Spherical Harmonics & d-Functions · Elementary Particle Physics

A detector never measures a spin. It measures an angular distribution, and these functions are what connect the two — which is why a spin assignment is always a fit and never a reading.

🎯 Why this matters

Because the distribution is a sum over

l

, a spin hypothesis predicts the shape and not the size. Normalisation divides straight out, so the measurement survives an unknown production rate, an unknown luminosity and an unknown efficiency at once.

Appendix 4 told you how angular momenta combine. Appendix 5 tells you what that looks like in a detector. These two families of functions are the bridge between a quantum number you cannot see and a histogram you can: give a state a spin, and these functions predict the angular distribution of what comes out of it. Fit the histogram, read off the spin.

That is not a side technique. It is how the book establishes the spin of the ρ, of the gluon, and of the Higgs boson.

📐 Physics you need first — why angles carry the spin

A decaying particle has no preferred direction of its own except the one its angular momentum defines. So when it decays, the directions in which the products fly are not uniform: the probability depends on the angle to that axis, and the dependence is fixed entirely by the spin — not by the force doing the decaying.

Concretely: quantum mechanics writes the amplitude to find a decay product at angle θ as one of these functions. Squaring it gives a probability per unit solid angle, which is exactly what an experiment measures by counting events in bins of $\cos\theta$ . Different spins give visibly different shapes, so the histogram is a spin-meter.

You need no quantum mechanics beyond this: the functions below are the shapes that different spins produce.

Spherical harmonics: the basis for anything on a sphere

\htmlClass{t-Y}{Y_l^m(\theta,\phi)} = \sqrt{\frac{(2l+1)\,\htmlClass{t-fac}{(l-m)!}}{4\pi\,\htmlClass{t-fac}{(l+m)!}}}\; \htmlClass{t-P}{P_l^m(\cos\theta)}\; \htmlClass{t-phase}{e^{im\phi}}

(R.5)

Appendix 5, p. 502. The prefactor is pure normalisation; all the shape is in P.

Every symbol, one at a time

Hover or tap a symbol above — it lights up in the equation and its meaning, units and type appear here.

💡 What this really says — a Fourier basis for the surface of a sphere

This is a Fourier basis for the surface of a sphere. On a circle you expand in

e^{im\phi}

; on a sphere you expand in

Y_l^m

. The index

m

is literally the same azimuthal Fourier index, and

l

counts how much structure the function has in the polar direction. Any angular distribution whatsoever can be written as a sum of these, and the coefficients are the “spectrum”.

lmparity − · 1 polar node · 3 states in this multiplet

$Y_1^0 = \sqrt{\tfrac{3}{4\pi}}\,\cos\theta$

r = |Y|², rotated about z. blue = positive amplitude, red = negative.

|Y|² (what you measure)
amplitude Y (sign matters for interference)

A detector counts events per solid angle, so it measures the solid curve. The dashed one is the amplitude, whose sign only shows up through interference with something else.

The book prints the first six explicitly, and the viewer above shows each one as you select it. Two features are worth naming, because Chapter 3 uses both:

Parity. Under $\vec r \to -\vec r$ (i.e. $\theta \to \pi - \theta$ , $\phi \to \phi + \pi$ ), $Y_l^m \to (-1)^l Y_l^m$ . So a state of orbital angular momentum $l$ carries parity $(-1)^l$ — that single fact is what lets §3.2 assign parities to hadrons at all.
Nodes. $|Y_l^m|^2$ vanishes at $l - |m|$ polar angles. A zero in an angular distribution is a direction in which the decay is forbidden, not merely rare, and finding one is strong evidence about the spin.

⚙️ Engineer’s bridge

$Y_l^m$ is the spherical harmonic transform, the sphere’s DFT. Same idea: an orthonormal basis, coefficients by projection, Parseval for the power.
$l$ behaves like a band index: higher $l$ means finer angular structure, exactly as higher frequency means finer time structure. A detector with limited angular resolution is a low-pass filter in $l$ .
$|Y|^2$ is a power spectrum: the phase $e^{im\phi}$ is discarded when you square, which is why measuring $|Y|^2$ alone loses the relative phase — and why interference between two amplitudes is the only way to get it back. That is the same reason a magnitude spectrum cannot reconstruct a signal.

