history_before_2005

来自「FreeFem++可以生成高质量的有限元网格。可以用于流体力学」· 代码 · 共 639 行 · 第 1/2 页

beginning:  november 21,  2001:  version 1.08.
----------------------------------------------

2001/11/22:  correction  of operator == and != 
2001/11/23 , version: 1.09: correction  (with g++) 
template<class A>  struct SameType, type of OK must be int and not bool 

2001/11/24   add fonctionality in plot,  bb=[[x1,y1],[x2,y2]] 
 add  loop if enter  character +,-,=,c,C,r  in graphic window.
2001/11/28
 correction bug initialization of QuadTree if less than 4 points in the quadtree
 files  QuadTree.cpp and FQuadTree.hpp
 add exec("xxx..."); //to execute on system command. "xxx ... " 
 add  dumptable(cout); //to show all internal table
2001/11/29:  
  Version ans graphique + ajout d'option dans plot (cf. doc)
  correction ajoute renum() des maillage crees. 
2001/12/10
  Correct missing check in plot  
     ex:  plot(1);  trap before now, genere a compile error
  Correct in the interpolation is full not conforme FE, 
  do not prolonged by continuity. change 
  add after line 479 in file lgfem.cpp
  if (outside && !KK.tfe->NbDfOnVertex && !KK.tfe->NbDfOnEdge) 
    return   SetAny<R>(0.0); 
2001/12/12
  correction in gibbs (mesh renum) reconstruct the array of
  triangle for each vertex (PB. of interpolation in non 
  convexe domain) Big bug.

Make version 1.14
-----------------

2001/12/14
  correction in trunc mesh, bug if empty mesh is created,
  and move  a little the test point in a triangle 
  is not exactly the barycenter.  

2002/01/14
  correction in probem.cpp line 1309  bug when we write   qft= ...  , in int2d 
   if (nargs[0]) return *GetAny<const Fem2D::QuadratureFormular*>((*nargs[1])...
  becomes:
   if (nargs[0]) return *GetAny<const Fem2D::QuadratureFormular*>((*nargs[0])...
2002/01/15
  in lgfem.cpp remove line 522  reffecran();//bug if not graphique some time
  in file MeshGeom.cpp line 108 bug if name == 0 ;  add test before the cout like 
    if(name) cout << " ... " << name << .... 
  correct the name of the exec file (FreeFee++ -> FreeFem++) in Makefile
2002/02/03
  in Mesh2.cpp in preinit() add call to srand(),
    to get the same  mesh with the same data. 
  correct 
  mesh tth=th; 
  
Make version 1.15
-----------------

2002/02/20
   Add periodic boundary condition see the manual //\index{periodic}

Make version 1.16
-----------------

  Add Parallele Mpi 
  correct bug in SegmentationFault.edp missing placing of delete[] operator
  put declaration in  for
  wait =xx change the default value of wait
  ofstream f("foo.txt",append)  open a file in append mode
   add NoUseOfWait=true;    never wait for run test easly\index{NoUseOfWait}

Make version 1.17 2002/03/20
----------------------------

  Make default iso value changing with zoom option.
  Add GMRES solver   dimKrylov= ,  tgv= //\index{GMRES}\index{problem!dimKrylov=}\index{solve!tgv=}\index{problem!tgv=}
 
april 2002
  Make the current version without CheckPtr, and 
  correct some bug in string allocation (forgetting +1 in some length)
  Improving the speed of the software
  add CPUTime global bool variable to print the CPU time of each instruction.
  correct some printing without verbosity 
  adding some checking in array management 
  Add option in the Makefile (gnumake)

Make version 1.18 the 2002/04/08  
--------------------------------
  
  Big correction in automatic cast see  bugv1.18.edp
  correction in interpolation of  label see also bugv1.18.edp
  Big correction in construction of non-symetrix matrix (see BUG)



Make version 1.19 2002/04/18
----------------------------

   Correction in Quadtree integer overflow when 
   interpolate solution of from one big domain to a very small one.

