File r34/lib/desir.doc artifact e0e3f6c202 part of check-in d58ccc1261



         ***************************************************************
         *                                                             *
         *                           DESIR                             *
         *                           =====                             *
         *                                                             *
         *      SOLUTIONS FORMELLES  D'EQUATIONS DIFFERENTIELLES       *
         *                                                             *
         *                   LINEAIRES ET HOMOGENES                    *
         *                                                             *
         * AU VOISINAGE DE POINTS SINGULIERS REGULIERS ET IRREGULIERS  *
         *                                                             *
         ***************************************************************

             Differential linear homogenous Equation Solutions in the
              neighbourhood of Irregular and Regular singular points

                            Version 3.1  -  Septembre 89


                       Groupe de Calcul Formel de Grenoble
                                laboratoire TIM3

                  (C. Dicrescenzo, F. Richard-Jung, E. Tournier)

                           E-mail: dicresc@afp.imag.fr


1) Introduction
2) Form of solutions
3) Interactive use
4) Direct use
5) Useful functions
6) Limitations
7) Implementation



1) INTRODUCTION
   ************

   This software enables the basis of formal solutions to be computed for an
   ordinary homogeneous differential equation with polynomial coefficients
   over Q of any order, in the neighbourhood of zero ( regular or irregular
   singular point, or ordinary point ).
   Tools have been added to deal with equations with a polynomial right-hand
   side, parameters and a singular point not to be found at zero.

   This software can be used in two ways : * direct ( DELIRE procedure )
                                           * interactive ( DESIR procedure)

   The basic procedure is the DELIRE procedure which enables the solutions of
   a linear homogeneous differential equation to be computed in the neigh-
   bourhood of zero.

   The DESIR procedure is a procedure without argument whereby DELIRE can be
   called without preliminary treatment to the data, that is to say, in an
   interactive autonomous way. This procedure also proposes some transfor-
   mations on the initial equation. This allows one to start confortably
   with an equation which has a non zero singular point, a polynomial
   right-hand side and parameters.

   This document is a succint user manual. For more details on the underlying
   mathematics and the algorithms used, the reader can refer to :

      E. Tournier : Solutions formelles d'equations differentielles - Le
        logiciel de calcul formel DESIR.
        These d'Etat de l'Universite Joseph Fourier (Grenoble - avril 87).

   He will find more precision on use of parameters in :

      F. Richard-Jung : Representation graphique de solutions d'equations
        differentielles dans le champ complexe.
        These de l'Universite Louis Pasteur (Strasbourg - septembre 88).



2) FORMS OF SOLUTIONS
   ******************

   We have tried to represent solutions in the simplest form possible. For
   that, we have had to choose different forms according to the complexity
   of the equation (parameters) and the later use we shall have of these
   solutions.

 "general solution"  =  {......, { split_sol , cond },....}
 ------------------

     cond = list of conditions or empty list (if there is no condition)
            that parameters have to verify such that split_sol is in the
            basis of solutions. In fact, if there are parameters, basis of
            solutions can have different expressions according to the values
            of parameters. ( Note : if cond={}, the list "general solution"
            has one element only.

     split_sol = { q , ram , polysol , r }
            ( " split solution " enables precise information on the solution
            to be obtained immediately )

 The variable in the differential operator being x, solutions are expressed in
 respect to a new variable xt, which is a fractional power of x, in the
 following way :

         q       : polynomial in 1/xt with complex coefficients
         ram     : xt = x**ram (1/ram is an integer)
         polysol : polynomial in log(xt) with formal series in xt coefficients
         r       : root of a complex coefficient polynomial ("indicial
                   equation").


                         qx  r*ram
 "standard solution"  = e   x      polysolx
  -----------------
         qx and polysolx are q and polysol expressions in which xt has been
         replaced by x**ram

 N.B. : the form of these solutions is simplified according to the nature of
        the point zero.
   - si 0 is a regular singular point : the series appearing in polysol are
     convergent, ram = 1 and q = 0.
   - if 0 is a regular point, we also have : polysol is constant in log(xt)
     (no logarithmic terms).




