A Derivation of free fields... coefficient question

topsquark
Science Advisor
Homework Helper
Insights Author
MHB
Messages
2,020
Reaction score
843
Editing... Please be patient!

I came across this difficulty comparing several versions of the Feynman rules for QED. I traced the problem back to a statement of Weinberg's in "The Quantum Theory of Fields (Vol 1)", Chapter 5, page 195. Weinberg is busy setting up for deriving free field expressions for different particle types. He has got to the point where he is deriving the field operators as an integral expression. Specifically, he derives two equations:
##\displaystyle \psi_l^+(x) = \sum_{\sigma , n} (2 \pi )^{3/2} \int d^3p u_l(\textbf{p}, \sigma, n) e^{i p \cdot x} a(\textbf{p}, \sigma, n)##

and
##\displaystyle \psi_l^-(x) = \sum_{\sigma , n} (2 \pi )^{3/2} \int d^3p v_l(\textbf{p}, \sigma, n) e^{-i p \cdot x} a^{\dagger} (\textbf{p}, \sigma, n)##

Now, I understand the overall features here. But Weinberg continues on to say
The factors ##(2 \pi )^{3/2}## could be absorbed into the definition of ##u_l## and ##v_l##, but it is conventional to show them explicitly in these Fourier integrals.

Now, it's a Fourier integral, so I do see the value in showing the ##2 \pi##s. But by the time the Feynman rules are derived, practically everyone has redefined the field operators so that the ##(2 \pi )^{3/2}## factor disappears.

And when we get to the amplitude, Weinberg's insistence on keeping the ##2 \pi##s alters the amplitude compared to other sources. As an example, see here, equations 2.1 and 2.2. Weinberg's set of Feynman rules gives an overall factor of ##(2 \pi)^{-6}## to these amplitudes.

How do we get rid of those extra ##2 \pi##s?

Thanks!

-Dan

Addendum: Problem solved! Weinberg's extra ##(2 \pi)^{-3/2}## factor in his version of the Feynman rules is there to get rid of that extra ##(2 \pi)^{3/2}## factor in the Fourier integral. I saw it just as I posted this.
 
Physics news on Phys.org
Congratulations on a very compact post.
 
This is a mess in the literature, and it drives you always nuts, if you like to compare results from different sources. Weinberg uses the non-relativistic convention for the anti-commutators (see p. 325 in Weinberg's book),
$$[\hat{a}(\vec{p},\sigma),\hat{a}^{\dagger}(\vec{p}',\sigma')]_+=[\hat{b}(\vec{p},\sigma),\hat{b}^{\dagger}(\vec{p}',\sigma')]_+=\delta^{(3)}(\vec{p}'-\vec{p}) \delta_{\sigma \sigma'}$$
and all other combinations of anti-commutators among annihilation and creation operators vanishing. That as the advantage that the corresponding momentum eigen states,
$$|a,\vec{p},\sigma \rangle=\hat{a}^{\dagger}(\vec{p},\sigma) |\Omega \rangle$$
and analogously for anti-particles are normalized to a ##\delta##-distribution.

What's never changing is the equal-time (anti-)commutators between the field operators and the conjugate field momenta, which are always (using the canonical formalism as a shortcut in comparison to the more tedious but much more insightful from a fundamental point of view, where everything is built up from Lorentz invariance and microcausality, as explained in the chapters before) read
$$[\hat{\phi}_j(t,\vec{x}),\hat{\Pi}_k(t,\vec{y})]_{\pm} = \mathrm{i} \delta^{(3)}(\vec{x}-\vec{y}) \delta_{jk}$$
for some fields ##\phi_j## and their canonical conjugated field momenta ##\hat{\Pi}_k## (for fermions/half-integer spin the upper and bosons/integer spin the lower sign).
 
Last edited:
vanhees71 said:
This is a mess in the literature, and it drives you always nuts, if you like to compare results from different sources. Weinberg uses the non-relativistic convention for the anti-commutators (see p. 325 in Weinberg's book),
$$[\hat{a}(\vec{p},\sigma),\hat{a}^{\dagger}(\vec{p}',\sigma')]_+=[\hat{b}(\vec{p},\sigma),\hat{b}^{\dagger}(\vec{p}',\sigma')]_+=\delta^{(3)}(\vec{p}'-\vec{p}) \delta_{\sigma \sigma'}$$
and all other combinations of anti-commutators among annihilation and creation operators vanishing. That as the advantage that the corresponding momentum eigen states,
$$|a,\vec{p},\sigma \rangle=\hat{a}^{\dagger}(\vec{p},\sigma) |\Omega \rangle$$
and analogously for anti-particles are normalized to a ##\delta##-distribution.

