Teilchenphysik:

Lecture 7: Feynman calculus

Next step: decays and scattering in actual theories

But actual theories are too complicated.

We will set up technology with a toy example

I How do you write down the contents of a theory?
I How does that determine the way the constituents interact?
I What do real calculations of decays look like?
I What do real calculations of scattering look like?
 2: The big picture so far

Big questions of particle physics:

I What are the fundamental particles in nature?
Answered in the first lecture. We’ll come back...
I How do they interact with each other?
Which ones are stable?
How do the unstable ones decay?
How do they scatter off of each other?
We have seen how to help organize these questions in terms of symmetries,
and how to define and formulate decay and scattering.

Next: How do you really calculate any of this?

 3: Why not to start with real examples

Real particle physics is complicated because the particles (almost) all carry
nonvanishing spin. The ways the spins combine is complicated, with relativistic
generalizations of Clebsch-Gordan coefficients.

Physicists learn by figuring out one piece at a time.

Let’s first see how decays and scattering work for spin-0 particles,
and then worry about adding the complications of spin.

Therefore we will consider a Toy Model

 4: Our toy model

Following the book, consider a theory of three scalar fields A, B, C:

mA > mB + mC mB > mC

They interact via a three-field interaction.

All physics is determined by the action
in field theories the action is a spacetime integral of a Lagrange density L:
Z Z
S= dt L , L= d 3x L , for us, L = ... + gABC

with g some coupling constant

 5: Feynman rules

Feynman developed a simple graphical method to keep track of all the ways
particles can move and interact: the Feynman rules

Particles arrive initially

Particles interact at vertices

Particles propagate
from vertex to vertex

Particles fly out

 6: Feynman rules and Momentum

How does energy-momentum move?

Particles bring in 4-momentum

It is conserved at vertices

It is carried along propagators

Particles carry out 4-momentum

 7: Super simple example: A decay

The process A → BC is allowed. One leading diagram:

Feynman rules always give −i times (2π )4 δ 4 (pin − pout ) times Matrix Element

−ig(2π )4 δ 4 (pAµ − pBµ − pCµ ) = −i(2π )4 δ 4 (pAµ − pBµ − pCµ )M M=g

 8: A decay: calculation

Fermi’s Golden Rule:

d 3 pB d 3 pC
Z
1
ΓA = (2π )3 δ 3 (~pB + ~pC )(2π )δ (mA − EB − EC ) × g 2
2mA (2π )6 2EB 2EC

Here I separated the momentum and energy parts of the δ 4 (..).

Start with baby case: mB = 0 = mC (or mA  mB , mC )
R
Use delta-function to perform d 3~pC :

~pC = −~pB EB = |~pB | EC = |~pC | = EB

Simplifies result:

g2 pB2 dpB
Z Z
Γ= d ΩB × δ (mA − 2EB )
(2π )2 2mA (2pB )2
 9: Integral of a Delta Function

g2
Z  Z 
dpB
Γ= d ΩB δ (mA − 2EB )
8 π 2 mA 4
CAREFUL!
Z ∞ Z ∞ Z ∞
2dx dy 1
dx δ (2 − 2x) Name y = 2x and δ (2 − 2x) = δ (2 − y) =
0 0 2 0 2 2

Don’t forget constants, functional dependences inside delta functions.

Z X 1
In general: dx δ (f (x)) =
|f 0 (x i )|
xi :f (xi )=0

R
Meanwhile, d ΩB = 4π as usual.
 10: Decay rate: final result

For the case mA  mB , mC we find:

g2
Z  Z 
dpB
Γ= d ΩB δ (mA − 2pB )
8 π 2 mA 4
g2 1 g2
= × 4π × =
8 π 2 mA 8 16π mA
The case where mB , mC are not so small is more complex. We need

pB2 dpB
Z  q q 
p q δ mA − pB2 + mB2 − pB2 + mC2
2 pB2 + mB2 2 pB2 + mC2

After some work (see book) we find:

c |pB |g 2
q
|pB | = mA4 + mB4 + mC4 − 2mA2 mB2 − 2mA2 mC2 − 2mB2 mC2 , Γ=
2mA 8π mA2
 11: Is the B stable?

Suppose mB > 2mC . There is enough energy for B to decay.

Can B decay into some C particles? Try drawing a diagram:

Why can’t I do it? Discrete symmetry: A → −A and B → −B.

