Scribd
Upload
133 views14 pages

Differential Geometry in Physics, 4 of 4 (Gabriel Lugo)

Phy in Geo

Uploaded by

Pharyanshu Kachhawa
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
133 views14 pages

Differential Geometry in Physics, 4 of 4 (Gabriel Lugo)

Phy in Geo

Uploaded by

Pharyanshu Kachhawa
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 14
Chapter 4 Surfaces in R® 4.1 Manifolds 4.1 Definition A C® coordinate chart is aC map # from an open subset of R? into R® x:UeR? —) R* (u,v) P24 (x(u,v) ulus), 2(u, 0) (4.1) We will always assume that the Jacobian of the map has maximal rank. In local coordinates, a coordinate chart is represented by three equations in two variables ws f'(ue), where =1,2,8, @=1,2 2) The local coordinate representation allows us to use the tensor index formalism introduced in earl chapters. The assumption the Jacobian J = (2x /2u") be of maximal rank, allows one to evoke the implicit function theorem. ‘Thus, in principle, one ean locally solve for one of the coordinates, say = in terms of the other two = f(e!27) (43) The locus of points in R® satisfying the equations x* = f*(u9), ean also be locally represented by an expression of the form P(zh, 2? 29) =0 (4a) ot x(u!,w2) : U > RE and y(v!, v?) : V — R® be two coordinate charts with a non empty intersection (x(U) N y(V)} # B. The two charts are said to be C™ equivalent if the map 6 = y~!x and its inverse @~ (see fig 4.1 Jaro infinitely differentiable, In more lucid terms, the def just states that two equivalent. charts x(w*) and y(v*) repre- sent different reparametrizations for the same set of points in R® Figure 4.1: Chart Equivalence 4B a CHAPTER 4, THEORY OF SURFACES 4.3. Definition A differentiably smooth surface in R is a set of points M in R® such that 1. If p © M then p belongs to some C® chart. 2. Ifp EM belongs to two different charts x and y, then the two charts are C™ equivalent. Intuitively, we may think of a surface as consisting locally of number of patches "sewn” to each other so as to form a quilt from a global perspective. “The first condition in the definition states that each local patch looks like a piece of R?, whereas the second differ y condition indicates that the patches are joined together smoothly. An other way to state this idea is 1o say that a surface a space that is locally Euclidean and it has a differentinble structure so that the notion of differentiation makes sense. If the Euclidean space is of dimension n, the *surface” is ealled an n- 4.6. bdf The element of arclength ds? = gagdu" @ du? (49) is also called the first fundamental form. We must caution the reader that this quantity is not a form in the sense of differential geometry since ds? involves the symmetric tensor product rather than the wedge product. The first fundamental form plays such a crucial role in the theory of surfaces, that will find it convenient to introduce yet a third more modeen version. Following the same development as in the theory of curves, consider a surface .M defined locally by a function (u!, 2) a(u!,u2). We say that a quantity Xp is a tangent vector at a point p € M, if Xp is a linear derivation on the space of C® real-valued fimetions {ff :.M —+ R} on the surface. ‘The set of all tangent vectors at a point As before, a vector field X’on th choice of a tangent vector at each point on the surface and the union of all tangent spaces is called the tangent bundle TM. ‘The coordinate chart map surface is a smooth a:R?— MeR® 46 CHAPTER 4. THEORY OF SURFACES induces a push-forward map ya TM defined by (VA) laguey= V(@ 2S) law Just as in the case of curves, when we revert back to classical notation to describe a surface as 2i(u'), what we really mean is (20.4)(u®), where « are the coordinate functions in R® . Particular examples of tangent vectors on M are given by the push-forward of the standard basis of TR®. ‘These tangent vectors which earlier we called xq are defined by a a Gra MA) baer Gast ef) he In this formalism, the frst fundamental form I is just the symmetric bilinear tensor defined by induced metric WAXY where X and Y are any pair of vector