1. Mar 12, 2013

### jtceleron

This question is from K. Huang, Quantum Field Theory: from operators to path integrals.

He says that, under a continuous infinitesimal transformation,
$\phi$(x)->$\phi$(x)+$\delta\phi$
the change of the Lagrangian density must be in the from
$\delta$L=$∂^{\mu}$$W_{\mu}$(x)

It is easily understood that this quantity must be Lorentz-invariant, but why should only be this form, not others (e.g. without derivatives).

2. Mar 12, 2013

### andrien

because only in this way,it can not affect the action integral.It is 4 divergence,so it will vanish.