Esempio n. 1
0
void mexFunction
(
    int nargout,
    mxArray *pargout [ ],
    int nargin,
    const mxArray *pargin [ ]
)
{
    CS_INT n, nel, s ;
    cs *A, *AT ;
    if (nargout > 1 || nargin != 3)
    {
        mexErrMsgTxt ("Usage: C = cs_frand(n,nel,s)") ;
    }
    n = mxGetScalar (pargin [0]) ;
    nel = mxGetScalar (pargin [1]) ;
    s = mxGetScalar (pargin [2]) ;

    n = CS_MAX (1,n) ;
    nel = CS_MAX (1,nel) ;
    s = CS_MAX (1,s) ;

    AT = cs_dl_frand (n, nel, s) ;
    A = cs_dl_transpose (AT, 1) ;
    cs_dl_spfree (AT) ;
    cs_dl_dropzeros (A) ;

    pargout [0] = cs_dl_mex_put_sparse (&A) ;
}
Esempio n. 2
0
/* add an entry to a triplet matrix; return 1 if ok, 0 otherwise */
int cs_entry (cs *T, int i, int j, double x)
{
    if (!CS_TRIPLET (T) || i < 0 || j < 0) return (0) ;     /* check inputs */
    if (T->nz >= T->nzmax && !cs_sprealloc (T,2*(T->nzmax))) return (0) ;
    if (T->x) T->x [T->nz] = x ;
    T->i [T->nz] = i ;
    T->p [T->nz++] = j ;
    T->m = CS_MAX (T->m, i+1) ;
    T->n = CS_MAX (T->n, j+1) ;
    return (1) ;
}
Esempio n. 3
0
/* cs_qrsol: solve least squares or underdetermined problem */
void mexFunction
(
    int nargout,
    mxArray *pargout [ ],
    int nargin,
    const mxArray *pargin [ ]
)
{
    CS_INT k, order ;
    if (nargout > 1 || nargin < 2 || nargin > 3)
    {
        mexErrMsgTxt ("Usage: x = cs_qrsol(A,b,order)") ;
    }
    order = (nargin < 3) ? 3 : mxGetScalar (pargin [2]) ;
    order = CS_MAX (order, 0) ;
    order = CS_MIN (order, 3) ;