Where it breaks: unlike a DFT basis you chose for convenience, this one is selected by rotational symmetry. You cannot pick a different basis and get equally simple physics, because the $Y_l^m$ of fixed $l$ are exactly the states that rotations mix among themselves and nothing else — the same irreducibility argument as the Clebsch–Gordan blocks.

d-functions: what a rotation does to a spin state

A Wigner d-function $d^{\,j}_{m'm}(\theta)$ is the amplitude that a state with angular-momentum projection $m$ along one axis is found with projection $m'$ along an axis tilted by $\theta$ . That is all it is — a rotation matrix element, written out.

Its importance is practical: in a two-body decay, the angle between the parent’s spin axis and the daughter’s direction is exactly such a tilt, so the decay amplitude is a d-function.

A parent of spin J decaying to two spin-0 particles emits them with the distribution |d^J₀₀(θ)|². The three curves are flatly different, so a histogram of cos θ measures the spin — no dynamics required.

J = 0: |d^0₀₀|²
J = 1: |d^1₀₀|²
J = 2: |d^2₀₀|²

J = 0 is flat; J = 1 is cos²θ and vanishes at 90°; J = 2 is ((3cos²θ − 1)/2)² with zeros at 54.7° and 125.3°. Count events, fit, read off the spin.

🔢 Worked example — reading a spin off a histogram

A neutral particle

X

decays to two spin-0 particles. Bin the events in

\cos\theta

, where θ is measured in the

X

rest frame between one daughter and the beam. The prediction is

|d^{\,J}_{00}(\theta)|^2

J = 0:\ \text{flat} \qquad J = 1:\ \cos^2\theta \qquad J = 2:\ \left(\tfrac{3\cos^2\theta - 1}{2}\right)^2

$J = 1$ vanishes at 90°. If your histogram has events at 90°, the parent is not spin 1 — one bin can kill a hypothesis.
$J = 2$ vanishes at $\theta$ where $\cos^2\theta = 1/3$ , i.e. 54.7° and 125.3°. Two symmetric dips are a spin-2 fingerprint.
$J = 0$ is featureless, and featurelessness is itself a measurement.

This is precisely the argument §9.18 uses to establish that the Higgs boson is spin 0 and not spin 2, and the one §6.1 uses to show that the gluon is a vector.

Switch the widget to any d-function to see the general case, including the half-integer ones the book prints for $j = 1/2$ .

jm′m

$d^{1}_{1,0} = -\tfrac{\sin\theta}{\sqrt{2}}$

d^1_{1,0}(θ)
|d|² — the angular distribution

The amplitude can go negative; the distribution you count cannot. Zeros of d are directions in which the decay is strictly forbidden — a selection rule you can see.

An erratum on p. 502 — and how to catch it

Appendix 5 opens the d-function section with a symmetry relation. As printed it reads

d^{\,j}_{m,m'}(\theta) = (-1)^{m-m'}\, d^{\,j}_{m,m'}(\theta) = d^{\,j}_{-m,-m'}(\theta),

and the middle expression carries the same indices as the left. Take the book’s own value from two lines further down, $d^1_{1,0} = -\sin\theta/\sqrt2$ , and substitute: the relation demands $d^1_{1,0} = (-1)^{1}\,d^1_{1,0}$ , i.e. $d^1_{1,0} = 0$ . It is not zero. The identity as printed is self-contradictory.

The standard relation has the indices swapped on one side:

d^{\,j}_{m',m}(\theta) = (-1)^{m-m'}\, d^{\,j}_{m,m'}(\theta) = d^{\,j}_{-m,-m'}(\theta),

and with that swap every one of the book’s six listed values is consistent. The snippet below shows both the contradiction and the fix.

