Analytical Mechanics This page intentionally left blank
Download 10.87 Mb. Pdf ko'rish
|
sn(u) = cn(u) dn(u), d du cn(u) = −sn(u) dn(u), d du dn(u) = −k 2 sn(u) cn(u). (A2.12)
If k = 0, the elliptic functions reduce to simple trigonometric functions (in this case ϕ = u), while for k = 1 the elliptic functions are no longer periodic and can be expressed through hyperbolic functions: k = 1,
sn(u) = tanh(u), cn(u) = dn(u) = 1 cosh(u)
. (A2.13)
The Jacobi elliptic functions, as functions of the complex variable u ∈ C, have the following complex periods: sn(u+2iK(k )) = sn(u), cn(u+2(K(k)+iK(k ))) = cn(u) and dn(u + 4iK(k )) = dn(u), where k = √ 1
2 . They are therefore an example of doubly periodic functions. For more information on elliptic functions and integrals, and for a more detailed study, we refer the reader to the beautiful classical books of Whittaker and Watson (1927) and Tricomi (1937). An interesting exposition of the history of elliptic functions can be found in Dieudonn´ e (1978, chapter 7). APPENDIX 3: SECOND FUNDAMENTAL FORM OF A SURFACE As seen in Section 1.6, the first fundamental form of a surface S expresses in the tangent space the notion of a scalar product of the Euclidean space in which the surface is embedded, and allows one to measure lengths, angles and areas. For planar curves the curvature measures how much the curve is far from being straight. To quantify how much a surface S in three-dimensional Euclidean space deviates from the tangent plane at one of its points P , one can study the unit normal vector of S in a neighbourhood of P . The second fundamental form of a surface, which we discuss here, expresses precisely the rate of change of the normal to the surface S for infinitesimal displacements on the surface. Since there exist two independent directions to move along the surface, the second fundamental form is a quadratic form. Let S be a regular surface, and x(u, v) be a local parametrisation. Let n be the normal unit vector n =
x u × x v |x u × x v | = x u × x v √ EG − F
2 . (A3.1) Consider a curve s → x(s) on the surface S parametrised by the arc length parameter s. Let t be the tangent unit vector to the curve and k(s) its curvature. The curvature vector of the curve k = dt
, (A3.2)
whose modulus is the curvature k(s), admits a unique decomposition k = k
n + k
g (A3.3)
into two vectors: the normal curvature vector k n = (k · n)n ∈ (T x(s) S)
, (A3.4)
and the geodesic curvature vector k g = k − k
n ∈ T
x(s) S. (A3.5) The modulus k g = |k g | is called the geodesic curvature of the curve. We observe that if the curve is a geodesic, then its geodesic curvature is zero. Since t
∈ T x(s)
S, it must be that t · n = 0, and hence by differentiation it follows that the normal curvature k n has the expression k n = k · n = − dx ds · dn ds . 710 Second fundamental form of a surface A3 On the other hand (ds) 2 = dx
· dx, and therefore k n = − dx · dn dx · dx . (A3.6)
Using the parametrisation of the surface, we have that dx = x
u du + x
v dv,
dn = n u du + n v dv,
(A3.7) where n
u = ∂n/∂u and n v = ∂n/∂v. Inserting the equations (A3.7) into the expression (A3.6) for the normal curvature we find k n = − (x u · n
u )(du)
2 + (x
u · n
v + x
v · n
u )(du)(dv) + (x v · n
v )(dv)
2 E(du)
2 + 2F (du)(dv) + G(dv) 2 .
D efinition A3.1 The numerator −dx · dn of (A3.8) is called the second fun- damental form of the surface S. It is a quadratic form on the tangent space to the surface S, given by −dx · dn = e(u, v)(du) 2 + 2f (u, v)(du)(dv) + g(u, v)(dv) 2 , (A3.9) where e(u, v) = −x u
u = x
uu · n,
2f (u, v) = −(x
u · n
v + x
v · n
u ) = 2x
uv · n,
g(u, v) = −x v · n v = x vv · n.