    if (mxIsComplex (pargin [0]) || mxIsComplex (pargin [1]))
    {
#ifndef NCOMPLEX
        cs_cl *A, Amatrix ;
        cs_complex_t *x, *b ;
        A = cs_cl_mex_get_sparse (&Amatrix, 0, pargin [0]) ;    /* get A */
        b = cs_cl_mex_get_double (A->m, pargin [1]) ;           /* get b */
        x = cs_dl_calloc (CS_MAX (A->m, A->n), sizeof (cs_complex_t)) ;
        for (k = 0 ; k < A->m ; k++) x [k] = b [k] ;            /* x = b */
        cs_free (b) ;
        if (!cs_cl_qrsol (order, A, x))                         /* x = A\x */
        {
            mexErrMsgTxt ("QR solve failed") ;
        }
        pargout [0] = cs_cl_mex_put_double (A->n, x) ;          /* return x */
#else
        mexErrMsgTxt ("complex matrices not supported") ;
#endif
    }
    else
    {
        cs_dl *A, Amatrix ;
        double *x, *b ;
        A = cs_dl_mex_get_sparse (&Amatrix, 0, 1, pargin [0]) ;     /* get A */
        b = cs_dl_mex_get_double (A->m, pargin [1]) ;               /* get b */
        x = cs_dl_calloc (CS_MAX (A->m, A->n), sizeof (double)) ;   /* x = b */
        for (k = 0 ; k < A->m ; k++) x [k] = b [k] ;
        if (!cs_dl_qrsol (order, A, x))                         /* x = A\x */
        {
            mexErrMsgTxt ("QR solve failed") ;
        }
        cs_dl_mex_put_double (A->n, x, &(pargout [0])) ;        /* return x */
        cs_free (x) ;
    }
}
/* allocate a sparse matrix (triplet form or compressed-column form) */
csr *csr_spalloc (int m, int n, int nzmax, int values, int triplet)
{
  csr *A = (csr*)calloc (1, sizeof (csr)) ;    /* allocate the csr struct */
  if (!A) return (NULL) ;                 /* out of memory */
  A->m = m ;                              /* define dimensions and nzmax */
  A->n = n ;
  A->nzmax = nzmax = CS_MAX (nzmax, 1) ;
  A->nz = triplet ? 0 : -1 ;              /* allocate triplet or comp.row */
  A->p = (int*)malloc (triplet ? CS_MAX(nzmax,1) * sizeof (int) : CS_MAX(m+1,1) * sizeof (int)) ;
  A->j = (int*)malloc (CS_MAX(nzmax,1) * sizeof (int)) ;
  A->x = values ? (double*)malloc (CS_MAX(nzmax,1) * sizeof (double)) : NULL ;
  return ((!A->p || !A->j || (values && !A->x)) ? csr_spfree (A) : A) ;
}
/* C = alpha*A + beta*B */
csr *csr_add (const csr *A, const csr *B, double alpha, double beta)
{
  int p, j, nz = 0, anz, *Cp, *Cj, *Bp, m, n, bnz, *w, values ;
  double *x, *Bx, *Cx ;
  csr *C ;
  if (!CS_CSC (A) || !CS_CSC (B)) return (NULL) ;         /* check inputs */
  if ( (A->m != B->m) || (A->n != B->n) ) return (NULL);
  m = A->m ; anz = A->p [m] ;
  n = B->n ; Bp = B->p ; Bx = B->x ; bnz = Bp [m] ;
  w = (int*)calloc (CS_MAX(n,1), sizeof (int)) ;                       /* get workspace */
  values = (A->x != NULL) && (Bx != NULL) ;
  x = values ? (double*)malloc (n * sizeof (double)) : NULL ;    /* get workspace */