Reproduce it

import numpy as np
from math import factorial, degrees, acos

def legendre_P(l, m, x):                     # Condon-Shortley phase included
    pmm, f = 1.0, 1.0
    for _ in range(m):
        pmm *= -f * np.sqrt(max(0.0, 1 - x*x)); f += 2
    if l == m: return pmm
    pmmp1 = x * (2*m + 1) * pmm
    if l == m + 1: return pmmp1
    for ll in range(m + 2, l + 1):
        pmm, pmmp1 = pmmp1, (x*(2*ll - 1)*pmmp1 - (ll + m - 1)*pmm) / (ll - m)
    return pmmp1

def Y(l, m, th):                             # real amplitude; e^{im phi} is a pure phase
    N = np.sqrt((2*l + 1)*factorial(l - abs(m)) / (4*np.pi*factorial(l + abs(m))))
    return N * legendre_P(l, abs(m), np.cos(th))

def d(j, mp, m, b):                          # Wigner small-d, d^j_{m',m}
    s, kmin, kmax = 0.0, int(max(0, m - mp)), int(min(j + m, j - mp))
    for k in range(kmin, kmax + 1):
        num = (-1)**(k - m + mp) * np.sqrt(factorial(int(j+m))*factorial(int(j-m))
                                           *factorial(int(j+mp))*factorial(int(j-mp)))
        den = factorial(int(j+m-k))*factorial(k)*factorial(int(j-k-mp))*factorial(int(k-m+mp))
        s += num/den * np.cos(b/2)**(2*j - 2*k + m - mp) * np.sin(b/2)**(2*k - m + mp)
    return s

th = 0.7
book = {(0,0): np.sqrt(1/(4*np.pi)),
        (1,0): np.sqrt(3/(4*np.pi))*np.cos(th),
        (1,1): -np.sqrt(3/(8*np.pi))*np.sin(th),
        (2,0): np.sqrt(5/(4*np.pi))*(1.5*np.cos(th)**2 - 0.5),
        (2,1): -np.sqrt(15/(8*np.pi))*np.sin(th)*np.cos(th),
        (2,2): 0.25*np.sqrt(15/(2*np.pi))*np.sin(th)**2}
ok = sum(abs(Y(l,m,th) - v) < 1e-12 for (l,m), v in book.items())
print(f"spherical harmonics: {ok}/6 match Appendix 5 exactly")

lhs, rhs = d(1,1,0,th), (-1)**(1-0) * d(1,1,0,th)
print(f"as printed:  d1_10 = {lhs:+.6f}  vs  (-1)^(m-m') d1_10 = {rhs:+.6f}  -> contradiction")
print(f"with the indices swapped: d1_01 = {d(1,0,1,th):+.6f}  = {rhs:+.6f}  -> holds")

z = degrees(acos(np.sqrt(1/3)))
print(f"J=2 distribution |d2_00|^2 vanishes at theta = {z:.1f} deg and {180-z:.1f} deg")

prints

spherical harmonics: 6/6 match Appendix 5 exactly
as printed:  d1_10 = -0.455531  vs  (-1)^(m-m') d1_10 = +0.455531  -> contradiction
with the indices swapped: d1_01 = +0.455531  = +0.455531  -> holds
J=2 distribution |d2_00|^2 vanishes at theta = 54.7 deg and 125.3 deg

⚠️ Three errata found so far

This is the third slip caught by checking the book’s data against itself, after

(\hbar c)^2

on p. 495 and the Ω_b⁻ isospin on p. 500. None of them changes any physics, and none of them is a reason to distrust the book — they are a reason to run the invariants. A table of numbers is an input like any other.

🔑 If you remember only three things

$l$ is not the spin. It is the orbital index of the expansion, and the two coincide only when the daughters are spinless.
The d-functions rotate the state, not the axes. Take the rotation the wrong way round and the interference terms change sign while the diagonal ones do not.
Isotropy is what spin 0 predicts. A flat fit is therefore a measurement in its own right, which is the opposite of how a flat plot reads almost anywhere else.

Where this is used

§3.2 Parity uses $(-1)^l$ to assign parities.
§4.4 Spin, parity and isospin of three-pion systems fits angular distributions exactly like the ones above.
§6.1 determines the gluon spin from the angle between jets; §9.18 does the same for the Higgs boson.
Appendix 4 supplies the coefficients that say which d-functions appear in a given decay.

✅ Check yourself — angular distributions

0/5 answered · 0 correct

1.A neutral particle decays to two spin-0 daughters. Your $\cos\theta$ histogram is flat within errors, with plenty of events at 90°. What can you conclude?
2.Why does the azimuthal factor $e^{im\phi}$ never show up in a measured angular distribution?
3.The site claims the symmetry relation printed on p. 502 is wrong. What is the argument?
4.Set the spherical-harmonic viewer to $l = 2$ . Which statements are true?
5.In what sense is a detector with poor angular resolution acting as a filter?