   Correction in embedded function (return problem)
   Correction in return type of function (right value an note left value)
   Correction cast bool to int, so lots of bug in expression 
   like  (region==2)*5 given alway 0, invisibl before 1.18 (because bug in
   automatics cast)

Make version 1.20 2002/04/25
----------------------------

  Correction of bug in interpolation on non convexe domain.
  correction of bug in periodic boundary condition (pb of sens(in french))
  correction some compilation bug under g++ v3.0 in RNM file
  correction find common point buildmesh (change threshold value)

Make version 1.21 2002/04/29
----------------------------

  Add Non linear Conjugued Gradient (CGNL) routine.
  correct bug in linearGC with non zero right and size.

Make version 1.22 2002/05/02
----------------------------

 all optimisation tool from //\url{http://coool.mines.edu/}
 Add BFGS optimisation tools form  cool //\index{BFGS}
 correct bug in CGNL, correct tgv=, in linear form //\index{CGNL}
 make algo.edp example 

Make version 1.23 2002/05/13
----------------------------

  correct bug of the symetric matrix are independ of  x,y,..
   add line at  285, in file problem.cpp: 
          MeshPointStack(stack)->set(T(pi),pi,Ku);

  correct bug in sign of int1d in linearform, 
    so change in exemple fluidstructure, schwarz-no-overlap, LaplaceP1, 
    aalapacien, lapacienprecon
  correct some bug in -= operator with  result of *. */ 
  add some operator:
    -square(u) = u^2  //\index{square}
    -intalledges(Th)(...) to compute integrale on all edges of all triangle //\index{intalledges}  
    for error indicator add global :
    -lenEdge the len of the current edge //\index{lenEdge}
    -hTriangle the size of the current triangle //\index{hTriangle}
    -area the area of the current triangle //\index{area} 

Make version 1.24
-----------------

  add in BamgFreeFem.cpp after line 211 to set name 
    Tn->name= new char[strlen("msh2bamg")+1];
    strcpy(Tn->name,"msh2bamg");
   add -  of linearform and bilinearform see LaplaceP1bis.edp 

   correct: July 8, 2002
   Vh u1,u2,v1,v2; 
   bug in problem(u1,u2,v1,v2,.... ) = 
   in file: FESSpace.hpp line 193:
   { throwassert(dim_which_sub_fem[N-1]>=0 && dim_which_sub_fem[N-1]< nb_sub_fem);
     for(int i=0,n0=0,l=0,i0=0; i<k;  i++,n0+=t.N,i0+=t.NbDoF)
         for (int j=0;j<t.pij_alpha.N();j++,l++) {
           pij_alpha[l].i=t.pij_alpha[j].i+i0;
           pij_alpha[l].p=t.pij_alpha[j].p;
           pij_alpha[l].j=t.pij_alpha[j].j+n0;
         }                         
    } 

    add plot of list of border //\index{plot!border}

    add init array:

    real[int] a=[1,2,3,5.6,7,8,9]; //\index{array!initialisation}

Make version 1.25 
-----------------

 13 aout 2002 Major Bug in : FESpace.hpp ligne 199

  All the non scalar problem with same kind
     of finite element do not work: 

    FESpace.hpp ligne 199
      for(int i=0,n0=0,l=0,i0=0; i<k;  i++,n0+=t.N,i0+=t.NbDoF)
        for (int j=0;j<t.pij_alpha.N();j++,l++) {
          pij_alpha[l].i=t.pij_alpha[j].i+i0;
          pij_alpha[l].p=t.pij_alpha[j].p;
          pij_alpha[l].j=t.pij_alpha[j].j+n0;
          }                         
    become
     // Warning the componant is moving first 
        for (int j=0,l=0;j<t.pij_alpha.N();j++) // for all sub DF
          for(int i=0,i0=0; i<k; i++,l++) // for componate
          {
          pij_alpha[l].i=t.pij_alpha[j].i*k+i;   //  DoF number
          pij_alpha[l].p=t.pij_alpha[j].p;       //  point of interpolation
          pij_alpha[l].j=t.pij_alpha[j].j+i*t.N; //  componante of interpolation
          }                         
 
   two corrections in problem.cpp :
     2 missing  delete     (example laplacienprecon.edp)
     2 delete -> delete [] (example testFE.edp) 

 26 septembre 2002 
   add New Finite element RTortho (a conforme //\index{RTortho}
    FE in $H(curl)$ like RT  conforme FE in $H(div)$.
    see exemple aaRT.epd
   correct the computation of dx(v) and dy(u) in [u,v] RT finite element
    function. 
   compile on hpux 11 with gcc 3.2 see Makefile-hp9000s700
   correct argument passing in GMRES routine

Make version 1.26 le 26 septembre 2002
--------------------------------------

 05 nov. 2002 Correct very small mistake in bamglib  part.