  C = csr_spalloc (m, n, anz + bnz, values, 0) ;          /* allocate result*/
  if (!C || !w || (values && !x)) return (csr_done (C, w, x, 0)) ;
  Cp = C->p ; Cj = C->j ; Cx = C->x ;
  for (j = 0 ; j < n ; j++)
    {
      Cp [j] = nz ;                   /* row j of C starts here */
      nz = csr_scatter (A, j, alpha, w, x, j+1, C, nz) ;   /* alpha*A(j,:)*/
      nz = csr_scatter (B, j, beta, w, x, j+1, C, nz) ;    /* beta*B(j,:) */
      if (values) for (p = Cp [j] ; p < nz ; p++) Cx [p] = x [Cj [p]] ;
    }
  Cp [m] = nz ;                       /* finalize the last row of C */
  csr_sprealloc (C, 0) ;              /* remove extra space from C */
  return (csr_done (C, w, x, 1)) ;    /* success; free workspace, return C */
}
/* infinity-norm of x */
static double norm (double *x, int n)
{
    int i ;
    double normx = 0 ;
    for (i = 0 ; i < n ; i++) normx = CS_MAX (normx, fabs (x [i])) ;
    return (normx) ;
}
/* read a problem from a file; use %g for integers to avoid int conflicts */
problem *get_problem (FILE *f, double tol)
{
    cs *T, *A, *C ;
    int sym, m, n, mn, nz1, nz2 ;
    problem *Prob ;
    Prob = cs_calloc (1, sizeof (problem)) ;
    if (!Prob) return (NULL) ;
    T = cs_load (f) ;                   /* load triplet matrix T from a file */
    Prob->A = A = cs_compress (T) ;     /* A = compressed-column form of T */
    cs_spfree (T) ;                     /* clear T */
    if (!cs_dupl (A)) return (free_problem (Prob)) ; /* sum up duplicates */
    Prob->sym = sym = is_sym (A) ;      /* determine if A is symmetric */
    m = A->m ; n = A->n ;
    mn = CS_MAX (m,n) ;
    nz1 = A->p [n] ;
    cs_dropzeros (A) ;                  /* drop zero entries */
    nz2 = A->p [n] ;
    if (tol > 0) cs_droptol (A, tol) ;  /* drop tiny entries (just to test) */
    Prob->C = C = sym ? make_sym (A) : A ;  /* C = A + triu(A,1)', or C=A */
    if (!C) return (free_problem (Prob)) ;
    printf ("\n--- Matrix: %g-by-%g, nnz: %g (sym: %g: nnz %g), norm: %8.2e\n",
            (double) m, (double) n, (double) (A->p [n]), (double) sym,
            (double) (sym ? C->p [n] : 0), cs_norm (C)) ;
    if (nz1 != nz2) printf ("zero entries dropped: %g\n", (double) (nz1 - nz2));
    if (nz2 != A->p [n]) printf ("tiny entries dropped: %g\n",
            (double) (nz2 - A->p [n])) ;
    Prob->b = cs_malloc (mn, sizeof (double)) ;
    Prob->x = cs_malloc (mn, sizeof (double)) ;
    Prob->resid = cs_malloc (mn, sizeof (double)) ;
    return ((!Prob->b || !Prob->x || !Prob->resid) ? free_problem (Prob) : Prob) ;
}
Esempio n. 8
0
static void
_compute_minmax(cs_int_t            n_vals,
                const cs_real_t     var[],
                cs_real_t          *min,
                cs_real_t          *max)
{
  cs_int_t  i;
  cs_real_t  _min = DBL_MAX, _max = -DBL_MAX;

