Page 397 - DMTH402_COMPLEX_ANALYSIS_AND_DIFFERENTIAL_GEOMETRY
P. 397

Complex Analysis and Differential Geometry




                    Notes          In other words, the principal normal  n  must be parallel to the normal N to the surface along the
                                                                
                                   curve segment from p to q.
                                   If C  is parameterized by arc length, this  means that  the acceleration must be normal to  the
                                   surface.
                                   It is then natural to define geodesics as those curves such that k  = 0 everywhere on their domain
                                                                                    g
                                   of definition.
                                   Actually, there is another  way of  defining geodesics in terms of vector fields and covariant
                                   derivatives, but for simplicity, we stick to the definition in terms of the geodesic curvature.

                                   Definition 1.  Given  a surface  X  :      E , a  geodesic  line,  or geodesic,  is  a  regular  curve
                                                                     3
                                   C : I  E  on X, such that k (t) = 0 for all t  I.
                                          3
                                                        g


                                                                        
                                     Notes    By regular curve, we mean that  C(t)  0  for all t  I, i.e., C is really a curve, and
                                     not a single point.
                                   Physically, a particle constrained to stay on the surface and not acted on by any force, once set in
                                   motion with  some non-null  initial velocity  (tangent  to the  surface),  will follow  a  geodesic
                                   (assuming no friction).
                                                               
                                   Since k  = 0 if the principal normal  n  to C at t is parallel to the normal N to the surface at X(u(t),
                                        g
                                                                 
                                   v(t)), and since the principal normal  n  is a linear combination of the tangent vector  C(t)  and
                                                                                                        
                                   the acceleration vector  C(t),  the normal N to the surface at t belongs to the osculating plane.
                                                      
                                   Since the tangential part of the curvature at a point is given by

                                                                                      
                                                       
                                                                                        ' 
                                                                       ' 
                                                     k g  g n     u    1 ij u u X    u    2 ij u u X ,
                                                                              "
                                                                                      '
                                                                     '
                                                                              2 
                                                             1 
                                                             "
                                                                       j
                                                                          u
                                                                       
                                                                     i
                                                                                        j
                                                                                      i
                                                                                           v
                                                                                         
                                                              i 1,2           i 1,2   
                                                                
                                                                                 
                                                               j 1,2          j 1,2   
                                                                                 
                                                                
                                   the differential equations for geodesics are
                                                                   1 
                                                                  "
                                                                           '
                                                                            '
                                                                 u      1 ij u u   0,
                                                                           i
                                                                            j
                                                                     i 1,2
                                                                     
                                                                     j 1,2
                                                                     
                                                                           '
                                                                   2 
                                                                  "
                                                                            '
                                                                 u      2 ij u u  0,
                                                                            j
                                                                           i
                                                                     i 1,2
                                                                     
                                                                     j 1,2
                                                                     
                                   or more explicitly (letting u = u  and v = u ),
                                                                     2
                                                            1
                                                                   2
                                                                                   2
                                                           u"   1 11 (u')   2 1 12 u'v'  1 22 (v')  0,
                                                                   2
                                                                                   2
                                                           v"   2 11 (u')   2 2 12 u'v'  2 22 (v')  0.
                                   In general, it is impossible to find closed-form solutions for these equations.
          390                               LOVELY PROFESSIONAL UNIVERSITY
   392   393   394   395   396   397   398   399   400   401   402