Elliptical Integrals and Functions

Transcription

1 Epilogue Everything created by humans has its beginning and its end. At the end of this book, we would like to say that this is just a first step towards a much deeper exposition of the topics discussed inside. First, because they are not treated with the completeness that we would like, and second, because there remain many examples in the physical world in which the elastica plays a principal role but which were not touched on here. We concretely have in mind the free rotations of a rigid body about a fixed point, the Dubins problem concerning the flight of airplanes, trajectories of a material ball rolling in the plane, the three-dimensional elasticas, and probably many others which we do not suspect at this time. We are, however, firmly convinced that glimpses of elastica are everywhere around us, and we have just to feel their presence and be struck again with admiration by this confluence of nature and geometry. We will certainly be thankful to everyone who sends us comments, remarks or suggestions. Springer International Publishing AG 17 I.M. Mladenov and M. Hadzhilazova, The Many Faces of Elastica, Forum for Interdisciplinary Mathematics 3, DOI 1.17/

2 Appendix A Elliptical Integrals and Functions Elliptic integrals and functions are mathematical objects, which nowadays are often omitted in the mathematical curricula of universities. One quite trivial explanation is the presence of plenty of efficient computational programs that can be implemented on modern computers. While the standard integration techniques allow us to obtain explicit expressions (in terms of trigonometric, exponential and logarithmic functions for every integral of the form R(x, P(xdx, where R(x, P(x is a rational function, and P(x is a linear or quadratic polynomial, we have to widen our vocabulary of elementary functions if we want to work with polynomials of higher degree. In particular, when P(x is a polynomial of the third or fourth degree, the corresponding function is called elliptic. Of course, when teaching calculus, one must stop somewhere, and it is reasonable to stay loyal to well-known linear and quadratic functions, while using numerical methods to calculate integrals of the third and fourth degree. The possibilities of easy-to-use computer systems for symbolic manipulation of the type represented by Maple R and Mathematica R makes this course of action even more understandable. The main point in this Appendix is that elliptic functions provide effective means for the description of geometric objects. The second is that the above-mentioned computer programs, through their built-in tools for calculation and visualization, are, in fact, a real motivation for the teaching and using of elliptic functions. In this Appendix, we will consider a few examples in order to prove that elliptic integrals and functions are necessary to get more interesting geometric and mechanical information than that given by direct numerical calculations. The history of the development of elliptic functions can be followed in Stillwell (1989. Clear statements of their properties and applications can be found in the books by Greenhill (1959, Hancock (1958, Bowman (1953 and Lawden (1989. A more recent approach to the problem from the viewpoint of dynamical systems is given by Meyer (1. Springer International Publishing AG 17 I.M. Mladenov and M. Hadzhilazova, The Many Faces of Elastica, Forum for Interdisciplinary Mathematics 3, DOI 1.17/

3 4 Appendix A: Elliptical Integrals and Functions A.1 Jacobian Elliptic Functions The easiest way to understand elliptic functions is to consider them as analogous to ordinary trigonometric functions. From the calculus, we know that arcsin(x = x 1 u Of course, if x = sin(t, π/ t π/, we will have t = arcsin(sin(t = sin(t 1 u (A.1 In this case, we can consider sin(t as the inverse function of the integral (A.1. The real understanding of trigonometric functions includes knowledge of their graphs, their connection with other trigonometric functions, such as in sin (θ + cos (θ = 1, and of course, the fundamental geometric and physical parameters in which they are included (i.e., circumferences and periodical movements. We will follow this example for elliptic functions too. Let us begin by fixing some k, k 1, which, from now on, will be called an elliptic molus and introce the following: Definition A.1 The Jacobi sine function sn(u, k is the inverse function of the following integral: sn(u,k dt u = 1 t 1 k t (A. More generally, we will call F(z, k = z dt 1 t 1 k t (A.3 the elliptic integral of the first kind. The elliptic integrals of the second and third kinds are defined by the equations E(z, k = Π(n, z, k = z z 1 k t dt 1 t dt (1 + nt (1 t (1 k t (A.4 When the argument z in F(z, k, E(z, k and Π(n, z, k is equal to one, these integrals are denoted, respectively, as K (k, E(k and Π(n, k and called complete elliptic integrals of the first, second and third kinds, respectively.