  for (i = 0; i < n_vals; i++) {
    _min = CS_MIN(_min, var[i]);
    _max = CS_MAX(_max, var[i]);
  }

#if defined(HAVE_MPI)

  if (cs_glob_n_ranks > 1) {
    MPI_Allreduce(&_min, min, 1, CS_MPI_REAL, MPI_MIN,
                  cs_glob_mpi_comm);

    MPI_Allreduce(&_max, max, 1, CS_MPI_REAL, MPI_MAX,
                  cs_glob_mpi_comm);
  }

#endif

  if (cs_glob_n_ranks == 1) {
    *min = _min;
    *max = _max;
  }
}
Esempio n. 9
0
/* infinity-norm of x */
static double norm (cs_complex_t *x, cs_long_t n)
{
    cs_long_t i ;
    double normx = 0 ;
    for (i = 0 ; i < n ; i++) normx = CS_MAX (normx, cabs (x [i])) ;
    return (normx) ;
}
Esempio n. 10
0
void *cs_realloc(void *p, int n, size_t size, int *ok) {

	void *pnew;
	pnew = realloc(p, CS_MAX (n,1) * size); /* realloc the block */
	*ok = (pnew != NULL); /* realloc fails if pnew is NULL */
	return ((*ok) ? pnew : p); /* return original p if failure */
}
/* C = A*B */
csr *csr_multiply (const csr *A, const csr *B)
{
  int p, j, nz = 0, anz, *Cp, *Cj, *Ap, m, k, n, bnz, *w, values, *Aj;
  double *x, *Ax, *Cx ;
  csr *C ;
  if (!CS_CSC (A) || !CS_CSC (B)) return (NULL) ;      /* check inputs */
  k = A->n;
  if (k != B->m ) return(NULL);
  m = A->m ; anz = A->p [A->m] ;
  n = B->n ; Ap = A->p ; Aj = A->j ; Ax = A->x ; bnz = B->p [k] ;
  w = (int*)calloc (CS_MAX(n,1), sizeof (int)) ;            /* get workspace */
  values = (Ax != NULL) && (B->x != NULL) ;
  x = values ? (double*)malloc (n * sizeof (double)) : NULL ; /* get workspace */
  C = csr_spalloc (m, n, anz + bnz, values, 0) ;       /* allocate result */
  if (!C || !w || (values && !x)) return (csr_done (C, w, x, 0)) ;
  Cp = C->p ;
  for (j = 0 ; j < m ; j++)
    {
      if (nz + n > C->nzmax && !csr_sprealloc (C, 2*(C->nzmax)+m))
        {
	  return (csr_done (C, w, x, 0)) ;            /* out of memory */
        }
      Cj = C->j ; Cx = C->x ;         /* C->j and C->x may be reallocated */
      Cp [j] = nz ;                   /* row j of C starts here */
      for (p = Ap [j] ; p < Ap [j+1] ; p++)
        {
	  nz = csr_scatter (B, Aj [p], Ax ? Ax [p] : 1, w, x, j+1, C, nz) ;
        }
      if (values) for (p = Cp [j] ; p < nz ; p++) Cx [p] = x [Cj [p]] ;
    }
  Cp [m] = nz ;                       /* finalize the last column of C */
  csr_sprealloc (C, 0) ;              /* remove extra space from C */
  return (csr_done (C, w, x, 1)) ;    /* success; free workspace, return C */
}
Esempio n. 12
0
/* cs_lusol: solve A*x=b using a sparse LU factorization */
void mexFunction
(
    int nargout,
    mxArray *pargout [ ],
    int nargin,
    const mxArray *pargin [ ]
)
{
    double tol ;
    CS_INT order ;
    if (nargout > 1 || nargin < 2 || nargin > 4)
    {
        mexErrMsgTxt ("Usage: x = cs_lusol(A,b,order,tol)") ;
    }
    order = (nargin < 3) ? 2 : mxGetScalar (pargin [2]) ;
    order = CS_MAX (order, 0) ;
    order = CS_MIN (order, 3) ;
    if (nargin == 2)
    {
        tol = 1 ;                           /* normal partial pivoting */
    }
    else if (nargin == 3)
    {
        tol = (order == 1) ? 0.001 : 1 ;    /* tol = 0.001 for amd(A+A') */
    }
    else
    {
        tol = mxGetScalar (pargin [3]) ;
    }
    if (mxIsComplex (pargin [0]) || mxIsComplex (pargin [1]))
    {
#ifndef NCOMPLEX
        cs_cl *A, Amatrix ;
        cs_complex_t *x ;
        A = cs_cl_mex_get_sparse (&Amatrix, 1, pargin [0]) ;    /* get A */
        x = cs_cl_mex_get_double (A->n, pargin [1]) ;           /* x = b */
        if (!cs_cl_lusol (order, A, x, tol))                    /* x = A\x */
        {
            mexErrMsgTxt ("failed (singular or out of memory)") ;
        }
        cs_cl_free (A->x) ;     /* complex copy no longer needed */
        pargout [0] = cs_cl_mex_put_double (A->n, x) ;          /* return x */
#else
        mexErrMsgTxt ("complex matrices not supported") ;
#endif
    }
    else
    {
        cs_dl *A, Amatrix ;
        double *x, *b ;
        A = cs_dl_mex_get_sparse (&Amatrix, 1, 1, pargin [0]) ;    /* get A */
        b = cs_dl_mex_get_double (A->n, pargin [1]) ;           /* get b */
        x = cs_dl_mex_put_double (A->n, b, &(pargout [0])) ;    /* x = b */
        if (!cs_dl_lusol (order, A, x, tol))                    /* x = A\x */
        {
            mexErrMsgTxt ("failed (singular or out of memory)") ;
        }
    }
}
Esempio n. 13
0
cs *cs_symperm(const cs *A, const int *pinv, int values) {