Total number of A plus B particles stays either even or odd.
12: Higher order corrections?

Multi-particle decays Loop effects

O(g 6 ) O(g 4 )
 13: Scattering

Consider first BC → BC. Two diagrams!

What should we do? Add them.

 14: Momenta

Let’s write out all the momenta for this process.

Use δ 4 (p1 +p2 −q) to set q = p1 +p2 Use δ 4 (p1 −p4 −q) to write q = p1 −p4
and use to rewrite δ 4 (q −p3 −p4 ) as and rewrite δ 4 (p2 −p3 +q) as
δ 4 (p1 +p2 −p3 −p4 ) δ 4 (p1 +p2 −p3 −p4 ).

i(−ig)2 i(−ig)2
M=i M=i
(p1 +p2 )2 − mA2 (p1 −p4 )2 − mA2
 15: Mandelstam Variables

Consider two incoming particles, (p1 , p2 ) and (m1 , m2 ) and two outgoing particles
(p3 , p4 ) and (m3 , m4 ). Frequently encounter

c 2 (p1 +p2 )2 ≡ s = c 2 (p3 +p4 )2

c 2 (p1 −p3 )2 ≡ t = c 2 (p2 −p4 )2
c 2 (p1 −p4 )2 ≡ u = c 2 (p2 −p3 )2

Not all independent: (p1 +p2 −p3 −p4 ) = 0 and p12 = c 2 m12 , p22 = c 2 m22 , p32 = c 2 m32 ,
p42 = c 2 m42

and therefore (after some work) s + t + u = c 4 (m12 + m22 + m32 + m42 )

For very relativistic systems t < 0 and u < 0, and

2 (1 − cos θ13 )s (1 + cos θ13 )s

s = Ecm (always), t =− , u=− .
2 2
16: Processes and Mandelstam

BC → BC BB → CC

   
1 1 1 1
g2 + g2 +
s − mA2 u − mA2 t − mA2 u − mA2
 17: Symmetry factors

Consider BB → CC. total cross-section:

2
d 3 p3 d 3 p4
Z 
1 4 4 4 1 1
σ= (2 π ) δ (p1 +p 2 −p3 −p4 ) g +
2E1 2E2 |v1 − v2 | (2π )6 2E1 2E2 t − mA2 u − mA2

Q: What p3 , p4 range should I integrate?

A: Only half! (p3 = x, p4 = y ) equivalent to (p4 = x, p3 = y ) and integrating over both
is double-counting physically distinct final states. Either
I Make sure to integrate over half of the final momentum range, leaving out all
“duplicate” integrations, or
I Integrate over everything but put in a “symmetry” factor 1
2
to correct for
double-counting.
If there were N final-state B particles, we would need a 1/N! factor or to integrate
over only 1/N! of the “naive” phase space.
 18: Doing the angular integral

Let’s compute total cross-section in s  mA2 limit in CM frame:

2
d 3 p3 d 3 p4
Z 
1 4 4 4 1 1
σ= (2π ) δ (p1 +p2 −p3 −p4 ) g +
2E1 2E2 |v1 − v2 | (2π )6 2E1 2E2 t − mA2 u − mA2

I |v1 − v2 | = 2 and 2E1 2E2 = s

I t = (1 − cos(θ13 ))s/2 and u = (1 + cos(θ13 ))s/2
I The δ 3 (~p1 + ~p2 − ~p3 − ~p4 ) = δ (~p3 + ~p4 ) forces ~p4 = −~p3 , E4 = E3 = E1 = Ecm /2.
I The |p3 | integral is just like in the 2-body decay.
2
g4 1
Z 2 Z 
p3 dp3 2 2
σ= δ (Ecm − 2p3 )2π sin(θ) d θ +
2s 4π 2 4p32 (1 − cos θ)s (1 + cos θ)s
Integral actually diverges as θ → 0 or θ → π , but if we include mA2 it stays finite.
 19: What about BB → BB?

This process actually requires a “box” diagram:

Cross-section is O(g 8 ) so for “small” g it is very suppressed!

20: Summary

I Total rates or widths are phase space times matrix element

I Feynman Rules exist to turn cartoons into determinations of the matrix
elements
I For a toy example, the decay Feynman rule is very simple
I Some processes are forbidden; others just arise at high order
I Two-body phase space is surprisingly simple
I Mandelstam variables are very useful in two-body scattering