Isospin and kaon charge-conjugation conventions

Goal

This note fixes a sign convention for isospin doublets and charge conjugation in amplitudes with pions and kaons, with the target application

\[ \chi_{c0},\chi_{c2} \to \pi\pi K\bar K .\]

The initial charmonium states have

\[ I=0, \qquad C=+, \qquad J^{PC}=0^{++},2^{++} .\]

The important point is that Clebsch–Gordan coefficients fix only the isospin recoupling signs. The charge-conjugation signs of meson pairs must be fixed from the meson currents.

Regeneration note: this tutorial is checked in as generated artifacts and is not rendered by docs/make.jl in CI. After editing this notebook, run

quarto render docs/isospin-kaon-conventions.qmd --to gfm

and commit the updated docs/generated/isospin-kaon-conventions.md file.

Isospin doublets

Use the light-quark doublet

\[ q= \begin{pmatrix} u\\ d \end{pmatrix},\]

with the SU(2)-conjugate doublet convention

\[ \widehat C q = -i\sigma_2 q^* = \begin{pmatrix} -\tilde d\\ \tilde u \end{pmatrix} .\]

Here the tilde denotes the isospin-conjugate antiquark object. This convention is the source of the relative minus sign in antikaon doublets.

For pseudoscalar kaons,

\[ K= \begin{pmatrix} K^+\\ K^0 \end{pmatrix}, \qquad K^+=u\bar s, \qquad K^0=d\bar s .\]

The corresponding antikaon doublet in the same convention is

\[ \bar K= \begin{pmatrix} -\bar K^0\\ K^- \end{pmatrix}, \qquad \bar K^0=s\bar d, \qquad K^-=s\bar u .\]

For vector kaons,

\[ K^*=\begin{pmatrix} K^{*+}\\ K^{*0} \end{pmatrix}, \qquad \bar K^*=\begin{pmatrix} -\bar K^{*0}\\ K^{*-} \end{pmatrix} .\]

These are isospin conventions. They do not yet determine the physical charge-conjugation sign of the meson state.

Charge conjugation must be applied to currents

A kaon is not just a light-quark doublet. It is represented by a quark bilinear. Therefore the safe operation is

\[ \text{quark-field convention} \longrightarrow \text{meson current} \longrightarrow \text{meson-state phase}.\]

For a bilinear current

\[ J_\Gamma=\bar q_1\Gamma q_2,\]

charge conjugation gives

\[ C\,J_\Gamma\,C^{-1} = \eta_\Gamma\,\bar q_2\Gamma q_1,\]

where

\[ \mathcal C^{-1}\Gamma\mathcal C = \eta_\Gamma\,\Gamma^T .\]

The most useful cases are

\[ \begin{array}{c|c|c} \text{state type} & \Gamma & \eta_\Gamma \\ \hline 0^- & i\gamma_5 & +1 \\ 0^+ & 1 & +1 \\ 1^- & \gamma_\rho & -1 \\ 1^{++}\text{-type axial} & \gamma_\rho\gamma_5 & +1 \\ 1^{+-}\text{-type axial} & \sigma_{\rho\sigma}\gamma_5\;\text{or equivalent} & -1 \\ \end{array}\]

The sign for the axial vector current is important. It is not caused by the absence or presence of an explicit factor of $i$. Charge conjugation is unitary, so it does not complex-conjugate $i$. The sign follows from the Dirac matrix identity.

For the vector current,

\[ \mathcal C^{-1}\gamma_\rho\mathcal C=-\gamma_\rho^T,\]

hence

\[ \eta_{\gamma_\rho}=-1.\]

For the axial-vector current,

\[ \mathcal C^{-1}(\gamma_\rho\gamma_5)\mathcal C = (-\gamma_\rho^T)(+\gamma_5^T) =-\gamma_\rho^T\gamma_5^T .\]

But

\[ (\gamma_\rho\gamma_5)^T = \gamma_5^T\gamma_\rho^T =-\gamma_\rho^T\gamma_5^T,\]

because $\{\gamma_\rho,\gamma_5\}=0$. Therefore

\[ \mathcal C^{-1}(\gamma_\rho\gamma_5)\mathcal C =(\gamma_\rho\gamma_5)^T, \qquad \eta_{\gamma_\rho\gamma_5}=+1 .\]

So, in the above isospin convention,

\[ C K = +\bar K, \qquad C K^* = -\bar K^* .\]

This is the origin of the common molecular convention

\[ |K\bar K^*;C=+\rangle = \frac{1}{\sqrt2} \left( |K\bar K^*\rangle - |K^*\bar K\rangle \right),\]

and

\[ |K\bar K^*;C=-\rangle = \frac{1}{\sqrt2} \left( |K\bar K^*\rangle + |K^*\bar K\rangle \right).\]

The relative minus in the $C=+$ molecular state is not a Clebsch sign. It is the vector-current charge-conjugation sign.

