5. Examples

5.1. Solving by the direct method MULT_FRONT

• Composition of the assembled matrix and of the second member:

We have previously calculated the elementary terms KEL, FEL.

K = ASSE_MATRICE (MATR_ELEM = KEL, NUME_DDL =NAKED,) F = ASSE_VECTEUR (MATR_ELEM = FEL, NUME_DDL =NAKED,)

• Factorization:

• Resolution:

• for the use of kinematic loads (with the elimination of imposed degrees of freedom), see the example given in command AFFE_CHAR_CINE [U4.44.03].

5.2. Solving by the MUMPS method

NAKED = NUME_DDL (MATR_RIGI = KEL)


K = ASSE_MATRICE (MATR_ELEM = KEL, NUME_DDL = NU)

F = ASSE_VECTEUR (VECT_ELEM = FEL, NUME_DDL = NU)

K = FACTORISER (reuse= K, MATR_ASSE = K, METHODE = 'MUMPS')

dep = RESOUDRE (CHAM_NO = F, MATR = K)

5.3. Resolution by the preconditioned conjugate gradient method

NAKED = NUME_DDL (MATR_RIGI = KEL)


K = ASSE_MATRICE (MATR_ELEM = KEL, NUME_DDL = NU)

F = ASSE_VECTEUR (VECT_ELEM = FEL, NUME_DDL = NU)

KPREC = FACTORISER (MATR_ASSE = K, METHODE = 'GCPC',

PRE_COND =” LDLT_INC “)

dep = RESOUDRE (CHAM_NO = F, MATR = K, MATR_PREC = KPREC,

NMAX_ITER = 1000, RESI_RELA = 1e-07 )

5.4. Solving by the PETSC method

NAKED = NUME_DDL (MATR_RIGI = KEL)


K = ASSE_MATRICE (MATR_ELEM = KEL, NUME_DDL = NU)

F = ASSE_VECTEUR (VECT_ELEM = FEL, NUME_DDL = NU)

K = FACTORISER (refuse=K, MATR_ASSE = K, METHODE = 'PETSC')

dep = RESOUDRE (CHAM_NO = F, MATR = K, MATR_PREC = K,

ALGORITHME =” GMRES “, NMAX_ITER = 1000, RESI_RELA = 1e-07)