	int i, j, p, q, i2, j2, n, *Ap, *Ai, *Cp, *Ci, *w;
	double *Cx, *Ax;
	cs *C;
	if (!CS_CSC (A))
		return (NULL); /* check inputs */
	n = A->n;
	Ap = A->p;
	Ai = A->i;
	Ax = A->x;
	C = cs_spalloc(n, n, Ap[n], values && (Ax != NULL), 0); /* alloc result*/
	w = (int *) cs_calloc(n, sizeof(int)); /* get workspace */
	if (!C || !w)
		return (cs_done(C, w, NULL, 0)); /* out of memory */
	Cp = C->p;
	Ci = C->i;
	Cx = C->x;
	for (j = 0; j < n; j++) /* count entries in each column of C */
	{
		j2 = pinv ? pinv[j] : j; /* column j of A is column j2 of C */
		for (p = Ap[j]; p < Ap[j + 1]; p++) {
			i = Ai[p];
			if (i > j)
				continue; /* skip lower triangular part of A */
			i2 = pinv ? pinv[i] : i; /* row i of A is row i2 of C */
			w[CS_MAX (i2, j2)]++; /* column count of C */
		}
	}
	cs_cumsum(Cp, w, n); /* compute column pointers of C */
	for (j = 0; j < n; j++) {
		j2 = pinv ? pinv[j] : j; /* column j of A is column j2 of C */
		for (p = Ap[j]; p < Ap[j + 1]; p++) {
			i = Ai[p];
			if (i > j)
				continue; /* skip lower triangular part of A*/
			i2 = pinv ? pinv[i] : i; /* row i of A is row i2 of C */
			Ci[q = w[CS_MAX (i2, j2)]++] = CS_MIN (i2, j2);
			if (Cx)
				Cx[q] = Ax[p];
		}
	}
	return (cs_done(C, w, NULL, 1)); /* success; free workspace, return C */
}
Esempio n. 14
0
void mexFunction
(
    int nargout,
    mxArray *pargout [ ],
    int nargin,
    const mxArray *pargin [ ]
)
{
    cs Amatrix, *A ;
    int m, n, mn, m2, n2, k, s, j, ij, sj, si, p, *Ap, *Ai ;
    double aij, *S, *Ax ;
    if (nargout > 1 || nargin < 1 || nargin > 2)
    {
        mexErrMsgTxt ("Usage: S = cs_thumb(A,k)") ;
    }
    A = cs_mex_get_sparse (&Amatrix, 0, 1, pargin [0]) ;    /* get A */
    m = A->m ;
    n = A->n ;
    mn = CS_MAX (m,n) ;
    k = (nargin == 1) ? 256 : mxGetScalar (pargin [1]) ;    /* get k */
    /* s = size of each submatrix; A(1:s,1:s) maps to S(1,1) */
    s = (mn < k) ? 1 : (int) ceil ((double) mn / (double) k) ;
    m2 = (int) ceil ((double) m / (double) s) ;
    n2 = (int) ceil ((double) n / (double) s) ;
    /* create S */
    pargout [0] = mxCreateDoubleMatrix (m2, n2, mxREAL) ;
    S = mxGetPr (pargout [0]) ;
    Ap = A->p ;
    Ai = A->i ;
    Ax = A->x ;
    for (j = 0 ; j < n ; j++)
    {
        sj = j/s ;
        for (p = Ap [j] ; p < Ap [j+1] ; p++)
        {
            si = Ai [p] / s ;
            ij = INDEX (si,sj,m2) ;
            aij = fabs (Ax [p]) ;
            if (ISNAN (aij)) aij = BIG_VALUE ;
            aij = CS_MIN (BIG_VALUE, aij) ;
            S [ij] = CS_MAX (S [ij], aij) ;
        }
    }
}
Esempio n. 15
0
/* 1-norm of a sparse matrix = max (sum (abs (A))), largest column sum */
double cs_norm (const cs *A)
{
    int p, j, n, *Ap ;
    double *Ax,  norm = 0, s ;
    if (!CS_CSC (A) || !A->x) return (-1) ;             /* check inputs */
    n = A->n ; Ap = A->p ; Ax = A->x ;
    for (j = 0 ; j < n ; j++)
    {
        for (s = 0, p = Ap [j] ; p < Ap [j+1] ; p++) s += fabs (Ax [p]) ;
        norm = CS_MAX (norm, s) ;
    }
    return (norm) ;
}
Esempio n. 16
0
/* allocate a sparse matrix (triplet form or compressed-column form) */
cs *cs_spalloc (int m, int n, int nzmax, int values, int triplet)
{
    cs *A = cs_calloc (1, sizeof (cs)) ;    /* allocate the cs struct */
    if (!A) return (NULL) ;		    /* out of memory */
    A->m = m ;				    /* define dimensions and nzmax */
    A->n = n ;
    A->nzmax = nzmax = CS_MAX (nzmax, 1) ;
    A->nz = triplet ? 0 : -1 ;		    /* allocate triplet or comp.col */
    A->p = cs_malloc (triplet ? nzmax : n+1, sizeof (int)) ;
    A->i = cs_malloc (nzmax, sizeof (int)) ;
    A->x = values ? cs_malloc (nzmax, sizeof (double)) : NULL ;
    return ((!A->p || !A->i || (values && !A->x)) ? cs_spfree (A) : A) ;
}
Esempio n. 17
0
/* get a MATLAB flint array and convert to int */
int *cs_mex_get_int (int n, const mxArray *Imatlab, int *imax, int lo)
{
    double *p ;
    int i, k, *C = cs_malloc (n, sizeof (int)) ;
