ANSWERS TO SELECTED EXERCISES1.1.1 α(t) = ( 1−t 21+t 2 ,2t1+t 2 ).1.1.4 We parametrize the curve by α(t) = (t, f(t)), a ≤ t ≤ b, <strong>and</strong> so length(α) =∫ ba ‖α′ (t)‖ dt = ∫ ba√1+(f ′ (t)) 2 dt.1.1.6 β(s) = ( 12 (√ s 2 +4+s), 1 2 (√ s 2 +4− s), √ 2 ln(( √ s 2 +4+s)/2) ) .1.2.1 c. κ =12 √ 2 √ 1−s 21.2.3 a. T = 1 2 (√ 1+s, − √ 1 − s, √ 12), κ =2 √ 2 √ , N =1/√ 2( √ 1 − s, √ 1+s, 0), B =1−s 212 (−√ 1+s, √ 1 − s, √ 12), τ =2 √ 2 √ ; c. T = √ √1(t, √1−s 2 2 1+t1+t 2 , 1), κ = τ =21/2(1 + t 2 ), N = √ 1 (1, 0, −t), B = √ √1(−t, √1+t 2 2 1+t1+t 2 , −1)21.2.5 κ =1/ s<strong>in</strong>h t (which we see, once aga<strong>in</strong>, is the absolute value of the slope).1.2.6 B ′ =(T×N) ′ = T ′ ×N+T×N ′ =(κN)×N+T×(−κT+τB) =τ(T×B) =τ(−N),as required.1.2.9 b. If all the osculat<strong>in</strong>g planes pass through the orig<strong>in</strong>, then there are scalar functionsλ <strong>and</strong> µ so that 0 = α + λT + µN for all s. Differentiat<strong>in</strong>g <strong>and</strong> us<strong>in</strong>g the Frenetformulas, we obta<strong>in</strong> 0 = T + κλN + λ ′ T + µ ( −κT + τB ) + µ ′ N; collect<strong>in</strong>g terms, wehave 0 = ( 1+λ ′ − κµ ) T + ( κλ + µ ′) N + µτB for all s. S<strong>in</strong>ce {T, N, B} is a basisfor R 3 ,we<strong>in</strong>fer, <strong>in</strong> particular, that µτ =0. (We could also just have taken the dotproduct of the entire expression with B.) µ(s) =0leads to a contradiction, so wemust have τ =0<strong>and</strong> so the curve is planar.1.2.11 We have α ′ × α ′′ = κυ 3 B, so α ′ × α ′′′ = (α ′ × α ′′ ) ′ = (κυ 3 B) ′ = (κυ 3 ) ′ B +(κυ 3 )(−τυN), so (α ′ × α ′′′ ) · α ′′ = −κ 2 τυ 6 . Therefore, τ = α ′ · (α ′′ × α ′′′ )/(κ 2 υ 6 ),<strong>and</strong> <strong>in</strong>sert<strong>in</strong>g the formula of Proposition 2.2 gives the result.1.2.23 a. Consider the unit normal A s,t to the plane through P = 0, Q = α(s), <strong>and</strong>R = α(t). Choos<strong>in</strong>g coord<strong>in</strong>ates so that T(0) = (1, 0, 0), N(0) = (0, 1, 0), <strong>and</strong>B(0) = (0, 0, 1), we apply Proposition 2.6 to obta<strong>in</strong>α(s) × α(t) =st(s − t) (−κ212 0 τ 0 st + ...,2κ 0 τ 0 (s + t)+...,−6κ 0 +2κ ′ 0 (s + t) − κ3 0 st + ...) ,α(s) × α(t)so A s,t =→ A =(0, 0, −1) as s, t → 0. Thus, the plane through P‖α(s) × α(t)‖with normal A is the osculat<strong>in</strong>g plane.114
SELECTED ANSWERS 1151.2.23 a. cont. Alternatively, let the equation of the plane through P , Q, <strong>and</strong> R beA s,t · x =0(where we choose A s,t to vary cont<strong>in</strong>uously with length 1). We wantto determ<strong>in</strong>e A = lim s,t→0 A s,t . For fixed s <strong>and</strong> t, consider the function F s,t (u) =A s,t · α(u). Then F s,t (0) = F s,t (s) =F s,t (t) =0, so, by the mean value theorem,there are ξ 1 <strong>and</strong> ξ 2 so that F s,t(ξ ′ 1 )=F s,t(ξ ′ 2 )=0,hence η so that F s,t(η) ′′ =0. NowF s,t(0) ′ = A s,t · T(0) <strong>and</strong> F s,t(0) ′′ = A s,t · κ 0 N(0). S<strong>in</strong>ce ξ i → 0 <strong>and</strong> η → 0ass, t → 0,we obta<strong>in</strong> A · T(0) = A · N(0) = 0, so A = ±B(0), as desired.1.3.4 Let L = length(C). Then by Theorem 3.5 we have 2π = ∫ L0 κ(s) ds ≤ ∫ L0cds = cL,so L ≥ 2π/c.2.1.2 a. E = a 2 , F =0,G = a 2 s<strong>in</strong> 2 u; d. E = G = a 2 cosh 2 u, F =02.1.4 Say all the normal l<strong>in</strong>es pass through the orig<strong>in</strong>. Then there is a function λ so thatx = λn. Differentiat<strong>in</strong>g, we have x u = λn u + λ u n <strong>and</strong> x v = λn v + λ v n. Dott<strong>in</strong>g withn, weget 0 = λ u = λ v . Therefore, λ is a constant <strong>and</strong> so ‖x‖ = const. Alternatively,from the statement x = λn we proceed as follows. S<strong>in</strong>ce n · x u = n · x v =0,wehavex · x u = x · x v =0. Therefore, (x · x) u =(x · x) v =0,so‖x‖ 2 is constant.2.1.6 We check that E = G =4/(1 + u 2 + v 2 ) 2 <strong>and</strong> F =0,sothe result follows fromExercise 5.2.1.7 b. One of these is: x(u, v) =(cos u + v s<strong>in</strong> u, s<strong>in</strong> u − v cos u, v).2.1.12 a. If a cosh(1/a) =R, the area is 2π ( a + R √ R 2 − a 2) .2.2.1 If u- <strong>and</strong> v-curves are l<strong>in</strong>es of curvature, then F =0(because pr<strong>in</strong>cipal directionsare orthogonal) <strong>and</strong> m = S(x u ) · x v = k 1 x u · x v =0. Conversely, sett<strong>in</strong>g S P (x u )=ax u + bx v ,we<strong>in</strong>fer that if F = m =0,then 0 = S P (x u ) · x v = Fa+ Gb = Gb, <strong>and</strong>so b =0. Therefore, x u (<strong>and</strong>, similarly, x v )isaneigenvector for S P . Moreover, ifS P (x u )=k 1 x u <strong>and</strong> S P (x v )=k 2 x v ,wedot with x u <strong>and</strong> x v , respectively, to obta<strong>in</strong>l = Ek 1 <strong>and</strong> n = Gk 2 .[]1/b 02.2.3 b. l = b, m = 0, n = cos u(a + b cos u), S P =,0 cos u/(a + b cos u)H = 1 12(b + cos u )a+b cos u , K =cos ub(a+b cos u) ; d. l = −a, m = 0, n = a, S P =[]−(1/a)sech 2 u 00 (1/a)sech 2 , H =0,K = −(1/a) 2 sech 4 u.u2.2.5 We know from Example 1 of Chapter 1, Section 2 that the pr<strong>in</strong>cipal normal of thehelix po<strong>in</strong>ts along the rul<strong>in</strong>g <strong>and</strong> is therefore orthogonal to n. Aswemove along arul<strong>in</strong>g, n twists <strong>in</strong> a plane orthogonal to the rul<strong>in</strong>g, so its directional derivative <strong>in</strong> thedirection of the rul<strong>in</strong>g is orthogonal to the rul<strong>in</strong>g.2.2.6 E = tanh 2 u, F =0,G = sech 2 u, −l = sech u tanh u = n, m =02.3.5 d. Γ v uv =Γ vvu = f ′ (u)/f(u), Γ uvv = −f(u)f ′ (u), all others 0.2.4.4 κ g = cot u 0 ;wecan also deduce this from Figure 3.1, as the curvature vector κN =(1/ s<strong>in</strong> u 0 )N has tangential component −(1/ s<strong>in</strong> u 0 ) cos u 0 x u = cot u 0 (n × T).