4 Appendix A: Elliptical Integrals and Functions 5 If we put t = sin φ, the above integrals are transformed, respectively, into F(φ, k = E(φ, k = Π(n, φ, k = φ φ φ dφ 1 k sin φ 1 k sin φ dφ (A.5 dφ (1 + n sin φ 1 k sin φ Let us note that when k 1, E(φ, 1 = sin φ, and therefore one can consider E(φ, k to be a generalization of the function sin φ. The Jacobi cosine function cn(u, k can be defined in terms of sn(u, k by means of the identity sn (u, k + cn (u, k = 1. (A.6 The third Jacobi elliptic function dn(u, k is defined by the equation dn (u, k + k sn (u, k = 1. (A.7 The integral definition of sn(u, k makes it clear that sn(u, = sin(u. Of course, cn(u, = cos(u as well. Besides sn, cn and dn, there are another nine functions that are widely used, and their definitions are given below: ns = 1 sn, nc = 1 cn, nd = 1 dn sc = sn cn, cs = cn sn, cn cd = dn, dn dc = cn, dn ds = sn sn sd = dn The derivatives of the elliptic functions can be found directly from their definitions (or vice versa, as in Meyer (1, where the elliptical functions are defined by their derivatives. For instance, the derivative of sn(u, k may be computed as follows. In (A.3, suppose that z = z(u. Then, df = df dz dz = 1 dz 1 z 1 k z

5 6 Appendix A: Elliptical Integrals and Functions But from (A. and (A.3, we know that for z = sn(u, k, wehavef(z, k = u. So, replacing z with sn(u, k and using / = 1, we obtain 1 1 = dsn(u, k 1 sn(u, k 1 k sn(u, k dsn(u, k = 1 sn(u, k 1 k sn(u, k (A.8 dsn(u, k = cn(u, k dn(u, k. After differentiation to (A.6 with respect of u and taking into account (A.8, we obtain dcn(u, k = sn(u, k dn(u, k, (A.9 Finally, after differentiating (A.7 and using (A.8 once more, we have ddn(u, k = k sn(u, k cn(u, k. (A.1 Symbolic computational programs such as Maple R or Mathematica R have embedded moles for working with elliptic functions, so these functions can be easily drawn. Graphs of the elliptic sin function sn, cos function cn and function dn are shown in Fig. A.1. We can see that sn(u, k and cn(u, k are periodic. We can define their period referring to the definitions above (see A. K (k = 1 dt 1 t 1 k t We can see that sn(k (k, k = 1. Obviously, from the graph, we are also convinced that K (k is 1/4 ofthesn(u, k period and that the period of dn(u, k is K (k. Of course, this can be checked analytically (e.g., see Woods (1934, p. 368, but this argument satisfies our objectives. Fig. A.1 Graphs of the elliptic sin function sn(u, k, elliptic cos function cn(u, k and the function dn(u, k drawn with k = 1

6 Appendix A: Elliptical Integrals and Functions 7 Using the computer program, we can look for a numerical solution, from which we can find K (k, i.e., to solve the equation sn(u, k = 1. Note that the equation sn (u, k + cn (u, k = 1 supposes that cn(u, k has the same period as sn(u, k, and therefore cn(k (k, k =. Now we have an idea of the algebraic and graphic properties of the elliptic functions. In order to complement our understanding, let us look at two simple examples one physical and one geometrical. Example A.1 (Penlum Let the angle of a penlum swing be denoted by x. Then, it is straightforward to derive that the equation of motion is: ẍ +(g/l sin(x =, where g is the acceleration e to gravity and l is the length of the penlum. If we take such units that give g/l = 1, the penlum equation becomes ẍ + sin(x =. Then, we can multiply by ẋ to obtain ẋ(ẍ + sin(x = ẋẍ + sin(x ẋ = ( x ( x ẋ ẋẍ + 4sin cos =, and, by integrating the last equation, end up with 1 ( ẋ + sin x = c. (A.11 Note also that because of the identity sin (x/ = 1 cos(x, the last equation expresses the conservation of energy of the motion of a particle with a unit mass. Now let z = sin(x/, and therefore ż = cos(x/ẋ = 1 z ẋ. Then, ( 4ż = (1 z ẋ =ẋ sin x ( ẋ =ẋ cos x ż = 1 ( 4 ẋ cos x. By the first part of the calculation, we have 1 4 ẋ = 1 ( c x sin = 1 ( c z and cos x = 1 z. Hence, ż = (A z (1 z, where A = c/. Taking a square root and separating the variables gives us t = z dz Au (A z (1 z = (1 u (1 Au = F( Au, A.

7 8 Appendix A: Elliptical Integrals and Functions That is, we see that the elliptic integrals appear even in the most standard of mechanical situations. Example A. (Ellipse Let us parameterize the ellipse by the polar angle, which we will denote by t, i.e., α(t = (x(t, z(t = (a sin(t, c cos(t, where t π, and a c. Then, the arclength integral is L = π ẋ +ż dt = 4 π/ = 4a π a cos (t + c sin (t dt 1 ε sin (t dt, in which ε= a c /a is the eccentricity of the ellipse. If we substitute sin(t=u, then dt = / 1 u, and in this way, we obtain 1 1 ε u L = 4a = 4a E(ε. 1 u So, we have been convinced once more that the elliptic integrals present themselves even in the most natural geometric problems. All Jacobian elliptic functions have integrals, given by the formulas below, that can be verified by direct differentiations snu = 1 ln(dnu kcnu, k dnu = arcsin(snu, ncu = 1 k ln(dcu + kscu, scu = 1 k ln(dcu + kncu, dsu = ln(nsu csu, dcu = ln(ncu + scu, cnu = 1 k arcsin(ksnu nsu = ln(dsu + csu n = arcsin(c 1 k c = 1 ln(ncu + scu k csu = ln(nsu + dsu s = 1 k k arcsin(kc.

