In Peskin & Schroeder's An Introduction to Quantum Field Theory, section 10.5, page 338, the book gives a two-loop renormalization example (in scalar $\phi^4$ theory).
Before we start the two-loops, let's us recall the renormalization condition in (10.19)
For the second renormalization condition, my understanding for two-loop case is
$$\text{One-loops + two-loops + counterterms}=-i\lambda. $$
However, the book in two-loop examples only considers the two loop case. The relevant two-loop feynman diagrams are given in (10.51)
The sentence below (10.51) reads
The value of last diagram in (10.51) is just a constant, which we can freely adjust to absorb any divergent terms that are independent of the external momenta.
-
Here, does the momentum independent divergence including double poles divergence? i.e. $(\frac{1}{\epsilon})^2$.
-
However, on page 339, the book refers
which is the result from one-loop renormalization. In this case, the $\delta_\lambda$ is already specified. And we cannot freely adjust it to absorb momentum independent divergence. This is contradict with the sentence below (10.51).
So I am wondering what's wrong with my logic?
Best Answer
Yes, see e.g. the first equation on p. 341.
The 1-loop counterterms are of order ${\cal O}(\lambda^2)$ while the 2-loop counterterms are of order ${\cal O}(\lambda^3)$ in the perturbative formal power series for the coupling constant $\lambda$. [The order$^1$ is determined by the number of $\times$-vertices in the Feynman diagrams that a counterterm is designed to (partially) cancel.] The 1-loop and 2-loop counterterms are hence independent terms.
--
$^1$The counting is more complicated if the Feynman diagram contains $\otimes$-vertices. Since the counterterm $\delta_{\lambda}$ is a power series expansion in $\lambda$ (the lowest power is quadratic), the corresponding $\otimes$-vertex doesn't contain a definite power of $\lambda$.