The $\overline{\rm MS}$ mass is function of the renormalization scale $\mu$. What does it mean to choose this scale $\mu$ as the $\overline{\rm MS}$ mass itself? I am giving some details below to make my question more concrete.

In M. Schwarz's book on QFT, the author discusses the effect of a fermion loop on the scalar mass. In Sec. 22.6.1 [Pg. 408], he considers a theory with scalar $\phi$ with mass $m$ and a Dirac fermion $\psi$ with mass $M$. This has the Lagrangian,

$$\mathcal{L}=-\frac{1}{2}\phi(\Box+m^2)\phi+\lambda\phi\bar{\psi}\psi+\bar{\psi}(i\gamma^\mu \partial_\mu-M)\psi \tag{22.41}$$

He then performs the renormalization exercise to calculate the counterterms in the on-shell scheme. He finally arrives at the difference between the pole mass and the $\overline{\rm MS}$ mass,

$$m_P^2-m^2_{\overline{\rm MS}}(\mu)=\frac{\lambda}{24\pi^2}(6M^2-m_P^2)-\frac{3\lambda^2}{4\pi^2}\int_0^1 dx[M^2-m_P^2 x(1-x)]\ln\frac{M^2-m_P^2 x(1-x)}{\mu^2} \tag{22.53}$$

This result is $\mu$ dependent, as expected. My confusion is when he considers the $\phi$ to be the Higgs boson and $\psi$ to be the top quark. With the choice of the parameters ($\lambda$, top mass, and $m_P$), he calculates that,

$$m_P^2-m^2_{\overline{\rm MS}}(m_{\overline{\rm MS}})=(18.6\,\mathrm{GeV})^2 \tag{22.54}$$

My question: How is it possible to put $\mu=m_{\overline{\rm MS}}$ in eq. (22.53) when it is $m_{\overline{\rm MS}}$ that we are calculating? Is there something simple that I am missing?

Thanks for any help.


1 Answer 1


There is nothing weird going on. Let $\mu$ denote the scale, and $f=f(\mu)$ the MS mass. We have an expression of the form $$ f(\mu)= c\log \mu $$ for some constant $c$. We can choose the scale such that $f(\mu)=\mu$, which is around $\mu=-c W(-1/c)$, with $W$ the Lambert function. Your equation is a little bit more complicated, and so solving for $\mu$ explicitly is not as straightforward, but one can always do so, at least numerically. In any case, the point is that the MS mass is a function of $\mu$, and one can always choose $\mu$ so that the function is identical to $\mu$ itself — this is nothing but an equation for $\mu$, which is in principle solvable (although in practice it can become messy).

For example, if you have Mathematica, the following code'll do the trick:

λ = .5;
M = 10;
mp = 3;

eq[μ_?NumericQ] := mp^2 - μ^2 - (λ/(24 π^2) (6 M^2 - mp^2) - (3 λ^2)/(4 π^2) NIntegrate[(M^2 - mp^2 x (1 - x)) Log[(M^2 - mp^2 x (1 - x))/μ^2], {x, 0, 1}]);

NSolve[eq[μ], μ]
(* μ -> 3.42547 *)

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.