    cs_mex_check (1, n, 1, 0, 0, 1, Imatlab) ;
    p = mxGetPr (Imatlab) ;
    *imax = 0 ;
    for (k = 0 ; k < n ; k++)
    {
        i = p [k] ;
        C [k] = i - 1 ;
        if (i < lo) mexErrMsgTxt ("index out of bounds") ;
        *imax = CS_MAX (*imax, i) ;
    }
    return (C) ;
}
Esempio n. 18
0
/* cs_amd: approximate minimum degree ordering */
void mexFunction
(
    int nargout,
    mxArray *pargout [ ],
    int nargin,
    const mxArray *pargin [ ]
)
{
    cs Amatrix, *A ;
    int *P, order ;
    if (nargout > 1 || nargin < 1 || nargin > 2)
    {
        mexErrMsgTxt ("Usage: p = cs_amd(A,order)") ;
    }
    A = cs_mex_get_sparse (&Amatrix, 0, 0, pargin [0]) ;    /* get A */
    order = (nargin > 1) ? mxGetScalar (pargin [1]) : 1 ;   /* get ordering */
    order = CS_MAX (order, 1) ;
    order = CS_MIN (order, 3) ;
    P = cs_amd (order, A) ;                         /* min. degree ordering */
    pargout [0] = cs_mex_put_int (P, A->n, 1, 1) ;  /* return P */
}
Esempio n. 19
0
/* get a MATLAB flint array and convert to CS_INT */
CS_INT *cs_dl_mex_get_int (CS_INT n, const mxArray *Imatlab, CS_INT *imax,
    int lo)
{
    double *p ;
    CS_INT i, k, *C = cs_dl_malloc (n, sizeof (CS_INT)) ;
    cs_mex_check (1, n, 1, 0, 0, 1, Imatlab) ;
    if (mxIsComplex (Imatlab))
    {
        mexErrMsgTxt ("integer input cannot be complex") ;
    }
    p = mxGetPr (Imatlab) ;
    *imax = 0 ;
    for (k = 0 ; k < n ; k++)
    {
        i = p [k] ;
        C [k] = i - 1 ;
        if (i < lo) mexErrMsgTxt ("index out of bounds") ;
        *imax = CS_MAX (*imax, i) ;
    }
    return (C) ;
}
Esempio n. 20
0
/* cs_lusol: solve A*x=b using a sparse LU factorization */
void mexFunction
(
    int nargout,
    mxArray *pargout [ ],
    int nargin,
    const mxArray *pargin [ ]
)
{
    cs *A, Amatrix ;
    csi order ;
    double *x, *b, tol ;
    if (nargout > 1 || nargin < 2 || nargin > 4)
    {
        mexErrMsgTxt ("Usage: x = cs_lusol(A,b,order,tol)") ;
    }
    A = cs_mex_get_sparse (&Amatrix, 1, 1, pargin [0]) ;    /* get A */
    b = cs_mex_get_double (A->n, pargin [1]) ;              /* get b */
    x = cs_mex_put_double (A->n, b, &(pargout [0])) ;       /* x = b */
    order = (nargin < 3) ? 2 : mxGetScalar (pargin [2]) ;
    order = CS_MAX (order, 0) ;
    order = CS_MIN (order, 3) ;
    if (nargin == 2)
    {
        tol = 1 ;                           /* normal partial pivoting */
    }
    else if (nargin == 3)
    {
        tol = (order == 1) ? 0.001 : 1 ;    /* tol = 0.001 for amd(A+A') */
    }
    else
    {
        tol = mxGetScalar (pargin [3]) ;
    }
    if (!cs_lusol (order, A, x, tol))                       /* x = A\x */
    {
        mexErrMsgTxt ("LU factorization failed (singular or out of memory)") ;
    }
}
/* C = A' */
csr *csr_transpose (const csr *A, int values)
{
  int p, q, i, *Cp, *Cj, n, m, *Ap, *Aj, *w ;
  double *Cx, *Ax ;
  csr *C ;
  if (!CS_CSC (A)) return (NULL) ;    /* check inputs */
  m = A->m ; n = A->n ; Ap = A->p ; Aj = A->j ; Ax = A->x ;
  C = csr_spalloc (n, m, Ap [m], values && Ax, 0) ;      /* allocate result */
  w = (int*)calloc (CS_MAX(n,1), sizeof (int)) ;         /* get workspace */
  if (!C || !w) return (csr_done (C, w, NULL, 0)) ;      /* out of memory */
  Cp = C->p ; Cj = C->j ; Cx = C->x ;
  for (p = 0 ; p < Ap [m] ; p++) w [Aj [p]]++ ;          /* col counts */
  csr_cumsum (Cp, w, n) ;                                 /* col pointers */
  for (i = 0 ; i < m ; i++)
    {
      for (p = Ap [i] ; p < Ap [i+1] ; p++)
        {
	  Cj [q = w [Aj [p]]++] = i ; /* place A(i,j) as entry C(j,i) */
	  if (Cx) Cx [q] = Ax [p] ;
        }
    }
  return (csr_done (C, w, NULL, 1)) ; /* success; free w and return C */
}
Esempio n. 22
0
static void
_update_bt_statistics(_box_tree_stats_t     *bts,
                      const fvm_box_tree_t  *bt)
{
  int dim;
  size_t i;
  size_t mem_required[3];