8 Appendix A: Elliptical Integrals and Functions 9 A. Weierstrassian Elliptic Functions Let us consider the elliptic integral of the first kind in the Weierstrassian approach u = z dz 4z3 g z g 3, (A.1 in which g and g 3 are arbitrary complex numbers. It can be proven that the above integral defines z as a unique function of u, denoted as (u, g, g 3, in which u is the argument and g and g 3 are the so-called invariants of the function. One should note that if the discriminant Δ = g 3 7g 3 of the cubic polynomial under the square root (A.1 is positive, then it takes real values for real values of z. Besides (u, Weierstrass introced another two functions-ζ(u and σ(u.they are defined by the equalities and ζ(u = 1 u u ( (ũ 1 dũ (A.13 ũ σ(u = u exp( u (ζ(ũ 1 ũ dũ. (A.14 Of the many interesting properties of the Weierstrassian functions, we will mention only those which have some direct relations to the applications used in the present text, namely, ζ (u = (u, σ (u σ(u = ζ(u ( u = (u, ζ( u = ζ(u, σ( u = σ(u. (A.15 References F. Bowman, Introction to Elliptic Functions with Applications (English Universities Press, London, 1953 A. Greenhill, The Applications of Elliptic Functions (Dover, New York, 1959 H. Hancock, Elliptic Integrals (Dover, New York, 1958 D. Lawden, Elliptic Functions and Applications (Springer, New York, 1989 K. Meyer, Jacobi elliptic functions from a dynamical systems point of view. Am. Math. Month. Am. 18, (1 J. Stillwell, Mathematics and its History (Springer, New York, 1989 F. Woods, Advanced Calculus (Ginn and Co., Boston, 1934

9 Index A Algebraic curve, 3 Amplitude function, 97 Antisymmetric matrix, 3 Artificial membrane, 7 E Elliptic molus, 49 Enzymes, 75 Essential equations, 4 Euler s elasticas equation, 114 B Bilayer lipid membranes, 8 Binormal, 16 Black bilayer, 71 F First fundamental form, 6 Fluid mosaic model, 76 Free elasticas equation, 114 Frenet-Serret formulas, C Cassini ovals, 6 Cauchy Riemann condition, 166 Christoffel symbols, 3 Codazzi-Mainardi-Peterson equations, 3 Coefficients of the first fundamental form, 6 Coefficients of the second fundamental form, 8 Co-lemniscate, 5 Co-moving frame, 48 Complementary elliptic molus, 173 Contact plane, 14 Covariant derivative, 158 Curvature, 3 Curvature radius, 4 Curvature vector, 5 Curvilinear coordinates, 5 G Gauss-Codazzi-Minardi equations, 3 Gauss curvature, 9 Gauss map, 9, 185 Generalised elasticas equation, 188 Glycocalyx, 7 I Infinitesimal operator, 165 Inflection point, 1 Integral membrane proteins, 75 Intrinsic equation of the curve, 6 Intrinsic equations, 4 Invariants of, 11 J Jacobian elliptic functions, 97 D Determining system of equations, 165 K Kronecker symbol, 158 Springer International Publishing AG 17 I.M. Mladenov and M. Hadzhilazova, The Many Faces of Elastica, Forum for Interdisciplinary Mathematics 3, DOI 1.17/

10 1 Index L Laplace-Beltrami operator, 16 Laplace-Young equation, 84 Lie equations, 164 Lie group generator, 165 Lipids, 73 Logarithmic spiral, 58 M Mean curvature, 9 Membrane, 69 Model membrane, 71 Monge parameterization, 43 Moving frame, 16 N Native membranes, 7 Natural equations, 4 Natural membrane, 7 Natural parameter, 3 Nodary, 97 Normal curvature, 33 Normal plane, 8 Normal vector, 6 O Orthogonal matrix, 3 Oscillating amplitude, 175 Osculating plane, 14 P Peripheral membrane proteins, 75 Plasma membrane, 7 Principal curvatures, 33 Principal normal line, 13 Principal unit normal vector, 13 R Receptors, 75 Reconstructed membrane, 7 Rectifying, 16 Regular point, 6 S Second curvature, 18 Second fundamental form, 8 Serret s curves, 59 Spontaneous curvature, 153 Structural proteins, 75 T Tangent, 8 Torsion, 18 Transcendental curve, 3 Transport proteins, 75 U Unlary, 97 Unit binormal vector, 16 Unit tangent vector, 7 V Vector field, 165 W Weingarten equations, 31