Tensor Calculus in Hyperbolic Coordinates We shall do for hyperbolic coordinates in two-dimensional space-time all the things that Schutz

does for polar coodinates in two-dimensional Euclidean space. The coordinate transformation
[Put drawing here.]

Introduce the coordinates ( , ) by t = sinh , x = cosh . t = tanh , t2 + x2 = 2 . (1) x The curve = const. is a straight line through the origin. The curve = const. is a hyperbola. As varies from 0 to and varies from to (endpoints not included), the region x > 0, x < t < x is covered one-to-one. In some ways is analogous to r and is analogous to , but geometrically there are some important dierences. From Exercises 2.21 and 2.19 we recognize that the hyperbola = const. is the path of a uniformly accelerated body with acceleration 1/. (The parameter is not the proper time but is proportional to it with a scaling that depends on .) From Exercises 1.18 and 1.19 we see that translation in (moving the points ( , ) to the points ( + 0 , )) is a Lorentz transformation (with velocity parameter 0 ). Let unprimed indices refer to the inertial coordinates (t, x) and primed indices refer to the hyperbolic coordinates. The equations of small increments are t t + = cosh + sinh , t = (2) x = sinh + cosh . Therefore, the matrix of transformation of (tangent or contravariant) vectors is V = V , Inverting this matrix, we have V = V ,

Then

cosh sinh
1

sinh cosh
1 sinh cosh

(3)

cosh sinh

(4)

(Alternatively, you could nd from (1) the formula for the increments ( , ) in terms of (t, x). But in that case the coecients would initially come out in terms of the inertial coordinates, not the hyperbolic ones. These formulas would be analogous to (5.4), while (4) is an instance of (5.89).)
If you have the old edition of Schutz, be warned that the material on p. 128 has been greatly improved in the new edition, where it appears on pp. 119120. 1

Basis vectors and basis one-forms Following p. 122 (new edition) we write the transformation of basis vectors e = e , e = cosh et + sinh ex , e = sinh et + cosh ex ; and the transformation of basis covectors
E = E ,

(5)

which is now written in a new way convenient for coordinate systems, 1 1 d = cosh dt sinh dx , d = sinh dt + cosh dx . (6)

To check that the notaion is consistent, note that (because our two matrices are inverses of each other) d (e ) = E (e ). Note that equations (6) agree with the classical formulas for the dierentials of the curvilinear coordinates as scalar functions on the plane; it follows that, for example, d (v) is (to rst order) the change in under a displacement from x to x + v. Note also that the analog of (6) in the reverse direction is simply (2) with replaced by d. The metric tensor Method 1: By denitions (see (5.30)) g = g(e , e ) = e e . So g = 2 , These facts are written together as ds2 = 2 d 2 + d 2 , or g The inverse matrix, {g }, is
1 2 0

g = 1,

g = g = 0.

2 0

0 1

0 1

Method 2: In inertial coordinates g Now use the

0 2 O

1 0

0 1

tensor transformation law g = g ,

which in matrix notation is g g g g = t x t x

t

1 0

0 1

t x

t x ,

which, with (3), gives the result. This calculation, while conceptually simple, is cumbersome and subject to error in the index conventions. Fortunately, there is a streamlined, almost automatic, version of it: Method 3: In the equation ds2 = dt2 + dx2 , write out the terms via (2) and simplify, treating the dierentials as if they were numbers: dx2 = ( cosh d + sinh d)2 + ( sinh d + cosh d.)2 = 2 d 2 + d 2 . Christoffel symbols A generic vector eld can be written v = v e . If we want to calculate the derivative of v with respect to , say, we must take into account that the basis vectors {e } depend on . Therefore, the formula for such a derivative in terms of components and coordinates contains extra terms, with coecients called Christoel symbols. [See () and the next equation in the old notes, or (5.43,46,48,50) in the book.] The following argument shows the most elementary and instructive way of calculating Christoel symbols for curvilinear coordinates in at space. Once we get into curved space we wont have inertial coordinates to fall back upon, so other methods of getting Christoel symbols will need to be developed. Dierentiate (5) to get e e e e = sinh et + cosh ex = e , = cosh et + sinh ex = 1 e , 1 = cosh et + sinh ex = e ,

= 0.
3

Since by denition e = e , x we can read o the Christoel symbols for the coordinate system ( , ): = 0, = 0, 1 ,

= =

= = 0, = , = 0.

Later we will see that the Christoel symbol is necessarily symmetric in its subscripts, so in dimension d the number of independent Christoel symbols is d (superscripts) d(d + 1) (symmetric subscript pairs) = 1 d2 (d + 1). 2 2

For d = 2, 3, 4 we get 6, 18, 40 respectively. In particular cases there will be geometrical symmetries that make other coecients equal, make some of them zero, etc.
Thanks to Charlie Jessup and Alex Cook for taking notes on my lectures in Fall 2005.