I am reading this paper (or this link), and I'm troubled by the custodial transformation. So, I will use the notations and equation labels appearing in this paper.

If we write the Higgs field into the four real components $$ \Phi=\left(\begin{array}{c} -\operatorname{Re} \phi_1 \\ \operatorname{Im} \phi_1 \\ \operatorname{Re} \phi_2 \\ \operatorname{Im} \phi_2 \end{array}\right) \tag{13} $$ where $\phi_1$ and $\phi_2$ are the complex components of $\Phi$ in the ordinary isospinor representation of the field, $$ \left(\begin{array}{l} \phi_1 \\ \phi_2 \end{array}\right) . $$ The Higgs field potential under (13) has the $SO(4)$ symmetry, since $$ \mathrm{SO}(4) \sim \mathrm{SU}(2)_{\text {isospin }} \times \mathrm{SU}(2)_{\text {custodial }}. $$

We denote the isospin generator with $\mathbf{I}$, the custodial generator with $\mathbf{K}$. Then the isospin and custodial transformation on the four vector $\Phi$ are $$ \Phi_i \rightarrow e^{i \epsilon \cdot \mathbf{I}} \Phi_i=\left(e^{i \epsilon \cdot \mathbf{T}}\right)_{i j} \Phi_j \\ \Phi_i \rightarrow e^{i \epsilon' \cdot \mathbf{K}} \Phi_i=\left(e^{i \epsilon' \cdot \mathbf{T}^{\prime}}\right)_{i j} \Phi_j. $$

So, now our goal is to calculate the 4 by 4 matrices $\mathbf{T}$ and $\mathbf{T^\prime}$. The calculation can be done by recasting the transformation law of $\Phi$ in isospinor form. Where, for the isospin transformation, $$ \left(\begin{array}{l} \delta \phi_1 \\ \delta \phi_2 \end{array}\right)=i \epsilon \cdot \tau\left(\begin{array}{l} \phi_1 \\ \phi_2 \end{array}\right) $$ while, for the custodial transformation, $$ \left(\begin{array}{l} \delta \phi_1 \\ \delta \phi_2^* \end{array}\right)=i \epsilon' \cdot \tau\left(\begin{array}{l} \phi_1 \\ \phi_2^* \end{array}\right) \tag{*} . $$

My question is for the $(*)$ equation. Why does the custodial transformation in isospinor format take this form?

  • 1
    $\begingroup$ Minor comment to the post (v2): Please consider to mention explicitly author, title, etc. of link, so it is possible to reconstruct link in case of link rot. $\endgroup$
    – Qmechanic
    Apr 11, 2023 at 8:53
  • $\begingroup$ Link rot preventative: doi, Carson et al. $\endgroup$ Apr 11, 2023 at 15:05

1 Answer 1


It should be a straight translation from this link answer, but I strongly suspect Carson et al., 1990 flipped a sign, at a deep level in their transition from a complex doublet to the real quartet representation they actually use. (They don't actually use this doublet/matrix representation in practice, and are merely meaning to connect to the readers' shared educational background, adjusting for their errant sign.)

I can explain your (*) equation, provided I insert a minus sign in front of $𝜙_1$ in it. You first observe the action of a weak isospin transformation with parameter $\vec \alpha$ on the complex doublet and its conjugate rep, $$ \phi = \begin{pmatrix} \phi_1 \\ \phi_2 \end{pmatrix} \qquad \phi \mapsto e^{i\vec{\alpha}\cdot \vec{\tau}/2} \phi \\ \tilde \phi =i\tau_2 \phi^*= \begin{pmatrix} \phi^{*}_2 \\ -\phi_1^* \end{pmatrix} \qquad \tilde \phi \mapsto e^{i \vec{\alpha}\cdot \vec{\tau}/2}\tilde \phi ~.$$

Thus, the 2×2 Higgs matrix is defined as a side-by-side juxtaposition of these two left-doublets serving as columns, $$ H\equiv \sqrt{2}(\tilde\phi, \phi)= \sqrt {2} \begin{pmatrix} \phi_2^{*} &\phi_1 \\ -\phi_1^* & \phi_2 \end{pmatrix}. $$

It is then evident that it transforms by left α isospin and also by manifestly commuting right β custodial isorotations K as $$ \bbox[yellow]{ e^{i\vec{\alpha}\cdot \vec{\tau}/2} \sqrt{2}(\tilde\phi , \phi )e^{i\vec{\beta}\cdot \vec{\tau}/2} = e^{i\vec{\alpha} \cdot \vec{\tau}/2}\sqrt {2} \begin{pmatrix} \phi_2^* &\phi_1 \\ -\phi_1^* & \phi_2 \end{pmatrix}e^{i\vec{\beta}\cdot \vec{\tau}/2} = e^{i\vec{\alpha}\cdot \vec{\tau}/2} H e^{i\vec{\beta}\cdot \vec{\tau}/2} ~ . } $$

Taking the adjoint of this equation and looking at the custodial action on the second column of $H^\dagger$, we see that $$ \begin{pmatrix} -\phi_1 \\ \phi_2^* \end{pmatrix} \qquad \mapsto \qquad e^{-i\vec{\beta}\cdot \vec{\tau}/2} \begin{pmatrix} -\phi_1 \\ \phi_2^* \end{pmatrix} \tag{*} $$ with the sign mismatch mentioned.

As indicated, it is trivial to demonstrate the action of I commutes with the action of K, so these two groups do not "know" about each other, i.e. they are in direct product, hence SO(4).

You may also trivially confirm that $$ \operatorname{Tr} H^\dagger H = 4 \phi^\dagger \phi = 4 \Phi^T \Phi, $$ invariant under $\mathrm{SU}(2)_{\text {isospin }} \times \mathrm{SU}(2)_{\text {custodial }}$.

NB Note added

I mapped the custodial $\tau_2 \to T'_2$ on p 2132, and matched correctly my (*), not theirs, on the same page, to their quartet. (Perhaps they meant to have $\small \begin{pmatrix} \phi_2 \\ \phi_1^* \end{pmatrix} \mapsto e^{-i\vec{\beta}\cdot \vec{\tau}/2} \begin{pmatrix} \phi_2 \\ \phi_1^* \end{pmatrix} $, instead?)

  • $\begingroup$ Hi professor, thank you very much! May I ask a related question in Carson et al.'s 1990 paper? In the third equation below equation (13), they say that the isospin transformation for the gauge fields is $\mathbf{A}^a \rightarrow e^{i \epsilon \cdot \mathbf{I}} \mathbf{A}^a=\left(e^{i \epsilon \cdot \Delta}\right)^{a b} \mathbf{A}^b$, where $(\Delta^a)^{bc}=-i\epsilon^{abc}$. I am troubled for this form. In gauge theory, we know the gauge field transforms like $A_\mu^a T^a\rightarrow U (A_\mu^a T^a)U^\dagger+ \frac{i}{g}U(\partial_\mu U^\dagger)$. And it seems that these two things don't match. $\endgroup$
    – Daren
    Apr 13, 2023 at 2:14
  • $\begingroup$ No hurry, thanks! $\endgroup$
    – Daren
    Apr 13, 2023 at 2:24
  • 1
    $\begingroup$ As stated at the beginning of sec II, they have fixed the gauge, and monitor the residual global symmetry. $\endgroup$ Apr 13, 2023 at 11:34
  • $\begingroup$ Ah, thanks! Clear! $\endgroup$
    – Daren
    Apr 15, 2023 at 9:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.