- #1

sit.think.solve

- 9

- 0

You are using an out of date browser. It may not display this or other websites correctly.

You should upgrade or use an alternative browser.

You should upgrade or use an alternative browser.

- Thread starter sit.think.solve
- Start date

- #1

sit.think.solve

- 9

- 0

- #2

cristo

Staff Emeritus

Science Advisor

- 8,140

- 74

After reading this, do you ahave any other specific questions?

- #3

garrett

Gold Member

- 412

- 45

The covariant derivative uses a connection, while the Lie derivative doesn't.

- #4

Doodle Bob

- 255

- 0

Just to spotlight one of these: The Lie derivative L_X(Y) is basically a dynamical invariant. It measures how compatible the flows of the vector fields are, i.e. how much they commute with each other. If X is generated by the flow [itex]\phi_t[/itex] and Y is generated by [itex]\psi_s[/itex], then the Lie derivative [X,Y] at point P is the tangent vector at time 0 of the curve given by: [itex]t \mapsto \psi_{-t} \circ \phi_{-t} \circ \psi_t \circ \phi_t (P)[/itex].

Berger describes this as moving P forward in time along the X-curves by t, then moving along the Y-curve by t, then moving backward in time along the X-curve and finally moving backward in time along the Y-curve. If you've ended up back at P, then [X,Y]=0 at P.

In particular, if [itex] \phi_s \circ \psi_t=\psi_t \circ \phi_s [/itex] for all s and t, then [X,Y]=0.

The covariant derivative acts similarly except instead of pushing Y along the X-curves via X's flow, we are pushing Y along X-curve via parallel transport. This explanation, though, is a bit of circular logic, since one usually uses the specific covariant derivative to generate the parallel transport.

Berger describes this as moving P forward in time along the X-curves by t, then moving along the Y-curve by t, then moving backward in time along the X-curve and finally moving backward in time along the Y-curve. If you've ended up back at P, then [X,Y]=0 at P.

In particular, if [itex] \phi_s \circ \psi_t=\psi_t \circ \phi_s [/itex] for all s and t, then [X,Y]=0.

The covariant derivative acts similarly except instead of pushing Y along the X-curves via X's flow, we are pushing Y along X-curve via parallel transport. This explanation, though, is a bit of circular logic, since one usually uses the specific covariant derivative to generate the parallel transport.

Last edited:

Share:

- Replies
- 15

- Views
- 3K

- Last Post

- Replies
- 6

- Views
- 1K

- Replies
- 5

- Views
- 1K

- Last Post

- Replies
- 5

- Views
- 755

- Replies
- 8

- Views
- 2K

- Replies
- 4

- Views
- 967

- Last Post

- Replies
- 4

- Views
- 2K

- Replies
- 37

- Views
- 6K

- Last Post

- Replies
- 7

- Views
- 644

- Replies
- 6

- Views
- 2K