# Killing Vector Equations

1. Jan 14, 2010

### TerryW

I'm working my way through D'Inverno's Understanding GR and have reached Chapter 7 with no real problems. I'm stuck though on p101 (and the corresponding problem 7.7). D'inverno sets out two forms of the geodesic equations, one of which is the Lagrangian with the Killing Vectors and then has a throwaway line (paraphrased as) '..it is possible ...using these two equations to read off the components of the Christoffel symbols..'

I have managed all the problems so far but I just cannot see this one. Am I missing something blindingly obvious?

2. Jan 14, 2010

### George Jones

Staff Emeritus
Welcome to Physics Forums!

Let's work through this. Can you find

$$\frac{\partial K}{\partial x^a}?$$

for $x^a$ equal to $t$, $r$, $\theta$, and $\phi$

If you can, post what you get; if you can't, ask more questions.

Last edited: Jan 14, 2010
3. Jan 14, 2010

### TerryW

Hi George, and thanks for helping.

For a=0, I reckon I'll get 1/2(g00,0 + g11,0 +g22,0 + g33,0).

I worked out that
da (dx/dsbdx/dsc = 0 when I was verifying the workings on the previous page.

I already know what the gααs are.

How am I doing?

Terry

4. Jan 14, 2010

### George Jones

Staff Emeritus
For a diagonal metric, this is almost, but not quite, correct.

Since $K = g_{ab} \dot{x}^a \dot{x}^b /2$,

$$\frac{\partial K}{\partial x^0} = \frac{1}{2} \left( \frac{\partial g_{ab}}{\partial x^0} \right) \dot{x}^a \dot{x}^b = \frac{1}{2} \left[ \left( \frac{\partial g_{00}}{\partial x^0} \right) \left( \dot{x}^0 \right)^2 + \left( \frac{\partial g_{11}}{\partial x^0} \right) \left( \dot{x}^1 \right)^2 + \left( \frac{\partial g_{22}}{\partial x^0} \right) \left( \dot{x}^2 \right)^2 + \left( \frac{\partial g_{33}}{\partial x^0} \right) \left( \dot{x}^3 \right)^2 \right]$$

for a diagonal metric.
Very well.

$$\frac{\partial K}{\partial x^0}?$$

If you want to see the code for a mathematical expression, click on the expression, and the code will be displayed in a code window. As you click on different expressions (in any post), new code windows don't appear, the contents of a single code window change.

This is a time consuming process that is sometimes (and sometimes not) worth the effort. For now, don't worry too much about learning how to do this.

5. Jan 14, 2010

### Altabeh

Sorry for asking an irrelevant question: Is this book really great for a self-study, George, or are there any other books which can be more helpful and interesting than D'Inverno's book?

6. Jan 14, 2010

### TerryW

Thanks George,

Oh yes, I forgot about the (dxa/ds)2 terms. It's the next part of the Lagrangian which is the problem though and it seems to be a bit of a circular argument which brings you back to the first equation for the geodesics with d2x0/ds2 and the Christoffel symbols.

Do you use Latex and then cut/paste the result into your posts?

I'm off to bed now. Hope to hear from you tomorrow.

Regards

Terry

7. Jan 14, 2010

### TerryW

Hi Altabeh,

I'm finding it OK but I did work my way all the way through Schaum's Tensor Calculus (doing all the problems!) before I started. I think I might have struggled with D'Inverno's chapters on Tensor Calculus if I hadn't done this.

Regards

Terry

8. Jan 14, 2010

### George Jones

Staff Emeritus
Oops, I forgot the latex \dot. This should read

$$\frac{\partial K}{\partial \dot{x}^0}?$$
I usually type the latex commands as I type my posts.

9. Jan 15, 2010

### Altabeh

And was Schaum's Tensor Calculus useful and easygoing?