fields in =< XY >, (4.10) Orthogonal Parametric Curves Let V and W be vectors tangent to a surface M defined locally by a chart x(u*). Since the vectors Xaipha Span the tangent space of M at each point, the vectors Vand IW" can be written as linear combinations Vex. Wx, ‘The functions V* and 1V* are called the curvilinear coordinates of the vectors, We can caleulate the length and the inner product of the vectors using the induced Riemann netic, wie = = WOVE < xa x9 > IWIF = gaa¥°V IWIP = gop’, and < Vx, WExa >= VOW < xa, X3 > gaaVW?. ‘The angle @ subtended hy the the vectors v and Wis the given by the equation TT WW) JV, VVW im GV VE Goa WF cos = Let u® = 6(t) and wu" (8) be two curves on the surface «8 and jue oY aun = at, and bu = Sa 4.2, THE FIRST FUNDAMENTAL FORM a7 represent infinitesimal tangent vectors (1.12) to the curves. ‘Thus, the angle between two infinitesimal vectors tangent to two intersecting eurves on the surface satisties dagu bul [Gap bid Gash BuP In particular, if the two curves happen to be the parametric curves, w! = constant and 1 constant then along one curve we have du! = 0, du? arbitrary, and along the second we have dur arbitrary and du? = 0. In this case, the cosine of the angle subtended by the infinitesimal tangent vectors reduces to cos = (4.12) gioduldu? Joist) Joaatrye ‘As a result, we have the following (4.13) 4.7 Proposition ‘The parametric lines are orthogonal if P = 4.8 Examples a) Sphere x = (asind cos, asind sind, acos6) xe = (acos cos, acosésing, —asin®) xs = (-asin#sind,asinBeos 6,) B= xeexpsa? Fos xe-xy G = x, ds? = ade? +a" sin? 0dé? b) Surface of Revolution x = (reas@,rsin8, f(r) x, = (cos 0,sind, (0) Xe = (-rsind,rcos?,0) Es xox = 14+ f%r) F = x -x=0 G = mar dst = [Lt fel]? + ae? «} Pseudosphere = (sin weos asin wsin v,a(cos wintans)) E = abeot?u F G = asin?u ds? = a’ cot? udu? + a) Torus 48 CHAPTER 4, THEORY OF SURFACES x = ({b+acos u) cosy, (b+ acos u) sin v, asin u) eee ae a0) G = (b+acosu)® ds? = a®du? + (b+ acosu)*dv ©) Helicoid x = (weose, usin v,av) el reo wee? du + (u? + aye? 1) Catenoid an " dst = 4.3. The Second Fundamental Form Let x = x(u%) be a coordinate patch on a surface .M. Since x, and xv are tangential to the surface, ‘we can construct @ unit normal n to the surface by taking ne BRR 7 Ie xxl om Now, consider a curve on the surface given by u* = 1%(s). Without loss of generality, we assu that the curve is parametrized by arclength s so that the curve has unit speed. Using the chain rule, we se that the unit tangent vector 7’ to the curve is given by dx _ dx du® _ du* de te ds ds (4.15) Since the curve lives on the surface and the the vector 7’ is tangent to the curve, it is elear that T is also tangent to the surface. On the other hand, the vector 7” = d7'/ds does not in general have this property, so what we will do is to decompose 7 into its normal and tangential components (see fig (42) To = Kytky inom + By (4.16) where Ky = [Kol] =< Tm > The scalar quantity sy is called the normal curvature of the curve and Ky is called the geodesic curvature vector, The normal curvature measures the the curvature of x(u°(s)) resulting, 4.3, THE SECOND FUNDAMEN’ AL FORM 49 Figure 4.2: Normal Curvature by the constraint of the curve to lie on a surface. ‘The geodesic curvature vector , measures the “sideward” component of the curvature in the tangent plane to the surface. Thus, if one draws a straight line on a flat piece of paper and then smoothly bend the paper into a surface, then the straight line would now acquire some curvature. Since the line was originally straight, there is no sideward component of curvature so Ky = 0 in this case. This means that the entire contribution to the curvature comes from the normal component, reflecting the fact that the only reason there is curvature here is due to the bend in the surface itself Similarly, ifone specifies a point.p € M and a direction vector Xy € T).M, one can geometrically envision the normal curvature by considering the equivalence class of all unit speed curves in M which contain the point p and whose tangent vectors line up with the direction of X. Of