Page 66 - Basic Structured Grid Generation
P. 66

Differential geometry of surfaces in E 3  55

                          Hence, with γ in place of α, eqn (3.67) becomes

                                         1 ∂a αβ  α β  d     1  β
                                               ˙ u ˙u −     a γβ ˙u  = 0,  γ = 1, 2.       (3.68)
                                        2f ∂u γ       dt  f
                          A simpler form of these equations results if we express the parametric form of the
                        solution curve in terms of arc-length s rather than t. Along a solution curve we now
                        have f = 1, a constant, by eqn (3.36). Thus eqn (3.68) becomes

                                                                   α
                                             d     du β     1 ∂a αβ du du β
                                                a γβ     −               = 0,              (3.69)
                                            ds      ds     2 ∂u γ  ds ds
                        that is,
                                                                        α
                                                        α
                                           2 β
                                          d u    ∂a γβ du du β  1 ∂a αβ du du β
                                       a γβ  2  +   α         −     γ         = 0,
                                           ds     ∂u   ds ds    2 ∂u   ds ds
                        which may be expressed as
                                          2 β
                                                                          β
                                         d u    1     ∂a γβ  ∂a γα  ∂a βα     du du α
                                      a γβ  2  +  2   α  +   β  −   γ          = 0,
                                         ds        ∂u     ∂u      ∂u    ds ds
                                                  β
                                                          α
                        exploiting the symmetry of (du /ds)(du /ds) with respect to β and α.Inother words,
                        using eqn (3.49),
                                                            β
                                              2 β
                                             d u          du du α
                                         a γβ  2  +[βα, γ ]      = 0,   γ = 1, 2.          (3.70)
                                             ds           ds ds
                        Re-writing α as δ, multiplying through by a αγ  (implying summation over γ ), and
                        using eqn (3.50), now gives
                                                β
                              2 α
                                                        2 α
                                                                    β
                             d u     αγ      du du δ   d u     α  du du δ
                                  + a  [βδ, γ ]      =      +   βδ       = 0,  α = 1, 2.   (3.71)
                              ds 2            ds ds     ds 2      ds ds
                          Equations (3.70) and (3.71) are second-order differential equations which define
                        geodesic curves, curves whose length is stationary (rather than necessarily being a
                        minimum) with respect to small variations given two fixed end-points. If we are given
                                                 α
                                         α
                        starting values of u and du /ds at a point on a surface, we can in principle solve
                        either of these pairs of equations and obtain a geodesic curve.
                          By the two-dimensional version of eqn (2.32), we can express eqn (3.71) in the
                        simple form
                                                      δ     du α
                                                               = 0.                        (3.72)
                                                      δs   ds
                          In practice eqns (3.71) may be quite complicated. For the surface of revolution as
                        defined by eqn (3.32) they become, using eqn (3.54),
                               2
                              d u      2   2 −1              du    2   2    2     dv    2


                                 + (f  + g )   (f f + g g )       − (f  + g )ff        = 0,
                              ds 2                           ds                   ds
                                                                2
                                                               d v            du      dv
                                                                         f
                                                                   + 2f  −1            = 0.
                                                               ds 2          ds    ds
                                                                                           (3.73)
   61   62   63   64   65   66   67   68   69   70   71