You are on page 1of 1

Math 109b - Problem Set 6 Evan Dummit

4.4.15a. Each transport is along a great circle, so the angle of the vector with the tangent to each great circle (along each segment of the transport) remains constant. Since the original position of v made an angle of 0 with the tangent at p0 along p0 q, the final position of v also makes an angle of 0 with the tangent at p0 along p0 r. Thus the angle between the original and final positions of v is just the angle q between the two tangent vectors at p0 . consider the cone with vertex p0 that is tangent to the sphere along q r. The cone has angle p ê 2 - j, so when it is mapped into 4.4.15b. As noted above, the transport from p0 to q moves v to the tangent vector to q along p0 q. Following example 4.4.1,

the plane by the usual isometry it becomes a circular sector of angle 2 p ÿ sinHp ê 2 - jL = 2 p cos j. Performing the parallel transport in the plane as in example 4.4.1 yields that the transport of v at r makes an angle of p cos q with the tangent at r along
p0 r. Then, again as noted above, the transport from r to p0 preserves the angle that the image of v makes with the tangent to the

sphere along r p0 , so the transport of v at p0 also makes an angle of q cos j with the tangent at p0 along p0 r. From (a), the angle between v and the tangent at p0 along p0 r is q, so since the final position of v lies between the tangents at p0 along p0 q and p0 r, the angle between v and its transport is q - q cos j. 4.4.17. Since a has finite arclength (as it is parametrized by arclength, and I is finite), its trace is bounded. So we may assume (by taking closures) that I = @0, lD and aHI L are (both) closed hence compact. Clearly x is differentiable since xs = a ' HsL + v B ' HsL = T HsL - v tHsL N HsL and xv = BHsL, by Frenet-Serret.
Ø Ø Ø Ø Ø Ø Ø Ø 2 Ø Ø 2

†xs μ xv §2 = T HsL - v tHsL N HsL μ BHsL = - N HsL - v tHsL T HsL = 1 + v2 tHsL2 is strictly positive, implying that xs μ xv is never zero

Then

hence that x has a one-to-one differential. For any p e aHI L there exists an e p > 0 such that x restricted to U p = It - e p , t + e p M μ I-e p , e p M has a continuous inverse for some
t e a-1 H pL. To see this, consider the set S of all points of intersection of two binormals of aH@t - d, t + dDL -- i.e., the "singular

points" of x. This is a closed set in 3 since x is continuous in s and t (cf. algebraic geometry). For sufficiently small d > 0, this set cannot contain p, for otherwise this would require a sequence of points 8 pk < Ø p all of whose binormals passed through p, which would force the curvature at p to be zero, contradicting the hypothesis that the curvature is nonzero. Then for d suffip e S , which is false. Then clearly the value e p = minId, infq e S † p - q§M has the required properties.

ciently small, infq e S † p - q§ > 0, since otherwise there would exist some 8qi < with †qi - p§ Ø 0; since S is closed this would imply

One has aHI L Œ ‹p e aHI L U p so since aHI L is compact, there exists a finite subcovering aHI L Œ ‹k i=1 U pi for some k e . Then taking
e = minIe pk M yields a set of local coordinates that parametrize the regular surface ‹k i=1 U pi û I μ H-e, eL = S as required.
Ø Ø Ø

By obviousity, a ' HsL is tangent to the trace of xHs, 0L, so a '' HsL is orthogonal to the tangent plane at every point of a. But this simply means that the projection of a '' HsL onto the tangent plane is 0, or D a ' HsL ê d s = 0 so that a is a geodesic.
Ø Ø Ø

Printed by Mathematica for Students