course, there are infinitely many such curves, but at an infinitesimal level, all these curves can be obtained by intersecting the surface with a “vertical” plane containing the vector X and the normal to M Alll curves in this equivalence class have the same normal curvature and their geodesic curvatures vanish. In this sense, the normal curvature is more of a property pertaining to a direction on the face at a point, whereas, the geodesic curvature really depends on the curve itself. It might be impossible for a hiker walking on the ondulating hills of the Ozarks to find a straight line trail, since the rolling hills of the terrain extend in all directions, However, it might be possible to walk on a path with zero geodesic curvature as long as the hiker can maintain the same compass directio ‘To find an explicit formula for the normal curvature we first differentiate equation (4.15) ‘Taking the inner product of the last equation with the normal and noticing that <>xq,m >= 0, we get ar good where bog =< Xagm > (4.18) 4.9 Definition ‘The expression TT = bagdu® du (4.19) is called the second fundamental form 50 CHAPTER 4, THEORY OF SURFACES 4.10 Proposition The second fundamental form is symmeteic. Proof: In the classical formulation of the second fundamental form the proof is trivial. We have Jaa since for a C™ patch x(u2), we have Xa = Xsq because the partial derivatives co We will denote the coefficients of the second fundamental form by Daa te. 6 = bn fF = bie =< xuem> 0 that equation (4.19) can be written as IT = edu? + 2fdude + gdv (4.20) and equation (4.17) as 1 _ Edu? +2Fdudv + Gav? 80 T= eee Bilao + ge a We would also like to point out that just as the first fundamental form can be represented as T =< dx,dx>, so can we represent the second fundamental form as I] =— ‘To see this it suffices to note that differentiation of the identity < xq, >= 0 implies that < Xap M >= — < Xamy > ‘Therefore, = = udu? = =< Xqp,m> dude" = - From a computational point a view, a more useful formula for the coefficients of the second fundamental formula can be derived by first applying the classical vector identity (Ax B)-(C xD) ee Fa (4.22) ‘ocompute Ix eff? (xu Xo) = (Ru Hy) = «(EE OE] oe (4.23) Consequently, the normal vector can be written as Tel” VEG= 4d, CURVATURE BL ‘Thus, we can write the coefficients bay directly as triple products involving derivatives of (x). ‘The expressions for these coefficients are p= HeXku JEG— (x Xe) ; oe ee (424) ‘The first funclamental form on a surface measures the (squate) of the dlistance hetween two infinitesimally separated points. ‘There is a similar interpretation of the second fundamental form as we show below. The second fundamental form measures the distance from a point on the surface to the tangent plane at a second infinitesimally separated point. To see this simple geometrical interpretation, consider a point xp = x(ug) €.M and a nearby point x(uj + du). Expanding on a Taylor series, we get x(ug + du") Xo + (xa)adu + Hx) optud! + We recall that the distance formula from a point x to a plane which contains xy is just the scalar projection of (x — xo) onto the normal. Since the normal to the plane at xq is the same as the unit normal to the surface and < x,,m >= 0, we find that the distance D is D = = 1 (xojaprm> dudu? = F< (oagm > dre 1 = lle [he first fundamental form (or rather, its determinant) also appears in calculus in the context of calculating the area of a parametrized surface. the reason is that if one considers an infinitesimal parallelogram subtended by the vectors x,du and x,dv, then the differential of surface area is given by the length of the cross product of these two infinitesimal tangent vectors. That is dS = Ijxy x x|| dude [ [ VEGF aa ‘The second fundamental form contains information about the shape of the surface at a point. For example, the discussion above indicates that if 6 = |bas| = eg — f2 > 0 then all the neighboring points lie on the same side of the tangent plane, and hence, the surface is concave in one direction, If at a point on a surface 6 > 0, the point is called an elliptic point, if 6 < 0, the