length solenoid, this analysis is actually INCORRECT. Can you
think of a conservation law that it violates?
Hereafter in these lectures, the symbol m is always the rest mass, and
sometimes denoted as m0, but the subscript 0 is redundant, and I’ll often
forget it and drop it.
1 ~ to distinguish it from energy E .
Notation: electric field will be denoted E
E = γm0c2
For particles of charge q, this is in fact only correct in absence of electric field.
The canonical coordinate E is the total energy, while what appears in the
above equation is only the dynamical (kinetic) part. The electric potential Φ
This equation plus the fact that −E, Px, Py , Pz are canonical and conjugate to
t, x, y, z can be used to derive all of the relativistic dynamics of charged
particles.
R P
(Maybe you recall this as: δ ( i Piq˙i − H)dt.)
Look again at 7. It’s true even in Newtonian mechanics and was known far
before Einstein.
It must have been a puzzle why E and P~ played similar roles when the central
equation was 2
1 ~ ~ + qΦ.
E= P − q A (9)
2m
E and P in this non-relativistic limit look like completely different concepts. But
in relativity they’re just different components of the 4-vector.
H(x, Px, y, Py , z, Pz ; t) =
q
qΦ + m20c4 + c2(Px − qAx)2 + c2(Py − qAy )2 + c2(Pz − qAz )2 (10)
We can use the Principle of Least Action to derive the usual equations of
motion:
dx ∂H dPx ∂H
= , =− ,
dt ∂Px dt ∂x
dy ∂H dPy ∂H
= , =− ,
dt ∂Py dt ∂y
dz ∂H dPz ∂H
= , =− ,
dt ∂Pz dt ∂z
dE ∂H
= . (11)
dt ∂t
1 ~ 2
H= P − q ~
A + qΦ,
(13)
2m
Find
∂H ∂(H − qΦ)
(H − qΦ)vx = (H − qΦ) = (H − qΦ) =
∂Px ∂Px
1 ∂(H − qΦ)2 2 2
= = c (Px − qAx) = c px (15)
2 ∂Px
E−qΦ
But H is the energy E. So px = c2
vx and similar for py and pz , therefore
E − qΦ
p~ = ~v . (16)
c2
Compare: p~ = m~v .
P~ is the total (or canonical) momentum containing both the dynamic part and
the spatial (magnetic field) part; p~ = P~ − q A
~ is the “kinetic part” of the
momentum.
However, this last is not the fundamental equation. Equation 8 (or 14) is the
fundamental equation; it gives ALL the dynamics.
For a particle of charge q, this “energy” is q times the voltage, V0, needed to
bring the particle exactly to rest.
dr~6
∇6 · =0 (18)
dt
R. Baartman, TRIUMF 2015 23
So the flow of points in phase space is divergenceless and therefore the 6-D volume of a set
of points is conserved. This is Liouville’s theorem.
Stated in mathematical form, let f (x, Px, y, Py , z, Pz ) be the distribution function in phase
space, then
df
= 0. (19)
dt
The total derivative can be expanded:
df = ∂f
∂t dt + ∂f
∂x dx + ∂f
∂y dy + ∂f
∂z dz + ∂f
∂Px dP x + ∂f
∂Py dP y + ∂f
∂Pz dPz , which we can write more
compactly as (~ ~ = (Px, Py , Pz ))
r = (x, y, z), P
∂f ~
df = r + ∇P f · dP
dt + ∇f · d~
∂t
into which we can substitute the equations of motion 11, so finally the differential equation 19
is:
∂f ∂H ∂H
+ ∇f · − ∇P f · = 0. (20)
∂t ∂P~ ∂~r
This is in fact the collisionless Boltzmann equation.
where the closed path is arbitrary but moves with the motion of the phase
space points. This is called the Poincaré-Cartan integral invariant. Click here
for a proof. For the case of time-independent H, simply
I
P~ · d~q = constant.