3) INTERACTIVE USE
   ***************

   To call the procedure : desir();
                           solution:=desir();

   The DESIR procedure computes formal solutions of a linear homogeneous
   differential equation in an interactive way.
   In this equation the variable must be x.
                                 ---------
   The procedure requires the order and the coefficients of the equation, the
   names of parameters if there are any, then if the user wants to transform
   this equation and how ( for example to bring back a singular point to zero
    - see procedures changehom, changevar, changefonc - ).

   This procedure DISPLAYS the solutions and RETURNS a list of general term
   { lcoeff, {....,{ general_solution },....}}. The number of elements in
   this list is linked to the number of transformations requested :
      * lcoeff : list of coefficients of the differential equation
      * general_solution : solution written in the general form




4) DIRECT USE
   **********

   procedure delire(x,k,grille,lcoeff,param);
   ==========================================

   This procedure computes formal solutions of a linear homogeneous differen-
   tial equation with polynomial coefficients over Q and of any order, in the
   neighborhood of zero, regular or irregular singular point. In fact it
   initializes the call of the NEWTON procedure that is a recursive procedure
   (algorithm of NEWTON-RAMIS-MALGRANGE)

   x      : variable
   k      : "number of desired terms".
            For each formal series in xt appearing in polysol,
            a_0+a_1 xt+a_2 xt**2+...+a_n xt**n+..., we compute the k+1 first
            coefficients a_0, a_1,...a_k.
   grille : the coefficients of the differential operator are polynomial in
            x**grille (in general grille=1)
   lcoeff : list of coefficients of the differential operator (in increasing
            order of differentiation)
   param  : list of parameters

   This procedure RETURNS the list of general solutions.
                              ----



5) USEFUL FUNCTIONS
  *****************

 -1) Reading of equation coefficients
     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

   procedure lectabcoef( ) ;
   ========================;

   This procedure is called by DESIR to read the coefficients of an equation,
   in  increasing order of differentiation, but can be used independently.
       -----------------------------------
   reading of n    : order of the equation.
   reading of parameters (only if a variable other than x appears in the
   coefficients)
   this procedure returns the list { lcoeff , param } made up of the list of
   coefficients and the list of parameters (which can be empty).

 -2) verification of results
     ~~~~~~~~~~~~~~~~~~~~~~~

   procedure solvalide(solutions,solk,k);
   =====================================;

   This procedure enables the validity of the solution number solk in the list
   "solutions" to be verify.
   solutions = {lcoeff,{....,{general_solution},....}} is any element of the
   list returned by DESIR or is {lcoeff,sol} where sol is the list returned by
   DELIRE.
                                   qx  r*ram
   If we carry over the solution e   x      polysolx in the equation, the
                         qx  r*ram
   result has the form e   x      reste, where reste is a polynomial in
   log(xt), with polynomial coefficients in xt. This procedure computes the
   minimal valuation V of reste as polynomial in xt, using k "number of
   desired terms" asked for at the call of DESIR or DELIRE, and DISPLAYS the
                                                               ram*(r+v)
   "theoretical" size order of the regular part of the result : x         .
   On the other hand, this procedure carries over the solution in the equation
   and DISPLAYS the significative term of the result. This is of the form :
               qx  a
              e   x  polynomial(log(xt)), with a>=ram*(r+v).

   Finally this procedure RETURNS the complete result of the carry over of the
   solution in the equation.

   This procedure cannot be used if the solution number solk is linked to a
   condition.


 -3) writing of different forms of results
     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

   procedure standsol(solutions);
   ==============================

   This procedure enables the simplified form of each solution to be obtained
   from the list "solutions", {lcoeff,{...,{general_solution},....}} which is
   one of the elements of the list returned by DESIR, or {lcoeff,sol} where
   sol is the list returned by DELIRE.

   This procedure RETURNS a list of 3 elements : { lcoeff, solstand, solcond }
     * lcoef = list of differential equation coefficients
     * solstand = list of solutions written in standard form
     * solcond  = list of conditional solutions that have not been written in
                  standard form. This solutions remain in general form.