10. Jan 15, 2010

### TerryW

Hi George,

My first venture into Latex:

$$\frac{\partial K}{\partial \dot{x}^a} = g_{ab}\dot{x}^b \ \ \ \ (2K = g_{ab}\dot{x}^a\dot{x}^b\ \ \ \frac{\partial g_{ab}}{\partial \dot{x}^a}\ =\ \frac{\partial g_{ab}}{\partial {x}^a}\frac{\partial{x}}{\partial \dot{x}^a} \rightarrow \0\ as\ \ x\rightarrow0)$$

Now you are going to ask me what happens when I take the derivative of

$$\frac{\partial K}{\partial \dot{x}^a}$$

with respect to the affine parameter represented by the dot over the x

11. Jan 15, 2010

### TerryW

Schaum's Tensor Calculus was very useful, but you have to do the problems. There are quite a few printing errors however which leave you pondering for a while. Whether or not it is easy going depends on you!:rofl:

12. Jan 15, 2010

### Altabeh

See that you are getting professional in using Latex...

$$\frac{\partial K}{\partial \dot{x}^a} = \frac{\partial}{\partial \dot{x}^a}(\frac{1}{2}g_{bc}\dot{x}^b\dot{x^c}) = ...$$

Now you continue this calculation and use the Kronecker's delta to manage combining two terms appearing in the operation. (Remember that g_ab is symmetric).

AB

13. Jan 15, 2010

### TerryW

Hi Altabeh,

I thought I'd done that already

$$\frac{\partial K}{\partial \dot{x}^a} = g_{ab}\dot{x}^b}$$

14. Jan 15, 2010

### George Jones

Staff Emeritus
Yes, this is correct. Did you arrive at this by using Kronecker deltas, as Altabeh suggested?
I'm not sure what this means.
Right!

15. Jan 15, 2010

### TerryW

Hi George,

Yes I did arrive at my answer by Kronecker deltas. The other bit was just an explanation of why the partial derivative of gab disappears.

So for the next bit;

$$\frac{d}{du}(g_{ab}\dot{x}^b) = \partial_{c}g_{ab}\dot{x}^b\dot{x}^c+g_{ab}\ddot{x}^b$$

Where now?

16. Jan 15, 2010

### Altabeh

Nice. Now you are required to calculate $$\frac{\partial}{\partial x^{a}}(\frac{1}{2}g_{bc}\dot{x}^b\dot{x}^c)$$. If you've already calculated these, then arrange all the terms obtained so far in the order given by the geodesic equation (7.46). After that just try to get rid of $$g_{ab}$$ in $$g_{ab}\ddot{x }^b$$. What should you do to lead to pure terms like $$\ddot{x }^b$$? (Note: Don't take the index b of $$\ddot{x}^b$$ seriously here. It must be something else if one still uses terms including $$\dot{x}^b\dot{x}^c$$.)

Last edited: Jan 15, 2010
17. Jan 15, 2010

### TerryW

Hi Altabeh,

$$\ddot{x}^a +\Gamma^a_{bc}\dot{x}^b\dot{x}^c = 0$$ ? (7.42 in D'Inverno)

If it is, it isn't the point of my original question which was to find out what D'Inverno means by "It is possible, by (7.42) (ie the equation above) to read off directly from

$$\frac{\partial{K}}{\partial{x}^a} - \frac{d}{du}\left(\frac{\partial{K}}{\partial\dot{x}^a}\right)=0$$ (7.46 in D'Inverno)
the components of the connection $$\Gamma^a_{bc}$$, and this proves to be a very efficient way of calculating $$\Gamma^a_{bc}$$."

I know various techniques for working out the $$\Gamma^a_{bc}$$ values for a given metric, I thought that this was alluding to an even smarter way of doing it but I can't see it.

18. Jan 15, 2010

### Altabeh

Yes, it is! But the point is that if you wanted to derive the Christoffel symbols for the desired metric here directly from the Euler-Lagrange equation, you would end up leading to what D'Inverno asked for in reality. We just gave a proof of how one would extract (7.42) from (7.46) which is the Euler-Lagrange equation.

So all you need now is to start from the line element in spherical coordinates, ds2, and devide both sides of it by du2, so this would be your $$\frac{1}{2}g_{bc}\dot{x}^b\dot{x}^c$$. Remember that here $$(ds/du)^2=2K$$ and for the sake of convenience, simply ignore that factor 2 which won't make any trouble ahead.

AB

19. Jan 15, 2010

### George Jones

Staff Emeritus
Sorry, I have been busy with work all day.

We have been proceeding too generally.

Exactly!

Or if you wanted to write down the equations of geodesic motion for a a given metric without first calculating the $\Gamma^\alpha {}_{\mu \nu}$ explicitly.

Let's start again. Write down the specific $K$ for the specific metric given in exercise 6.31.

20. Jan 16, 2010

### TerryW

Hi George and Altabeh,

The penny has just dropped! I've just done $$\Gamma^0_{00}$$ for the Schwartzchild metric and can see my way to doing the rest.

Thanks very much both of you for helping with this.

Regards

Terry