52 Chapter 2. <strong>Surfaces</strong>: Local Theoryc. (More challeng<strong>in</strong>g) Show that, more generally, for any θ <strong>and</strong> m ≥ 3, we haveH = 1 (( 2π ) ( 2π(m − 1) ) )κ n (θ)+κ n θ + + ···+ κn θ + .mmm10. a. Apply Meusnier’s Formula to a latitude circle on a sphere of radius a to calculate thenormal curvature.b. Prove that the curvature of any curve ly<strong>in</strong>g on the sphere of radius a satisfies κ ≥ 1/a.11. Prove or give a counterexample: If M is a surface with Gaussian curvature K>0, then thecurvature of any curve C ⊂ M is everywhere positive. (Remember that, by def<strong>in</strong>ition, κ ≥ 0.)♯ 12.Suppose that for every P ∈ M, the shape operator S P is some scalar multiple of the identity,i.e., S P (V) =k(P )V for all V ∈ T P M. (Here the scalar k(P )maywell depend on the po<strong>in</strong>tP .)a. Differentiate the equationsD xu n = n u = − kx uD xv n = n v = − kx vappropriately to determ<strong>in</strong>e k u <strong>and</strong> k v <strong>and</strong> deduce that k must be constant.b. We showed <strong>in</strong> Proposition 2.2 that M is planar when k =0. Show that when k ≠0,M is(a portion of) a sphere.13. a. Prove that if α is a l<strong>in</strong>e of curvature <strong>in</strong> M, then (n◦α) ′ (t) =−k(t)α ′ (t), where k(t) isthepr<strong>in</strong>cipal curvature at α(t) <strong>in</strong>the direction α ′ (t). (More colloquially, differentiat<strong>in</strong>g alongthe curve α, wejust write n ′ = −kα ′ .)b. Suppose two surfaces M <strong>and</strong> M ∗ <strong>in</strong>tersect along a curve C. Suppose C is a l<strong>in</strong>e of curvature<strong>in</strong> M. Prove that C is a l<strong>in</strong>e of curvature <strong>in</strong> M ∗ if <strong>and</strong> only if the angle between M <strong>and</strong>M ∗ is constant along C. (In the proof of ⇐=, be sure to <strong>in</strong>clude the case that M <strong>and</strong> M ∗<strong>in</strong>tersect tangentially along C.)14. Prove or give a counterexample:a. If a curve is both an asymptotic curve <strong>and</strong> a l<strong>in</strong>e of curvature, then it must be planar.b. If a curve is planar <strong>and</strong> an asymptotic curve, then it must be a l<strong>in</strong>e.15. a. How is the Frenet frame along an asymptotic curve related to the geometry of the surface?b. Suppose K(P ) < 0. If C is an asymptotic curve with κ(P ) ≠0,prove that its torsionsatisfies |τ(P )| = √ −K(P ). (H<strong>in</strong>t: If we choose an orthonormal basis {U, V} for T P (M)with U tangent to C, what is the matrix for S P ? See the Remark on p. 46.)16. Cont<strong>in</strong>u<strong>in</strong>g Exercise 15, show that if K(P ) < 0, then the two asymptotic curves have torsionof opposite signs at P .17. Prove that the only m<strong>in</strong>imal ruled surface with no planar po<strong>in</strong>ts is the helicoid. (H<strong>in</strong>t: Considerthe curves orthogonal to the rul<strong>in</strong>gs. Use Exercises 8b <strong>and</strong> 1.2.19.)