   This procedure has no meaning for "conditional" solutions. In case, a value
   has to be given to  the parameters, that can be done either by calling the
   procedure SORPARAM that displays and returns these solutions in the
   standard form, either by calling the procedure SOLPARAM which returns
   these solutions in general form.

  procedure sorsol(sol);
  ======================

   This procedure is called by DESIR to write the solution sol, given in
   general form, in standard form with enumeration of different conditions (if
   there are any).
   It can be used independently.


 -4) Writing of solutions after the choice of parameters
     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

  procedure sorparam(solutions,param);
  ====================================

   This is an interactive procedure which displays the solutions evaluated :
   the value of parameters is requested.
      * solutions :  {lcoeff,{....,{general_solution},....}}
      * param : list of parameters.

   It returns the list formed of 2 elements :
      * list of evaluated coefficients of the equation
      * list of standard solutions evaluated for the value of parameters.

  procedure solparam(solutions,param,valparam);
  =============================================

   This procedure evaluates the general solutions for the value of parameters
   given by valparam and returns these solutions in general form.
       * solutions :  {lcoeff,{....,{general_solution},....}}
       * param : list of parameters
       * valparam : list of parameters values

   It returns the list formed of 2 elements :
       * list of evaluated coefficients of the equation
       * list of solutions in general form, evaluated for the value of
         parameters.


 -5) Transformations
     ~~~~~~~~~~~~~~~

  procedure changehom(lcoeff,x,secmember,id);
  ===========================================

   Differentiation of an equation with right-hand side.
       * lcoeff : list of coefficients of the equation
       * x : variable
       * secmember : right-hand side
       * id : order of the differentiation.


   It returns the list of coefficients of the differentiated equation.
   It enables an equation with polynomial right-hand side to be transformed
   into a homogeneous equation by differentiating id times,
   id = degre(secmember) + 1.

  procedure changevar(lcoeff,x,v,fct);
  ====================================

   Changing of variable in the homogeneous equation defined by the list,lcoeff
   of its coefficients : the old variable x and the new variable v are linked
   by the relation x = fct(v).

   It returns the list of coefficients in respect to the variable v of the new
   equation.

   examples of use :
   ---------------
   - translation enabling a rational singularity to be brought back to zero.
   - x = 1/v brings the infinity to 0.

  procedure changefonc(lcoeff,x,q,fct);
  =====================================

   Changing of unknown function in the homogeneous equation defined by the
   list lcoeff of its coefficients :
       * lcoeff : list of coefficients of the initial equation
       * x : variable
       * q : new unknown function
       * fct : y being the unknown function  y = fct(q)

   It returns the list of coefficients of the new equation.

   example of use :
   --------------
   this procedure enables the computation,in the neighbourhood of an irregular
   singularity, of the "reduced" equation associated to one of the slopes (the
   Newton polygon having a null slope of no null length). This equation gives
   much informations on the associated divergent series.


 -6) Optional writing of intermediary results
     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

   switch trdesir : when it is ON, at each step of the Newton algorithm, a
          =======
   description of the Newton polygon is displayed (it is possible to follow
   the break of slopes), and at each call of the FROBENIUS procedure ( case of
   a null slope ) the corresponding indicial equation is displayed.

   By default, this switch is OFF.


6)   LIMITATIONS
     ***********

 -1) This DESIR version is limited to differential equations leading
   to indicial equations of degree <= 3. To pass beyond this limit, a further
   version written in the D5 environment of the computation with algebraic
   numbers has to be used.

 -2) The computation of a basis of solutions for an equation depending on
   parameters is assured only when the indicial equations are of degree <= 2.


7) IMPLEMENTATION
   **************

   This software uses the 3.3 version of REDUCE.



REDUCE Historical
REDUCE Sourceforge Project | Historical SVN Repository | GitHub Mirror | SourceHut Mirror | NotABug Mirror | Chisel Mirror | Chisel RSS ]