# Stress energy tensor transformation

## Homework Statement

Show that if you add a total derivative to the Lagrangian density ##L \to L + \partial_\mu X^\mu##, the energy momentum tensor changes as ##T^{\mu\nu} \to T^{\mu\nu}+\partial_\alpha B^{\alpha\mu\nu}## with ##B^{\alpha\mu\nu}=-B^{\mu\alpha\nu}##.

## The Attempt at a Solution

So we have ##T_{\mu\nu}=\frac{\partial L}{\partial(\partial_\mu \phi)}\partial_\nu \phi-g_{\mu\nu}L##, where ##\phi## is the field that the Lagrangian depends on. If we do the given change on the Lagrangian, the change in ##T_{\mu\nu}## would be ##\frac{\partial (\partial_\alpha X^\alpha)}{\partial(\partial_\mu \phi)}\partial_\nu \phi-g_{\mu\nu}\partial_\alpha X^\alpha =\partial_\alpha \frac{\partial X^\alpha}{\partial(\partial_\mu \phi)}\partial_\nu \phi-g_{\mu\nu}\partial_\alpha X^\alpha##. From here I thought of using this: ##g_{\mu\nu}\partial_\alpha X^\alpha=g_{\mu\nu}\partial_\alpha \phi \frac{\partial X^\alpha}{\partial \phi}## But I don't really know what to do from here. Mainly I don't know how to get rid of that ##g_{\mu\nu}##. Can someone help me?

Orodruin
Staff Emeritus
Homework Helper
Gold Member
2021 Award
the change in ##T_{\mu\nu}## would be ##\frac{\partial (\partial_\alpha X^\alpha)}{\partial(\partial_\mu \phi)}\partial_\nu \phi-g_{\mu\nu}\partial_\alpha X^\alpha =\partial_\alpha \frac{\partial X^\alpha}{\partial(\partial_\mu \phi)}\partial_\nu \phi-g_{\mu\nu}\partial_\alpha X^\alpha##.

This is not correct. First of all, you are making the assumption that you can just insert ##\partial_\mu X^\mu## in the expression for ##T_{\mu\nu}## and obtain the correct ##T_{\mu\nu}##. This is not so because the expression assumes a dependence only on derivatives up to first order existing in ##L##. If your ##X^\mu## contains any derivatives of ##\phi## (as it would have to to make it Lorentz invariant), then ##\partial_\mu X^\mu## will contain derivatives of second order and higher. Second, you are assuming that
$$\frac{\partial (\partial_\alpha X^\alpha)}{\partial(\partial_\mu \phi)} = \partial_\alpha \frac{\partial X^\alpha}{\partial(\partial_\mu \phi)},$$
which is simply not true.

Instead, I suggest that you start from the expression for ##\delta \mathscr S## in terms of the Lagrangian for the transformation ##\delta y^\mu = k^\mu## and ##\delta \phi = 0##, which is what gives the energy-momentum tensor as the current ##J^\mu = T^{\mu}_\nu k^\nu## and takes the form
$$\delta \mathscr S = \int [\{L(\phi(y-\delta y))-L(\phi(y))\} + \partial_\mu(L \delta y^\mu)] d^4x.$$

Edit: Also note that you have some errors in your index placements.
Edit 2: You should not have to make any assumptions on the actual field content.