point is called hyperbolic or a saddle point, and if b= 0, the point is called parabolic 4.4 Curvature Curvature and all related questions which surround curvature, constitute the central object of study in differential geometry. One would like to be able to answer questions such as, what quantities retain invariant as one surface is smoothly changed into another? ‘There is certainly something intrinsically different from a cone, which we can construct from a flat piece of paper and a sphere which we can not, What is it that makes these two surfaces so different? How does one caleulate the shortest path between two objects when the path is constrained to be on a surface? 52 CHAPTER 4, THEORY OF SURFACES ‘These and many other questions of similar type can be quantitatively answered through the study curvature, We cannot overstate the great importance of this subject; perhaps it suffices to say that without a clear understanding of curvature, there would not be a general theory of relativity, xno concept of black holes, and even more disastrous, no Star Trek ‘The study of curvature of a hypersurface in R” (a surface of dimension n — 1) begins by trying to understand the covariant derivative of the normal to the surface. The reason is simple, If the normal to a surface is constant, then the surface is a flat hyperplane, ‘Thus, it is variations in the normal that indicate the presence of curvature, For simplicity, we constrain our discussion to surfaces in R® , but the formalism we use is applicable to any dimension, We will also introduce in this section the modern version of the second fundamental form 4.11 Definition Let X be a vector field on a surface M in R® , and let.’ be the normal vector. ‘The map L given by LX =-VxNV (4.25) is called the Gauss map In this definition we will be careful to differentiate between operators which ive on the surface and operators which live in the ambient space. We will adopt the convention of overlining objects hich live in the ambient space, the operator ¥ above being an example of one such object. ‘The Gauss map is clearly a good place to start, since it is the rate of change ofthe normal in an arbitrary direction which we wish to quantify 4.12 Definition ‘The Lie bracket [¥',¥] of two vector fields X and ¥ on a surface M is defined as the commutator [X.Y]=XY =YX, (4.26) (YA) = YX). 413 The Lie bracket of two vectors X,¥ € T(ealM) is another vector in T(M). Proof: If suffices to prove that the bracket is a linear derivation on the space of C™ fu Consider vectors X,Y, Z € T(calM) and smooth functions fg in M. ‘Then IY HZ) = XU +2) - 0 42K) XW UN + 210) = VX) = 20K) X(V(A) = YX) + XZ) = ZK) POY] +X. 2100, meaning that if f is a function on M then [X,Y] and X+¥, 2) (X + Y) ZN) — ZUX + YN) X(Z(D)) + YVAN) ~ 2X) ~ 2D) N(Z(f) ~ Z(X(N) + (ZU) — ZV) 1X, 216) + 2100) so the bracket is linear on each slot. Furthermore (Y)S +a) = XY(E +9) -YIXUS +9) X(V(N) + ¥(a)} — V(X) + X(@)) X(Y (EY) — YONA) + XY (a) — YEN) = YF Io), 4d, CURVATURE 53 (WX. ¥]tf9) X{Y (Fay) — Y(X(F9)) XTFY (9) + 9¥ (4) — YUFX(9) + 9X) X (SPY (g) + FX(Y (9) + XY) + XUV) HYD (9) — V(X) — YX) = 9 (XU) = AX) — OX) + XO) — YX = FX Ila) + 9X. YI, «that the bracket acts as linear derivation of functions. 4.14 Proposition ‘The Gauss map isa linear transformation on 7(M), Proof: Linearity follows from the linearity of ©, s0 it suffices to show that Ls X —> LX maps X € T(M) to a vector LX € T(M). Since NV is the unit normal to the surface, < N,V >= 1, 30 that any derivative of < N,N > is. Assuming that the connection is compatible with the metric, Vx + 2 = 2=0. Therefore, LX is orthogonal to V, and hence it lies in TM). We recall at this point that the in the previous section we gave two equivalent definitions < dx,dx >, and < X,¥ >) of the first fundamental form. We will now do the same for the second fundamental form. 4.15 Definition ‘The second fundamental form is the bilinear map, I(X,¥) =< EXY > (427) 4.16 Remark It should be noted that the two definitions of the second fundamental form are consistent, This is easy to see if one chooses X to have components x, and Y to have