Page 47 -
P. 47

dominated convergence theorem  Let {f }     If V is finite dimensional, then dim(V ) =
                                                     n
                                1
                 be a sequence in L such that f → f almost  dim(V ). See also dual basis.
                                                                ∗
                                           n
                 everywhere and there exists a nonnegative g ∈
                  1
                 L such that |f |≤ g almost everywhere for all
                             n
                             1
                 n. Then f ∈ L and  f = lim     f .
                                                 n
                                          n→∞              duality techniques (Aubin-Nitsche trick)
                                                           These are used to establish a priori estimates
                 donor    A compound which breaks a chem-  for the discretization error of finite element
                 ical bond, yielding a substituent group which  schemes in norms weaker than the natural norms
                 forms a new bond in a bimolecular chemical or  associated with the continuous variational
                 biochemical reaction. See acceptor.       problem.  For example, let H, V be Hilbert
                                                           spaces, V continuously embedded in H, and
                                                           a : V × V → C a continuous and V-elliptic
                 dot product   See angle between vectors.
                                                           sesqui-linear form. Write u ∈ V, u ∈ V for
                                                                                             h
                                                                                        h
                                                           the solutions of
                 double helix  The structure of DNA in all
                 biological species is in the form of a double helix
                                                                   a(u, v) = f(v) ∀v ∈ V,
                 made of two chain molecules. Each chain is a
                 polymer made of four types of nucleotide: A
                 (adenine), G(guanine), T(thymine), andC(cyto-    a(u ,v ) = f(v ) ∀v ∈ V ,
                                                                       h
                                                                                        h
                                                                                    h
                                                                               h
                                                                    h
                 sine). The structure that was first proposed by
                 J.D. Watson and F.H.C. Crick immediately leads
                                                           where V is some closed subspace of V (a con-
                                                                 h
                 to a possible mechanism of biological heredity.
                                                           forming finite element space) and f ∈ V .For

                 This was later confirmed by experiments and
                                                           φ ∈ H denote by g(φ) ∈ V the solution of
                 hence provides a molecular basis for genetics.
                                                                 a(v, g(φ)) = (φ, v) H  ∀v ∈ V.
                 drawing    See rendering.
                                                           Then, choosing φ := u − u we obtain via
                                                                                   h
                 dual   See dual vector space, complement.  Galerkin orthogonality
                                                                      2
                 dual basis  Let E be a finite dimensional vec-   u − u   = a(u − u ,g(u − u )) ≤
                                                                                         h
                                                                    h H
                                                                                h
                 tor space with basis (e ,...,e ) . The dual basis
                                  1     n
                   1
                         n
                 (α ,...,α ) of the dual space E is defined by
                                           ∗
                         j
                                  j
                  j
                                                                      h V
                 α (e ) = δ , where δ = 1if j = i and 0 other-   a  u − u   inf v h ∈V h   g(u − u ) − v
                                                                                      h
                                                                                           h V
                                 i
                         i
                    i
                 wise.
                                                           If g(φ) possesses extra regularity beyond merely
                                                           belonging to V , the second term on the right-
                 dual vector space  Let V be a vector space
                                                           hand side will become very small compared to
                 (over the field K); the set V of all linear func-
                                       ∗
                                                            u − u   , if the resolution of the finite ele-
                                                                 h H
                 tionals α : V → K can be endowed with a struc-
                                                           ment space is increased. Thus, when considering
                 ture of vector space by defining (λα + µβ)(v) =
                                                           families of finite element spaces, the H-norm of
                 λ α(v) + µ β(v) (where λ, µ ∈ K; v ∈ V ). The
                                                           the discretization error may converge asymptot-
                 vector space V so obtained is called the (alge-
                             ∗
                                                           ically faster than the V -norm.
                 braic) dual vector space of V .
                   If V is a topological vector space the set of
                 all linear and continuous functionals α : V →
                 K with the standard linear structure introduced  Duffing equation  The Duffing equation
                 above is called the topological dual space of V ,
                                                                                3
                 and it is still denoted by V .                        ¨ x + x + "x = 0
                                      ∗
           © 2003 by CRC Press LLC
   42   43   44   45   46   47   48   49   50   51   52