 19 nov 2002:
  add:
   -tanh function //\index{tanh}
   -qf2pT4P1 a triangular QuadratureFormular  
	(4P1 the qf2pT QuadratureFormular) //\index{qf2pT4P1}
   small change une meshadapt to just compute the metrix see exemple
      convectapt.edp 
    make interpolation matrix between to FEspace (not wet finish)

 3/12/2002 
  make correction to get a not to bad linenumber in error.
  add add test is used of unset x,y to make a error.
  add checkmovemesh(Th,[x+u,y+v]) function to return the //\index{checkmovemesh}
  value of the area of the minimal triangle of the movemesh.
  see mesh.edp example.

 23/12/2002:
  add eigen value solver  //\index{eigenvalue}
  see eigen README_ARPACK to compile
  see  examples++-eigen to test.

Make  version 1.28 2/1/2003 
---------------------
09/1/2003: 
 add sub array option in array:  //\index{array!sub}
  real[int] tab(100);
  tab(:) = array from 0 to 99=tab.n-1
  tab(2:10)  sub array from 2 to 10
  tab(2:10:2)  sub array from 2 to 10 by step 2
  tab(2:10:2)[0] = 5 ; => modificaton of tab[2] = 5;
  tab(2:10:2)[1] = 6 ; => modificaton of tab[2+2] = 6;
  
  
16/01/2003
add macro generation like cpp preprosseur://\index{macro}
this is usefull to make automatic diff
exemple: 
real cc=2;
macro f(u) (cc*(exp(u)-1)) //
macro df(f) (cc*(exp(f)))  //
real u=1;
cout << (cc*(exp(u)-1)) << endl;
cout << f(u) << endl;
cout << df(f(u)) << endl;

see macro.edp  for more detail:
  
22/01/2003 
in file lex.cpp add .c_str() for compilation problem 
on g++ 2.95   
  
15/04/2003
add new finite element:
 P1b  P1 + Bubble  //\index{P1b}
 P1dc P1 discontinious //\index{P1dc}
 P2dc P2 discontinious //\index{P2dc}

correct bug periodic BC with vectorial FESpace, 
add the code.

Make version 1.31
-----------------  

23/04/2003
  small correction to compile with  g++ 2.95.2 in eigen value tools

26/04/2003
  add function triangulate(filename) to build the //\index{triangulate}
  Delaunay Triangulation of a set of points in R^2.
 to build a function form a set of : x y f(x,y)
 see exemple mesh.edp  


make version 1.32 (29/04/2003)
-----------------

 -add optimization in automatique interpolation 
  CPU of all tutoral exemples 172 s( mon my Mac) new version 1.33 (with graphics)
  CPU of all tutoral exemples 167 s( mon my Mac) new version 1.33 (without graphics)
  CPU of all tutoral exemples 169 s( mon my Mac) new version 1.32 (without graphics)
  bofbof.
  
 -correct bug in macro generation, a macro existe just in a block {..}.

 //  lex.cpp  
    if ( ret =='{') { /*cout << " listMacroDef->push_back"<< endl; */
      listMacroDef->push_back( MapMacroDef() );}
  else if (ret == '}') {/*cout << " listMacroDef->pop_back"<< endl;  */
       listMacroDef->pop_back( );}

 -small correction to be compatible with g++ version 3.3

Make version 1.33 01/07/2003
-----------------------------

 - small correction in parallelempi.cpp add a ; line 159

 - comment line 519 in file AFunction.hpp (pb with g++ 3.3.1)
	 // operator Expression() const {return f;}
  (double definition)see line 527
	  operator  const  E_F0 *  () const {return f;}

 - correct Makefile for mpi version