More generally, if $X$ is an excited kaon created by $\bar s\Gamma_X q$, define

\[ C X = \eta_X \bar X .\]

Then a two-channel kaonic system satisfies

\[ C|K\bar X\rangle = \eta_X |X\bar K\rangle, \qquad C|X\bar K\rangle = \eta_X |K\bar X\rangle,\]

up to the fixed isospin-doublet phases. The $C$-projected combination is therefore

\[ |K\bar X;C=\lambda\rangle = \frac{1}{\sqrt2} \left( |K\bar X\rangle +\lambda\eta_X |X\bar K\rangle \right), \qquad \lambda=\pm1 .\]

For $X=K^*$, $\eta_X=-1$, hence the $C=+$ combination has a minus sign. For an axial kaon of $1^{++}$-type, $\eta_X=+1$, so the $C=+$ combination has a plus sign. For an axial kaon of $1^{+-}$-type, $\eta_X=-1$, so the $C=+$ combination has a minus sign.

Axial kaons and the two $1^+$ nonets

The quark model contains two different $1^+$ structures:

\[ {}^3P_1: \quad J^{PC}=1^{++},\]

and

\[ {}^1P_1: \quad J^{PC}=1^{+-}.\]

They have different charge-conjugation signs in the non-strange neutral sector,

\[ \eta_C({}^3P_1)=+1, \qquad \eta_C({}^1P_1)=-1.\]

For isovector states one usually quotes $G$-parity,

\[ G=C\,e^{i\pi I_2}.\]

With the usual convention, for $I=1$,

\[ G=-C.\]

Thus

\[ a_1(1^{++}):\quad G=-, \qquad b_1(1^{+-}):\quad G=+ .\]

The strange axial kaons do not have a definite $C$-parity by themselves, but their signs in a $K\bar K_1\pm K_1\bar K$ construction must be inherited from the underlying nonet/current basis. In practice:

\[ K_{1A}\sim {}^3P_1 \quad\Rightarrow\quad \eta_{K_{1A}}=+1,\]

whereas

\[ K_{1B}\sim {}^1P_1 \quad\Rightarrow\quad \eta_{K_{1B}}=-1.\]

Physical $K_1(1270)$ and $K_1(1400)$ are mixtures of $K_{1A}$ and $K_{1B}$. Therefore one should assign the charge-conjugation sign in the $(K_{1A},K_{1B})$ basis first, then rotate to the physical $K_1$ states. Assigning a single memory-based sign directly to $K_1(1270)$ or $K_1(1400)$ is unsafe.

$K\bar K$ isospin states

Using

\[ K= \begin{pmatrix}K^+\\K^0\end{pmatrix}, \qquad \bar K= \begin{pmatrix}-\bar K^0\\K^-\end{pmatrix},\]

ordinary Clebsch–Gordan coefficients give

\[ |K\bar K;I=1,M=0\rangle = \frac{1}{\sqrt2} \left( -|K^+K^-\rangle +|K^0\bar K^0\rangle \right),\]

and

\[ |K\bar K;I=0,M=0\rangle = -\frac{1}{\sqrt2} \left( |K^+K^-\rangle +|K^0\bar K^0\rangle \right).\]

Thus

\[ |K^+K^-\rangle =-\frac{1}{\sqrt2} \left( |I=1,M=0\rangle +|I=0,M=0\rangle \right),\]

and

\[ |K^0\bar K^0\rangle =\frac{1}{\sqrt2} \left( |I=1,M=0\rangle -|I=0,M=0\rangle \right).\]

Three- and four-body construction

For pions, use the isovector convention

\[ \pi^+ = |1,+1\rangle, \qquad \pi^0 = |1,0\rangle, \qquad \pi^- = |1,-1\rangle .\]