(A3.10) It is immediate to check that the following relations hold: e(u, v) = x uu · x u × x v √ EG − F 2 = 1 √ EG − F 2 x uu y uu z uu x u y u z u x v y v z v , f (u, v) = x uv · x u × x v √ EG − F 2 = 1 √ EG − F 2 x uv y uv z uv x u y u z u x v y v z v , g(u, v) = x vv · x u × x v √ EG − F 2 = 1 √ EG − F 2 x vv y vv z vv x u y u z u x v y v z v . (A3.11) Example A3.1 Consider the
sphere of radius r with
the parametrisation x =
2 = r 2 (du)
2 + r
2 cos
2 u(dv)
2 , and hence √ EG
2 = r
2 cos u. From the definition A3 Second fundamental form of a surface 711 of the normal unit vector it follows that n = −(cos u cos v, cos u sin v, sin u), and it is immediate to check that the second fundamental form is given by e = r, f = 0,
g = r cos 2 u. Remark A3.1 From (A3.8) it follows that the normal curvature k n depends only on the point P (of coordinates (u, v)) on the surface and on the tangent space T P S (determined by du/dv or by dv/du): all curves through a point P of the surface tangent to the same direction have the same normal curvature. We can hence study how the normal curvature k n varies as the direction in a fixed point of the surface varies. Since the first fundamental form is positive definite, the sign of the normal curvature k n depends only on the second fundamental form. There are three possible cases. (1) If at a point P of the surface eg − f 2
applied to different directions always has the same sign, and the point is then called elliptic; the centres of curvature of all the normal sections to the surface passing through the point P lie on the same side of the surface. This situation is satisfied, for example, at all points of a sphere or of an ellipsoid. (2) If eg −f 2 = 0, there exists a direction in which the normal curvature vanishes. The point is then called parabolic. An example is given by any point of a cylinder. (3) If eg −f 2
ies: the surface S crosses its tangent plane and the point is called hyperbolic. This is what happens if the point P is a saddle point. We now look for the directions along which the normal curvature has a max- imum or a minimum. A direction in the tangent space T P S to the surface at the point P is determined by λ = dv/du, and the expression of the normal curvature k n in terms of λ can be obtained immediately from equation (A3.8): k n = k n (λ) = e + 2f λ + gλ 2 E + 2F λ + Gλ 2 . (A3.12) Hence the condition for a maximum or a minimum follows from requiring that dk n dλ (λ) = 0,
i.e. 2(f + gλ)(E + 2F λ + Gλ 2 )
2 ) (E + 2F λ + Gλ 2 ) 2 = 0. (A3.13)
Since the first fundamental form is positive definite, the denominator of (A3.13) is never zero; the condition for the normal curvature to be stationary is (E + F λ)(f + gλ) = (e + f λ)(F + Gλ), (A3.14)
712 Second fundamental form of a surface A3 which when substituted into (A3.12) gives k n = (e + f λ) + λ(f + gλ) (E + F λ) + λ(F + Gλ) = f + gλ
F + Gλ = e + f λ E + F λ = e(du) + f (dv) E(du) + F (dv) = f (du) + g(dv) F (du) + G(dv) . (A3.15) Hence we find that the maximum and minimum values of k n are solutions of the system (e − k n E)(du) + (f − k n
(f − k
n F )(du) + (g − k n
(A3.16) and hence of the eigenvalue problem for the second fundamental form F II
to the first fundamental form F I : det(k
n F I − F II ) = Ek n − e F k n − f
F k n − f Gk n − g
= 0. (A3.17)
The maximum and minimum values of k n are given by the roots of the characteristic polynomial (EG
− F 2 )k 2 n − (eG + Eg − 2fF )k n + eg
− f 2 = 0. (A3.18) D efinition A3.2 The two roots k 1 and k
2 of (A3.18) are called the principal curvatures of the surface S at the point P . Moreover the mean curvature M is the arithmetic mean of the principal curvatures: M = k
+ k 2 2 = Eg + eG
− 2fF 2(EG
− F 2 ) , (A3.19)
while the Gaussian curvature K is defined as the square of the geometric mean of the principal curvatures: K = k 1
2 = eg − f 2 EG − F 2 . (A3.20) Note that on the basis of the latter formula the classification given in Remark A3.1 can be reformulated in terms of the sign of K. One can prove (cf. Dubrovin et al. 1991a) that the vanishing of the mean curvature characterises the minimal surfaces (i.e. the surfaces of minimal area). The Gaussian curvature measures how far the metric of the surface is from the Euclidean metric. Indeed, we have the following. T heorem A3.1 A necessary and sufficient condition for a surface to be isomet- ric to an open set of a Euclidean plane is that the Gaussian curvature K is identically zero. A3 Second fundamental form of a surface 713 Clearly, the second fundamental form, and consequently the Gaussian curvature, are defined independently of the first. However, Gauss proved that K is in fact determined by the first fundamental form. T heorem A3.2 (Egregium theorem of Gauss) The Gaussian curvature depends only on the first fundamental form and on its derivatives: K =
1 √ EG − F 2 ∂ ∂u F E
v − EG
u 2E √ EG − F
2 + ∂ ∂v 2EF
u − F E
u − EE
v 2E √ EG − F
2 , (A3.21) where E v = ∂E/∂v, G u = ∂G/∂u, etc. Remark A3.2 If the coordinate system u, v that parametrises the surface is orthogonal, and hence if F = 0, equation (A3.21) simplifies to K =
− 1 √ EG ∂ ∂u 1 √ E ∂ ∂u √ G + ∂ ∂v 1 √ G ∂ ∂v √ E . If also f = 0, then (A3.18) becomes EGk 2
− (gE + eG)k n + eg = 0, from which it follows that the principal curvatures are k 1 = e/E, k 2 = g/G
(corresponding in (A3.12) to the two cases λ = 0, λ → ∞), and hence M = eG + gE
2EG , K = eg EG . For a more detailed discussion of the theory of the curvature of a surface, and for its formulation on a Riemannian manifold, we refer the reader to the texts already cited. In addition, we recommend the survey article by Osserman (1990) which illustrates the various, fascinating developments of modern Riemannian geometry. Problems
1. Prove that the second fundamental form for surfaces of revolution, given by the parametrisation x = (u cos v, u sin v, ψ(u)) has coefficients e = ψ (u)
1 + (ψ (u)) 2 , f = 0, g =
uψ (u) 1 + (ψ (u)) 2 .
