Lorentz covaraince of differntial operator

In summary: I mean simpler...way to do it:\frac{\partial}{\partial x'^\nu} = \frac{\partial}{\partial x^\mu} \frac{\partial}{\partial x^\mu}
  • #36
Yes, the two indices on the components of the metric transform "covariantly" (i.e. with the same matrix as the basis vectors of the tangent space) while the index on the components of the coordinates of an event transforms "contravariantly" (with the inverse of the matrix that transforms the basis vectors of the tangent space).

The former transformation matrix is denoted by [itex]\Lambda^{-1}[/itex] and the latter by [itex]\Lambda[/itex]. This means that [tex]g_{\bar \mu \bar \nu} = \frac{\partial x^\rho}{\partial x^{\bar \mu}}\frac{\partial x^\sigma}{\partial x^{\bar \nu}} g_{\rho\sigma}[/tex] isn't equivalent to [itex]g'=\Lambda^Tg\Lambda[/itex] as I said. It's equivalent to [itex]g'=(\Lambda^{-1})^T g\Lambda^{-1}[/itex].

I got a bit confused by the right-hand side of [tex]g_{\bar \mu \bar \nu} = \frac{\partial x^\rho}{\partial x^{\bar \mu}}\frac{\partial x^\sigma}{\partial x^{\bar \nu}} g_{\rho\sigma} = \Lambda^\rho {}_{\bar \mu} \, \Lambda^\sigma {}_{\bar \nu} \, g_{\rho \sigma}[/tex] in your post #33. I would have written this as [tex]g'_{\mu\nu} = \frac{\partial x^\rho}{\partial x'^{\mu}}\frac{\partial x^\sigma}{\partial x'^{\nu}} g_{\rho\sigma} = (\Lambda^{-1})^\rho{}_\mu (\Lambda^{-1})^\sigma{}_\nu\, g_{\rho\sigma}= \Lambda_\mu{}^\rho\, \Lambda_\nu{}^\sigma \, g_{\rho \sigma}.[/tex]
I don't use the notation that puts primes, bars or whatever on the indices instead of on the variables, so I didn't know that people who use it write row [itex]\mu[/itex] column [itex]\nu[/itex] of [itex]\Lambda^{-1}[/itex] as [itex]\Lambda^\mu{}_{\bar\nu}[/itex].
 
Last edited:
Physics news on Phys.org
  • #38
I don't think it's necessary to change it. It's just a different notational convention than the one I'm using. They're letting us know that we're dealing with the inverse by moving the bar to lower index, so they don't also have to let us know by changing the horizontal position of the indices.
 
  • #39
Okay, I think I get it.

The equation from the wikipedia article

[tex]g_{\bar \mu \bar \nu} = \frac{\partial x^\rho}{\partial x^{\bar \mu}}\frac{\partial x^\sigma}{\partial x^{\bar \nu}} g_{\rho\sigma} = \Lambda^\rho {}_{\bar \mu} \, \Lambda^\sigma {}_{\bar \nu} \, g_{\rho \sigma}[/tex]


actually implicitly defines

[tex]\Lambda^\rho { }_{\bar \mu} = \frac{\partial x^\rho}{\partial x^\bar \mu}[/tex]

and

[tex]\Lambda^\sigma { }_{\bar \nu} = \frac{\partial x^\sigma}{\partial x^\bar \nu}[/tex]

While a Lorentz Transformation from event (t,x,y,z) to [itex](\tau,\bar x, \bar y, \bar z)[/itex]

would be given as

[tex]\Lambda^{\bar \sigma} { }_\nu = \frac{\partial x^{\bar \sigma}}{\partial x^\nu}[/tex]

I didn't realize the versatility of the [itex]\Lambda[/itex] symbol.
 
  • #40
Hi,

I'm reviving this because I think I get the notation now but the original question still troubles me.
Fredrik said:
Let's focus on the problem at hand. You got this part right: [tex]\frac{\partial}{\partial x'^\nu} = \frac{\partial x^\mu}{\partial x'^\nu} \frac{\partial}{\partial x^\mu}[/tex] The next step is [tex]=(\Lambda^{-1})^\mu{}_\nu\frac{\partial}{\partial x^\mu}=\Lambda_\nu{}^\mu\frac{\partial}{\partial x^\mu}[/tex]
We have [itex]x'^{\mu}=\Lambda^{\mu}_{\ \ \nu}x^{\nu}\ \ \ (1)[/itex]

So [itex]dx'^{\mu}=\Lambda^{\mu}_{\ \ \nu}dx^{\nu}\ \ \ (2)[/itex]

which also means that [itex]dx^{\mu}=(\Lambda^{-1})^{\mu}_{\ \ \nu}dx'^{\nu}\ \ \ (3)[/itex]

When you did [tex]\frac{\partial x^\mu}{\partial x'^\nu} \frac{\partial}{\partial x^\mu}=(\Lambda^{-1})^\mu{}_\nu\frac{\partial}{\partial x^\mu}[/tex] I assumed initially that you used (3) to get [tex]\frac{\partial x^\mu}{\partial x'^\nu} \frac{\partial}{\partial x^\mu}= \frac{(\Lambda^{-1})^{\mu}_{\ \ \nu}\partial x'^{\nu}}{\partial x'^\nu} \frac{\partial}{\partial x^\mu}[/tex] and canceled the [itex]\partial x'^{\nu}[/itex] from the numerator and denominator.

But they cannot be canceled out like that since the numerator is a summation over the dummy index [itex]\nu[/itex] whereas the index in the denominator is a fixed one. So what exactly did you do to get that equality?

