In Peskin and Schroeder from what I can tell there are four different forms the coupling constant can take:

  • $\lambda_0$: This is the bare coupling and that which appears in the Lagrangian e.g. for $\phi^4$: $$\mathcal{L}=\frac{1}{2}\partial_\mu\phi_0\partial^\mu \phi_0-\frac{m^2}{2}\phi^2_0-\frac{\lambda_0}{4!} \phi_0^4$$
  • $\lambda$: This is the renormalized coupling which can be found via e.g. minimal subtraction.

  • $\lambda'$: This is the effective coupling which appears in the effective Lagrangian of Wilson's Approach to renormalization and takes the form: $$\lambda'=(\lambda+\Delta \lambda)(1+\Delta Z)^{-2} b^{d-4}$$

  • $\bar \lambda$: This is the running coupling constant which appears in the renormalization group equation: $$\frac{d}{d \log(p/M)}\bar \lambda(p;\lambda)=\beta(\bar \lambda)$$

I am looking at how all of these are related. In most cases it is obvious but I am unsure about how $\lambda'$ and $\lambda$ are related to $\bar \lambda$. My question is therefore; how are they related?


1 Answer 1


We start with a Lagrangian full of bare couplings and fields $\{Z_B,g_B,m_B,...\}$. Then we regularize the theory by a hard cut-off $\Lambda $ or by dimensional reg $d=4-\varepsilon$ to tame the infinities. The amplitudes can then be formally calculated in perturbative expansion on the bare coupling $g_B$: $$ \mathcal{M}(s_0,t_0,u_0)=\sum_{n=1}^{\infty}A_n(Z_B,m_B,...,\Lambda\, or\,\varepsilon)(g_{B})^{n} $$ Each term $A_n(Z_B,m_B,...\Lambda\, or\,\varepsilon)$ usually diverge as we make $\Lambda\rightarrow\infty$ or $\varepsilon\rightarrow 0$, so this can be viewed only as formal expression. The idea of renormalizability is to absorb these divergence doing bare couplings $\rightarrow$ renormalized couplings.

Now, that the infinities are tamed as long as we hold $\varepsilon\neq0$ or $\Lambda\neq\infty$, we can do the bare couplings $\rightarrow$ renormalized couplings (i.e. fixing a renormalization condition). $$ \mathcal{M}(s_0,t_0,u_0)=g_0 $$ for a given renormalization scale $\mu_0$ contained in $(s_0,t_0,u_0)$. Together with others renormalization conditions we obtain a set of renormalized couplings $\{Z_0,g_0,m_0,...\}$. Note that this depends on the regularization procedure and the scale $\mu_0$. This renormalized couplings are finite and can be obtained directly or indirectly by experimental results.

The running coupling is the $\{Z(\mu),g(\mu),m(\mu)\}$, and can be obtained by perturbative calculation, expanding the beta functions in the renormalized couplings $\{Z_0,g_0,m_0,...\}$. Note that the running of the couplings are telling you how the renormalization conditions should change if you change the scale by $\mu_0\rightarrow\mu$ in order to preserve all the physical quantities (preserve the predictions of the theory).

Now, using the renormalization condition $\mathcal{M}(s_0,t_0,u_0)=g_0$ and the formal expansion of $\mathcal{M}(s_0,t_0,u_0)$ in terms of the bare couplings you can relate the bare and the renormalized couplings. Knowing the beta-functions you can relate the renormalized couplings of two diferent scales $\mu_0$ and $\mu$ of the same theory.

So, all this relate the bare couplings, the renormalized couplings and the running couplings.

Now, the Wilson approach is somehow different. We start with a theory that is already regulated by a cut-off $\Lambda_0$ and the couplings can be viewed as functions of the cut-off $\{Z(\Lambda_0),g(\Lambda_0),m(\Lambda_0),...\}$. As we integrate over modes, changing $\Lambda_0\rightarrow\Lambda$ we obtain an effective Lagrangian with effective couplings $\{Z(\Lambda),g(\Lambda),m(\Lambda),...\}$.

Note that now there is no renormalization condition and no renormalization scale $\mu_0$. The couplings are running with the scale of the cut-off regularization instead of the renormalization scale. The relation between these two approach are only achieved when we do the limit $\Lambda \rightarrow \infty$ on both. Now, using some physical quantity in both sides we may relate the effective couplings $\{Z(\Lambda),g(\Lambda),m(\Lambda),...\}$ with the ruining couplings $\{Z(\mu),g(\mu),m(\mu)\}$.

What is reveled trough this relation is that both approaches are parametrizing the renormalization group of a "same theory", with different parametrization (is the same theory only after the limit $\Lambda\rightarrow\infty$):

  • The first one is parametrizing the renormalization group by the renormalization scale $\mu$ and a renormalization condition $\{Z(\mu),g(\mu),m(\mu)\}$ and push the regularization to infinity $\Lambda\rightarrow\infty$ at the end.
  • The Wilsonian approach is parametrizing the renormalization group by the regularization scale $\Lambda$ and effective couplings $\{Z(\Lambda),g(\Lambda),m(\Lambda),...\}$, and then at the end of calculation we push $\Lambda\rightarrow\infty$ as before.

Note that the effective coupling will diverge as $\Lambda\rightarrow\infty$, for the same reasons as the bare couplings does, in order to have finite physical quantities as predictions.

What is important to note is that running $\mu$ with $\Lambda$ fixed is the same as running $\Lambda$ with $\mu$ fixed. When we do $\Lambda\rightarrow\infty$, in both cases, the dependence on the the details of each regularization and approach will drop out. For more details about the relation between this two approachs see this

  • $\begingroup$ @Noguiera: What you said at the end regarding Wilson's approach is incorrect. See my answer to physics.stackexchange.com/q/372306 for more details. The relation to the old Gell-Mann Low RG is one thing Wilson struggled a lot to explain to high energy physicists. Looks like, he did not succeed. $\endgroup$ Dec 21, 2017 at 14:58
  • $\begingroup$ @Abdelmalek I edit the answer. I'm awere that there is a relation between the two, but is not so direct as the previous relation that I presented. I edit the answer to make more clear the fact that there is indeed a relation, an indirect one, based on the Renormalization Group $\endgroup$
    – Nogueira
    Dec 21, 2017 at 17:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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