714 Second fundamental form of a surface A3 2. Compute the second fundamental form for the ellipsoid with the paramet- risation x = (a cos u cos v, b cos u sin v, c sin u), where a > b > c > 0. Verify that in the case a = b we again find the expression already derived for surfaces of revolution, and in the case a = b = c the formula derived for the sphere. 3. Prove that the second fundamental form for the torus parametrised by x = (cos v(1 + a cos u), sin v(1 + a cos u), a sin u), with 0 < a < 1, has coefficients given by
e = a, f = 0,
g = (1 + a cos u) cos u. 4. Compute the second fundamental form of the circular paraboloid x = (u cos v, u sin v, u 2 ). 5. Determine the elliptic, parabolic and hyperbolic points of the torus. 6. Compute the second fundamental form for a surface S which is the graph of the function ψ(x, y), and prove that its Gaussian curvature has value K =
∂ 2 ψ ∂x∂x ∂ 2 ψ ∂x∂y
∂ 2 ψ ∂y∂x ∂ 2 ψ ∂y∂y
1 + ∂ψ ∂x 2 + ∂ψ ∂y 2 2 . 7. Prove that the Gaussian curvature of an ellipsoid with semi-axes a, b, c is K = 1
2 b 2 c 2 x 2 a 4 + y 2 b 4 + z 2 c 4 2 . 8. Prove that the Gaussian curvature of a surface of revolution x = (u cos v, u sin v, ψ(u)) is given by K = ψ (u)ψ (u) u(1 + (ψ (u)) 2 ) 2 . For example, for the circular paraboloid ψ(u) = u 2 , we have K = 4
2 ) 2 , which vanishes in the limit u → ∞, in agreement with geometrical intuition. 9. Prove that the Gaussian curvature of the catenary x = u cos v, u sin v, c cosh −1 u c , where c > 0 is a fixed constant, is K = −c 2 /u 4 , and that the mean curvature is M = 0 (the catenary is an example of a ‘minimal surface’).
APPENDIX 4: ALGEBRAIC FORMS, DIFFERENTIAL FORMS, TENSORS The use of differential forms allows one to generalise to the case of manifolds of any dimension the ordinary concepts of work of a vector field along a path, of flow through a surface and in general the results of classical vector analysis. The use of differential forms is important for a deeper understanding of Hamiltonian mechanics (see Abraham and Marsden 1978, Arnol’d 1979a, and Meyer and Hall 1992), although in the present text we have avoided their use (except for differential 1-forms). In this appendix we limit ourselves to a brief introduction to the study of differential forms, and refer the interested reader to one of the numerous treatises on the subject (e.g. Flanders 1963, or the cited books of Abraham and Marsden and of Arnol’d) for a more detailed study and for the proofs we omit. In addition, we systematically adopt the repeated index summation convention (covariant and contravariant, below and above, respectively, following the classical notation). A4.1
Algebraic forms Let V be a real vector space of dimension l. D efinition A4.1 The dual space V ∗ of V is the space of all linear maps ϑ : V → R. The elements ϑ ∈ V ∗ are called covectors or (algebraic) 1-forms. It is immediate to check that V ∗ is a real vector space, and that dim V ∗ = dim V = l. The sum of two covectors ϑ 1 , ϑ
2 ∈ V
∗ is defined by the formula (ϑ 1
2 )(v) = ϑ
1 (v) + ϑ
2 (v),
(A4.1) for every v ∈ V , and the product with a real number λ yields (λϑ)(v) = λϑ(v). (A4.2) If e
1 , . . . , e l is any basis of V , we can associate with it the dual basis e 1∗ , . . . , e l ∗
∗ , defined by the conditions e i
(e j ) = δ i j = 1, if i = j, 0, otherwise, (A4.3) and every covector ϑ can be expressed through its components: ϑ = ϑ i
i ∗ . (A4.4) 716 Algebraic forms, differential forms, tensors A4.1 It is not difficult to check that if e 1 , . . . , e l is a new basis of V , and M is the l × l matrix whose entries M j i
j i of e i expressed in the basis e 1
l , we have e i
= A i k e k ∗ , (A4.5)
where A i k M k j = δ i j , i.e. A = (M T ) − 1 , and the components of the vectors v = v i e i = v
i e i and of the covectors ϑ = ϑ i e i ∗ = ϑ i e i ∗ are transformed according to the following rules: v i = A i j v j , ϑ i = M j i ϑ j . (A4.6) Because of this transformation property, the components v i of the vectors are called contravariant and the components ϑ i of the covectors are called covariant. Indeed, they are transformed, respectively, through the matrix A, the (transposed) inverse of the change of basis, and the matrix M of the change of basis. D efinition A4.2 An (algebraic) k-form is a map ω : V k → R, where V k =
Download 10.87 Mb. Do'stlaringiz bilan baham: |
ma'muriyatiga murojaat qiling