For a three-body state such as $\pi K\bar K$, couple first the kaon pair,

\[ \left|\left(K\bar K\right)I_{K\bar K}M_{K\bar K};\pi\right\rangle,\]

then couple to the pion:

\[ \left|IM;I_{K\bar K}\right\rangle = \sum_{\mu_\pi,\mu_K,\mu_{\bar K}} \langle \tfrac12\mu_K,\tfrac12\mu_{\bar K}|I_{K\bar K}M_{K\bar K}\rangle \langle I_{K\bar K}M_{K\bar K},1\mu_\pi|IM\rangle |K_{\mu_K}\bar K_{\mu_{\bar K}}\pi_{\mu_\pi}\rangle .\]

Here $\bar K$ means the isospin-conjugate antikaon doublet, so the physical $K^-$ already carries the convention-dependent sign through

\[ \bar K_{-1/2}=K^- .\]

For $\pi\pi K\bar K$, useful couplings are

\[ (\pi\pi)_{I_{\pi\pi}}(K\bar K)_{I_{K\bar K}} \to I=0,\]

or sequential chains such as

\[ \pi\,(K\bar K^*)_{I,C}, \qquad K\,(\pi\bar K)_{I}, \qquad \bar K\,(\pi K)_{I} .\]

The first type needs both isospin and charge-conjugation bookkeeping if the kaon system contains excited kaons.

Constraints for $\chi_{c0}$ and $\chi_{c2}$

For annihilation decays of

\[ \chi_{c0},\chi_{c2}: \qquad I^GJ^{PC}=0^+0^{++},\;0^+2^{++},\]

one should enforce on the full final state

\[ I_{\rm tot}=0, \qquad C_{\rm tot}=+ .\]

For topologies of the form

\[ (\pi\pi)(K\bar K),\]

this is mostly an isospin-recouping problem. The $K\bar K$ signs are fixed by the antikaon doublet convention above.

For topologies with excited kaons, for example

\[ K\bar K^*, \qquad K^*\bar K, \qquad K\bar K_0^*, \qquad K_2^*\bar K,\]

one must additionally apply charge conjugation at the level of the corresponding meson current. The rule is

\[ C\left(\bar s\Gamma q\right)C^{-1} = \eta_\Gamma\,\bar q\Gamma s .\]

The value of $\eta_\Gamma$ depends on the spin-parity structure of the kaon resonance:

\[ \begin{array}{c|c|c} \text{current type} & \Gamma & \eta_\Gamma \\ \hline 0^- & i\gamma_5 & +1 \\ 1^- & \gamma_\rho & -1 \\ 0^+ & 1 & +1 \\ 1^{++}\text{-type axial} & \gamma_\rho\gamma_5 & +1 \\ 1^{+-}\text{-type axial} & \sigma_{\rho\sigma}\gamma_5\;\text{or equivalent} & -1 \\ 2^+ & \text{derivative tensor current} & \text{fix from chosen current} \end{array}\]

The last line is intentionally not compressed into a memory rule: for tensor or derivative currents, the sign should be derived from the explicit interpolating current used in the amplitude convention.

Practical implementation recipe

1. Fix the light-quark convention:

\[ (u,d) \mapsto (-\tilde d,\tilde u).\]

1. Define all kaon and antikaon doublets using this convention:

\[ K=(K^+,K^0)^T, \qquad \bar K=(-\bar K^0,K^-)^T .\]

1. Use ordinary Clebsch–Gordan coefficients for all isospin coupling.

2. Never use Clebsches to infer the charge-conjugation sign of a $K\bar K^*$-type pair.

3. For each excited kaon, define the meson current $\bar s\Gamma q$ and evaluate

\[ C(\bar s\Gamma q)C^{-1}.\]

1. Build $C=+$ and $C=-$ combinations only after the current-level sign is known.

2. For $\chi_{c0}/\chi_{c2}\to\pi\pi K\bar K$, keep only full amplitudes with

\[ I_{\rm tot}=0, \qquad C_{\rm tot}=+.\]

Minimal checks

The convention gives

\[ |K\bar K;I=0\rangle =-\frac{1}{\sqrt2} \left(|K^+K^-\rangle+|K^0\bar K^0\rangle\right),\]

and

\[ |K\bar K^*;C=+\rangle =\frac{1}{\sqrt2} \left(|K\bar K^*\rangle-|K^*\bar K\rangle\right).\]

These two signs have different origins:

• the first is an SU(2) doublet convention;
• the second is the vector-current sign under charge conjugation.

Keeping this separation avoids sign mistakes in $\pi\pi K\bar K$ amplitude models.