Last edited:
Thank you for your reply. So I tried this: $$\frac{\delta(L+\partial_\mu X^\mu)}{\delta y^\nu}=\Big(\frac{\partial L}{\partial \phi}+\frac{\partial(\partial_\mu X^\mu)}{\partial \phi}-\partial_\mu\frac{\partial L}{\partial (\partial_\mu \phi)}-\partial_\mu\frac{\partial (\partial_\mu X^\mu)}{\partial (\partial_\mu \phi)}\Big)\frac{\delta \phi}{\delta y^\nu}+\partial_\mu\Big(\frac{\partial L}{\partial(\partial_\mu\phi)}\frac{\delta \phi}{\delta y^\nu}+\frac{\partial (\partial_\mu X^\mu)}{\partial(\partial_\mu\phi)}\frac{\delta \phi}{\delta y^\nu}\Big)$$ So this is the normal derivation from my book, but with the replacement made to the Lagrangian. We also have $$\frac{\partial L}{\partial \phi}-\partial_\mu\frac{\partial L}{\partial (\partial_\mu \phi)}=0$$ as the field satisfies the equations of motion. Then we have $$\frac{\delta \phi}{\delta y^\nu}=\partial_\nu \phi$$ which is the infinitesimal transformation for a scalar field. As the Lagrangian, even wit the new term, is a scalar field we also have: $$\frac{\delta (L+\partial_\mu X^\mu)}{\delta y^\nu}=\partial_\nu L+\partial_\nu(\partial_\mu X^\mu)$$ Equating the 2 expressions we get:
$$\Big(\frac{\partial(\partial_\mu X^\mu)}{\partial \phi}-\partial_\mu\frac{\partial (\partial_\mu X^\mu)}{\partial (\partial_\mu \phi)}\Big)\partial_\nu\phi+\partial_\mu\Big(\frac{\partial L}{\partial(\partial_\mu\phi)}\partial_\nu\phi+\frac{\partial (\partial_\mu X^\mu)}{\partial(\partial_\mu\phi)}\partial_\nu\phi\Big)=\partial_\nu L+\partial_\nu(\partial_\mu X^\mu)$$ Now from the old stress energy tensor we have $$\frac{\partial L}{\partial(\partial_\mu\phi)}\partial_\nu\phi-g_{\mu\nu}L$$ so the part I need to write in the form $$\partial_\mu (M_{\mu\nu})$$ is what remains from there: $$\Big(\frac{\partial(\partial_\alpha X^\alpha)}{\partial \phi}-\partial_\mu\frac{\partial (\partial_\alpha X^\alpha)}{\partial (\partial_\mu \phi)}\Big)\partial_\nu\phi+\partial_\mu\Big(\frac{\partial (\partial_\alpha X^\alpha)}{\partial(\partial_\mu\phi)}\partial_\nu\phi\Big)=\partial_\nu(\partial_\alpha X^\alpha)$$ I replaced the index for the summation for X with ##\alpha##. Based on what I was told my answer should look like this $$T_{\mu\nu} \to T_{\mu\nu} +\partial_\alpha\Big( \frac{\partial X^\alpha}{\partial(\partial_\mu \phi)}\partial\nu \phi - \frac{\partial X^\mu}{\partial(\partial_\alpha \phi)}\partial_\nu \phi \Big)$$. If we apply the product rule in the second term in our equation, we get rid of one more term and we end up with: $$\Big(\frac{\partial(\partial_\alpha X^\alpha)}{\partial \phi}-\partial_\mu\frac{\partial (\partial_\alpha X^\alpha)}{\partial (\partial_\mu \phi)}\Big)\partial_\nu\phi+\partial_\mu\Big(\frac{\partial (\partial_\alpha X^\alpha)}{\partial(\partial_\mu\phi)}\Big)\partial_\nu\phi+\frac{\partial (\partial_\alpha X^\alpha)}{\partial(\partial_\mu\phi)}\partial_\mu\partial_\nu\phi=\partial_\nu(\partial_\alpha X^\alpha)$$ $$\frac{\partial(\partial_\alpha X^\alpha)}{\partial \phi}\partial_\nu\phi+\frac{\partial (\partial_\alpha X^\alpha)}{\partial(\partial_\mu\phi)}\partial_\mu\partial_\nu\phi=\partial_\nu(\partial_\alpha X^\alpha)$$ Now I am not sure what to do. Is it ok so far? Any suggestions now? Thank you again for help!

Orodruin
Staff Emeritus
Homework Helper
Gold Member
2021 Award
You really need to stop thinking of the field ##\phi## altogether and start thinking more in terms of how the action changes when you do the ##\delta y^\mu## transformation and how that affects ##\delta L##.

Also note that some of your terms have four ##\mu## indices. That is never a good sign.

You really need to stop thinking of the field ##\phi## altogether and start thinking more in terms of how the action changes when you do the ##\delta y^\mu## transformation and how that affects ##\delta L##.
Also note that some of your terms have four ##\mu## indices. That is never a good sign.
I am a bit confused, don't I need the ##\phi##, as what I am doing is like a chain rule?

Orodruin
Staff Emeritus
Homework Helper
Gold Member
2021 Award
I am a bit confused, don't I need the ##\phi##, as what I am doing is like a chain rule?
No, it is irrelevant what is inside the ##X^\mu## as your transformation of the fields is ##\delta \phi = 0##, all changes in the action will be expressible in terms of ##X^\mu## and its derivatives.

No, it is irrelevant what is inside the ##X^\mu## as your transformation of the fields is ##\delta \phi = 0##, all changes in the action will be expressible in terms of ##X^\mu## and its derivatives.
I am not sure why the change in ##\phi## is zero. When you derive E-L equations, you actually need ##\delta \phi## to be non-zero.

Orodruin
Staff Emeritus
Homework Helper
Gold Member
2021 Award
I am not sure why the change in ##\phi## is zero. When you derive E-L equations, you actually need ##\delta \phi## to be non-zero.
I am not talking about the equations of motion, I am talking about the 1-parameter symmetry related to the conserved current that is the energy-momentum tensor contracted with the spacetime translation vector ##k^\nu##, i.e., ##J^\mu = T^\mu_\nu k^\nu##.

You really need to stop thinking of the field ##\phi## altogether and start thinking more in terms of how the action changes when you do the ##\delta y^\mu## transformation and how that affects ##\delta L##.

Also note that some of your terms have four ##\mu## indices. That is never a good sign.
Ok, so for the change in action we have: $$\frac{\delta S}{\delta y^\mu}=\int(\frac{\delta L}{\delta y^\mu}+\frac{\delta \partial_\alpha X^\alpha }{\delta y^\mu})$$ And the first term gives the old result. Is this ok so far?

I am not talking about the equations of motion, I am talking about the 1-parameter symmetry related to the conserved current that is the energy-momentum tensor contracted with the spacetime translation vector ##k^\nu##, i.e., ##J^\mu = T^\mu_\nu k^\nu##.
Any further hint would be greatly appreciated

Orodruin
Staff Emeritus