What's never changing is the equal-time (anti-)commutators between the field operators and the conjugate field momenta, which are always (using the canonical formalism as a shortcut in comparison to the more tedious but much more insightful from a fundamental point of view, where everything is built up from Lorentz invariance and microcausality, as explained in the chapters before) read
$$[\hat{\phi}_j(t,\vec{x}),\hat{\Pi}_k(t,\vec{y})]_{\pm} = \mathrm{i} \delta^{(3)}(\vec{x}-\vec{y}) \delta_{jk}$$
for some fields ##\phi_j## and their canonical conjugated field momenta ##\hat{\Pi}_k## (for fermions/half-integer spin the upper and bosons/integer spin the lower sign).
I know. I like his style of presentation, but occasionally it's useful to reference other sources and his conventions can get in the way. I've been writing some stuff on the Feynman rules for QED and his electron propagator is:
## \dfrac{\left ( \dfrac{-i}{(2 \pi )^4 } \right ) \dfrac{ -i \not \! q + m}{q^2 + m^2 - i \epsilon} ##

I haven't found that anywhere else. Most everyone else seems to use
##\dfrac{i (\not \! q + m)}{q^2 - m^2 - i\epslion}##

This part I already knew: His convention for the ##\gamma## matrices is i times the "typical" choice, and he's using different metric. (Frankly, I use his metric, but anyway...)

I presume that there's a reason why he would choose this, and I can certainly work with it now that I know where the factors come from, but it can be really annoying to check my calculations with something I find online!

-Dan

Addendum: What did I do wrong with the LaTeX? That should be right!
 
topsquark said:
I know. I like his style of presentation, but occasionally it's useful to reference other sources and his conventions can get in the way. I've been writing some stuff on the Feynman rules for QED and his electron propagator is:
## \left ( \dfrac{-i}{(2 \pi )^4 } \right ) \dfrac{ -i \not \! q + m}{q^2 + m^2 - i \epsilon} ##

I haven't found that anywhere else. Most everyone else seems to use
##\dfrac{i (\not \! q + m)}{q^2 - m^2 - i\epsilon}##

This part I already knew: His convention for the ##\gamma## matrices is i times the "typical" choice, and he's using different metric. (Frankly, I use his metric, but anyway...)

I presume that there's a reason why he would choose this, and I can certainly work with it now that I know where the factors come from, but it can be really annoying to check my calculations with something I find online!

-Dan

Addendum: What did I do wrong with the LaTeX? That should be right!
That's the LaTeX-corrected posting :-). What puzzles me most is the i in front of the q-slash.
 
Last edited:
vanhees71 said:
That's the LaTeX-corrected posting :-). What puzzles me most is the i in front of the q-slash.
Weinberg defines the ##\gamma##s as
##\gamma^0 = -i \begin{pmatrix} 0 & I \\ I & 0 \end{pmatrix}##

##\gamma^i = -i \begin{pmatrix} 0 & \sigma^i \\ -\sigma^i & 0 \end{pmatrix}##

which makes his
##-i \not \! q + m = -i \gamma^0 q_0 - i \gamma^i q_i + mI = \begin{pmatrix} 0 & -imI -i q_0I - i \sigma^i q_i \\-m - i Iq_0 + i \sigma^i q_i & 0\end{pmatrix}##

##\hphantom{XXXXX} = - \begin{pmatrix} 0 & imI + iq_0I + \sigma^i q_i \\ imI + i Iq_0 - i \sigma^i q_i & 0 \end{pmatrix}##
vs the "usual" standard
##i ( \not \! q + m ) = \begin{pmatrix} im & i Iq_0 + i \sigma^i q_i \\ i Iq_0 - i \sigma^i q_i & -im \end{pmatrix}##

I'm still looking into this to see how this makes the amplitude come out right, but I'm thinking we essentially get an overall factor of i out of it. As we are going to square the modulus in the end, this won't matter.

-Dan
 
Toponium is a hadron which is the bound state of a valance top quark and a valance antitop quark. Oversimplified presentations often state that top quarks don't form hadrons, because they decay to bottom quarks extremely rapidly after they are created, leaving no time to form a hadron. And, the vast majority of the time, this is true. But, the lifetime of a top quark is only an average lifetime. Sometimes it decays faster and sometimes it decays slower. In the highly improbable case that...
I'm following this paper by Kitaev on SL(2,R) representations and I'm having a problem in the normalization of the continuous eigenfunctions (eqs. (67)-(70)), which satisfy \langle f_s | f_{s'} \rangle = \int_{0}^{1} \frac{2}{(1-u)^2} f_s(u)^* f_{s'}(u) \, du. \tag{67} The singular contribution of the integral arises at the endpoint u=1 of the integral, and in the limit u \to 1, the function f_s(u) takes on the form f_s(u) \approx a_s (1-u)^{1/2 + i s} + a_s^* (1-u)^{1/2 - i s}. \tag{70}...
Back
Top