§2. The Gauss Map <strong>and</strong> the Second Fundamental Form 5318. Suppose U ⊂ R 3 is open <strong>and</strong> x: U → R 3 is a smooth map (of rank 3) so that x u , x v , <strong>and</strong> x ware always orthogonal. Then the level surfaces u = const, v = const, w = const form a triplyorthogonal system of surfaces.a. Show that the spherical coord<strong>in</strong>ate mapp<strong>in</strong>g x(u, v, w) =(u s<strong>in</strong> v cos w, u s<strong>in</strong> v s<strong>in</strong> w, u cos v)furnishes an example.b. Prove that the curves of <strong>in</strong>tersection of any pair of surfaces from different systems (e.g.,v = const <strong>and</strong> w = const) are l<strong>in</strong>es of curvature <strong>in</strong> each of the respective surfaces. (H<strong>in</strong>t:Differentiate the various equations x u · x v =0,x v · x w =0,x u · x w =0with respect to themiss<strong>in</strong>g variable. What are the shape operators of the various surfaces?)19. In this exercise we analyze the surfaces of revolution that are m<strong>in</strong>imal. It will be convenient towork with a meridian as a graph (y = h(u), z = u), as opposed to our customary parametrizationof surfaces of revolution.a. Use Exercise 1.2.4 <strong>and</strong> Proposition 2.5 to show that the pr<strong>in</strong>cipal curvatures arek 1 =h ′′(1 + h ′2 ) 3/2 <strong>and</strong> k 2 = − 1 h · 1√ .1+h ′2b. Deduce that H =0if <strong>and</strong> only if h(u)h ′′ (u) =1+h ′ (u) 2 .c. Solve the differential equation. (H<strong>in</strong>t: Either substitute z(u) =lnh(u)or<strong>in</strong>troduce w(u) =h ′ (u), f<strong>in</strong>d dw/dh, <strong>and</strong> <strong>in</strong>tegrate by separat<strong>in</strong>g variables.) You should f<strong>in</strong>d that h(u) =1ccosh(cu + b) for some constants b <strong>and</strong> c.20. By choos<strong>in</strong>g coord<strong>in</strong>ates <strong>in</strong> R 3 appropriately, we may arrange that P is the orig<strong>in</strong>, the tangentplane T P M is the xy-plane, <strong>and</strong> the x- <strong>and</strong> y-axes are <strong>in</strong> the pr<strong>in</strong>cipal directions at P .a. Show that <strong>in</strong> these coord<strong>in</strong>ates M is locally the graph z = f(x, y) = 1 2 (k 1x 2 + k 2 y 2 )+...,...where limx,y→0 x 2 + y 2 =0.(Youmay start with Taylor’s Theorem: if f is C2 ,wehavewheref(x, y) =f(0, 0) + f x (0, 0)x + f y (0, 0)y +limx,y→0...x 2 + y 2 = 0.)12(fxx (0, 0)x 2 +2f xy (0, 0)xy + f yy (0, 0)y 2) + ...,b. Show that if P is an elliptic po<strong>in</strong>t, then a neighborhood of P <strong>in</strong> M ∩ T P M is just theorig<strong>in</strong> itself. Show that if P is a hyperbolic po<strong>in</strong>t, such a neighborhood is a curve thatcrosses itself at P <strong>and</strong> whose tangent directions at P are the asymptotic directions. Whathappens <strong>in</strong> the case of a parabolic po<strong>in</strong>t?21. Let P ∈ M be a non-planar po<strong>in</strong>t, <strong>and</strong> if K ≥ 0, choose the unit normal so that l, n ≥ 0.a. We def<strong>in</strong>e the Dup<strong>in</strong> <strong>in</strong>dicatrix to be the conic <strong>in</strong> T P M def<strong>in</strong>ed by the equation II P (V, V) =1. Show that if P is an elliptic po<strong>in</strong>t, the Dup<strong>in</strong> <strong>in</strong>dicatrix is an ellipse; if P is a hyperbolicpo<strong>in</strong>t, the Dup<strong>in</strong> <strong>in</strong>dicatrix is a hyperbola; <strong>and</strong> if P is a parabolic po<strong>in</strong>t, the Dup<strong>in</strong><strong>in</strong>dicatrix is a pair of parallel l<strong>in</strong>es.b. Show that if P is a hyperbolic po<strong>in</strong>t, the asymptotes of the Dup<strong>in</strong> <strong>in</strong>dicatrix are given byII P (V, V) =0, i.e., the set of asymptotic directions.