Thank you
 
  • #41
(1) is the four components of the matrix equation [itex]x'=\Lambda x[/itex]. Multiply with [itex]\Lambda^{-1}[/itex] from the left, and you get [itex]\Lambda^{-1}x'=x[/itex]. The [itex]\mu[/itex]th component of this equation is [tex]x^\mu=(\Lambda^{-1})^\mu{}_\nu x'^\nu=\Lambda_\nu{}^\mu x'^\nu.[/tex] The expression [tex]\frac{\partial x^\mu}{\partial x'^\nu}[/tex] denotes the [itex]\nu[/itex]th partial derivative of the function [itex]x^\mu:\mathbb R^4\rightarrow\mathbb R[/itex] defined by [tex]x^\mu(x'^0, x'^1, x'^2, x'^3)=\Lambda_\nu{}^\mu x'^\nu=\Lambda_0{}^\mu x'^0+\Lambda_1{}^\mu x'^1+\Lambda_2{}^\mu x'^2+\Lambda_3{}^\mu x'^3.[/tex] All you have to do now is to compute the partial derivatives.

Edit: Regarding the "cancelling out" issue, note that the chain rule says that [tex]\frac{\partial f}{\partial x'^\nu}=\frac{\partial x^\mu}{\partial x'^\nu}\frac{\partial f}{\partial x^\mu}.[/tex]
 
Last edited:
  • #42
Sorry but I still cannot see how it all cancels out.

Let us consider [tex]\frac{\partial}{\partial x'^{2}}[/tex]

The chain rule says [tex]\frac{\partial}{\partial x'^{2}}=\frac{\partial x^{0}}{\partial x'^{2}}\frac{\partial}{\partial x^{0}}+\frac{\partial x^{1}}{\partial x'^{2}}\frac{\partial}{\partial x^{1}}+\frac{\partial x^{2}}{\partial x'^{2}}\frac{\partial}{\partial x^{2}}+\frac{\partial x^{3}}{\partial x'^{2}}\frac{\partial}{\partial x^{3}}[/tex]

Now, we express the first term on the RHS as follows:[tex]\frac{\partial x^{0}}{\partial x'^{2}}\frac{\partial}{\partial x^{0}} = \frac{\Lambda_{0}^{\ \ 0}\partial x'^{0}}{\partial x'^{2}}\frac{\partial}{\partial x^{0}}+\frac{\Lambda_{1}^{\ \ 0}\partial x'^{1}}{\partial x'^{2}}\frac{\partial}{\partial x^{0}}+\frac{\Lambda_{2}^{\ \ 0}\partial x'^{2}}{\partial x'^{2}}\frac{\partial}{\partial x^{0}}+\frac{\Lambda_{3}^{\ \ 0}\partial x'^{3}}{\partial x'^{2}}\frac{\partial}{\partial x^{0}}[/tex]

And we get similar expressions for the three other terms. But the expression we should get is only the third term on the RHS (where we can "cancel out" the [itex]\partial x'^{2}[/itex] term) i.e. [tex]\frac{\partial x^{0}}{\partial x'^{2}}\frac{\partial}{\partial x^{0}}=\Lambda_{2}^{\ \ 0}\frac{\partial}{\partial x^{0}}[/tex]

This correctly satisfies [tex]\frac{\partial}{\partial x'^{\nu}}=\Lambda_{\nu}^{\ \ \mu}\frac{\partial}{\partial x^{\mu}}[/tex]

Where do the remaining terms in the earlier expression go?

Thank you :smile:

EDIT: Oh dear! Of course, they're all zero! Mega brain fart there. Sorry about that, and thank you so much.
 
Last edited:
  • #43
qbert said:
I think something that needs to be mentioned is that there are two "kinds" of 4-vector.
Ones that transform like the coordinates: [itex] x' = \Lambda x [/itex], in physics we
call these contravariant vectors or just vectors. we generally write them in index notation
as a beast with an upper index like [itex] x^i[/itex]. in this notation we have
[itex] x'{}^{i} = \Lambda^i_{\;j} x^j [/itex].

the second kind are like the derivatives and transform with the inverse matrix:
[itex] p' = \Lambda^{-1} p[/itex]. in physics we call these covariant vectors
or co-vectors. we generally write them in index notation as something with a lower
index like [itex]p_i[/itex]. in this notation we have
[itex] p'_{i} = \left( \Lambda^{-1} \right)_i^{\;j} p_j [/itex]

Ok one more thing based on qbert's post.

The four vector [tex](\frac{\partial}{\partial x^{0}},\frac{\partial}{\partial x^{1}},\frac{\partial}{\partial x^{2}},\frac{\partial}{\partial x^{3}})[/tex] is a covariant one because it transforms the "opposite" way i.e. with the inverse matrix. So should it be represented as [itex]\partial_{\mu}[/itex] since covariant vectors have the lower index?

Also, does [tex](\frac{\partial}{\partial x_{0}},\frac{\partial}{\partial x_{1}},\frac{\partial}{\partial x_{2}},\frac{\partial}{\partial x_{3}})[/tex] then become the contravariant four vector i.e. [itex]\partial^{\mu}[/itex]?

Lastly, would it make sense to think of the covariant vectors as row vectors and contravariant vectors as column vectors? I read in another thread that this is usually ok (with the caveat that it is not always correct) but it seems to be a bit inconsistent if applied here.

Thank you and sorry to drag the thread on for so long.
 
Last edited:
  • #44
Short answer: yes, yes and yes.
 
  • #45
Great, thank you!
 
Back
Top