components x». With these choices, LX has components —n, and 11(X,¥) becomes bag = — < Xa,ms > We also note that there is third funclamental form defined by TIX, Y) =< LX,LY >=< LEXY > (4.28) In classical notation, the thitd fundamental form would be denoted by < dn, dn >. As one would expect, the third fundamental form contains third order ‘Taylor series information about the surface. We will not treat 11/(X,¥) in much detail in this work. 4.17 Definition ‘The torsion of a connection V is the operator T such that YX.Y T(X,Y) = Fx — Pv X — [X,Y] (4.29) A connection is called torsion free if T(X,Y) = 0. In this case, UxY PrN =1N,¥) We will say much more later about the torsion and the importance of torsion free connections. For the time being, it suffices to assume that for the rest of this section, all connections are torsion fice, Using this assumption, it is possible to prove the following very important theorem, or CHAPTER 4, THEORY OF SURFACES 4.18 Theorem The Gauss map is a self adjoint operator on TM. Proof: We have already shown that L: TM —+ TM is a linear map. Recall that an operator Lona linear space is self adjoint if < LX,¥ >=< X,LY >, so that the theorem is equivalent to proving that that the second fundamental for is symmetric (/1[X,Y] = 1[¥,X]). Computing the difference of these two quantities, we get MIX,Y]- HX] = -< LY,X> <¥xNY>-< Ty 5s Since < X,N >=< ¥,N 0, and the connection is compatible with the metric, we know that = = hence MX,Y]- 1.x] = -, 0 Gif [X,¥1E TM) One of the most important topics in an introductory course linear algebra deals with the spectrum of self adjoint operators. ‘The main result in this area states that if one considers the eigenvalue equation LX 6X (4.30) then the eigenvalues are always real and eigenvectors corresponding to different eigenvalues are or ‘thogonal. In the current situation, the veetor spaces in question are the tangent spaces at each point of a surface in R® , so the dimension is 2. Hence, we expect two eigenvalues and two eigenvectors LX, Xs mX (4.31) wiX2, (4.32) 4.19 Definition ‘The eigenvalues ¢y and sy of the Gauss map L are called the prineipal eur vatures and the eigenvectors Xi and X» are called the principal directions. -veral possible situations may occur depending on the classification of the eigenvalues at each point p on the surface: 1. Ie: # xy and both eigenvalues are posi ve, then p is ealled an elliptie point 2 Weepee <0, then p is called a hyperbolic point, 3 ey 12 £0, then p is called an umbilic point. 4. iff mW» = 0, then p is called a parabolic point It is also well known from linear algebra, that the the determinant and the trace of a self adjoint operator are the only invariants under a adjoint (similarity) transformation. Clearly these invariants are important in the case of the operator L, and they deserve to be given special names. 4.20 Definition ‘The determinant & = det() is called the Gaussian curvature of M and H = (1/2)Tr(L) is called the mean curvature. 4d, CURVATURE 55 Since any self-adjoint operator is diagonalizable and in a diagonal basis, the matrix repres Lis ding(c, m2) follows immediately that K = mm Ho= Heston) (4.38) 4.21 Proposition Let X and Y be any linearly independent vectors in 7M). ‘Th LX x LY (LX XY) + (X x LY) K(X «¥) 2H(X x) (4.38) Proof Since LX, LY € T(M), they can be expresses as linear combinations of the basis vectors X and Y IX = aX+hY LY = aX +be¥ computing the cross product, we get a by LX & LY oo ee (EX XY) + (Xx LY) = (a tho\(X x) det{E)(X x ¥). Similarly ‘THEY(X *Y) (@2HVIX x) 4.22 Proposition K = Hn (438) 2 Proof: Starling with equations (4.34) take ¢ vector identity (1.22). We immediately get 36 CHAPTER 4, THEORY OF SURFACES ‘The result follows by taking X =x, and 4.23 Theorem (Buler) Let X; and Xz be unit eigenvectors of L and let X = (cos )X1+ (sin 8)Xa, ‘Then in? IN(X,X) Prooft Easy, Just compute 11(X,X) =< LX, >, using the fact the LX, = wiX1 , LY: and noting that the eigenvectors are orthogonal. We get, 1608? Ow: X > = < (cond) Xi + (sinO)e2N2, (c098)X1 + (sin 4).X2 > = 1008? < Xi, Ni > teas = pcos? 0+ Kysin? a,

You might also like