  assert(bts != NULL);

  dim = fvm_box_tree_get_stats(bt,
                               bts->depth,
                               bts->n_leaves,
                               bts->n_boxes,
                               bts->n_threshold_leaves,
                               bts->n_leaf_boxes,
                               bts->mem_used,
                               mem_required);

  bts->dim = dim;

  for (i = 0; i < 3; i++)
    bts->mem_required[i] = CS_MAX(bts->mem_required[i], mem_required[i]);
}
Esempio n. 23
0
/* p = amd(A+A') if symmetric is true, or amd(A'A) otherwise */
CS_INT *cs_amd (CS_INT order, const cs *A)  /* order 0:natural, 1:Chol, 2:LU, 3:QR */
{
    cs *C, *A2, *AT ;
    CS_INT *Cp, *Ci, *last, *W, *len, *nv, *next, *P, *head, *elen, *degree, *w,
        *hhead, *ATp, *ATi, d, dk, dext, lemax = 0, e, elenk, eln, i, j, k, k1,
        k2, k3, jlast, ln, dense, nzmax, mindeg = 0, nvi, nvj, nvk, mark, wnvi,
        ok, cnz, nel = 0, p, p1, p2, p3, p4, pj, pk, pk1, pk2, pn, q, n, m, t ;
    unsigned CS_INT h ;
    /* --- Construct matrix C ----------------------------------------------- */
    if (!CS_CSC (A) || order <= 0 || order > 3) return (NULL) ; /* check */
    AT = cs_transpose (A, 0) ;              /* compute A' */
    if (!AT) return (NULL) ;
    m = A->m ; n = A->n ;
    dense = CS_MAX (16, 10 * sqrt ((double) n)) ;   /* find dense threshold */
    dense = CS_MIN (n-2, dense) ;
    if (order == 1 && n == m)
    {
        C = cs_add (A, AT, 0, 0) ;          /* C = A+A' */
    }
    else if (order == 2)
    {
        ATp = AT->p ;                       /* drop dense columns from AT */
        ATi = AT->i ;
        for (p2 = 0, j = 0 ; j < m ; j++)
        {
            p = ATp [j] ;                   /* column j of AT starts here */
            ATp [j] = p2 ;                  /* new column j starts here */
            if (ATp [j+1] - p > dense) continue ;   /* skip dense col j */
            for ( ; p < ATp [j+1] ; p++) ATi [p2++] = ATi [p] ;
        }
        ATp [m] = p2 ;                      /* finalize AT */
        A2 = cs_transpose (AT, 0) ;         /* A2 = AT' */
        C = A2 ? cs_multiply (AT, A2) : NULL ;  /* C=A'*A with no dense rows */
        cs_spfree (A2) ;
    }
    else
    {
        C = cs_multiply (AT, A) ;           /* C=A'*A */
    }
    cs_spfree (AT) ;
    if (!C) return (NULL) ;
    cs_fkeep (C, &cs_diag, NULL) ;          /* drop diagonal entries */
    Cp = C->p ;
    cnz = Cp [n] ;
    P = cs_malloc (n+1, sizeof (CS_INT)) ;     /* allocate result */
    W = cs_malloc (8*(n+1), sizeof (CS_INT)) ; /* get workspace */
    t = cnz + cnz/5 + 2*n ;                 /* add elbow room to C */
    if (!P || !W || !cs_sprealloc (C, t)) return (cs_idone (P, C, W, 0)) ;
    len  = W           ; nv     = W +   (n+1) ; next   = W + 2*(n+1) ;
    head = W + 3*(n+1) ; elen   = W + 4*(n+1) ; degree = W + 5*(n+1) ;
    w    = W + 6*(n+1) ; hhead  = W + 7*(n+1) ;
    last = P ;                              /* use P as workspace for last */
    /* --- Initialize quotient graph ---------------------------------------- */
    for (k = 0 ; k < n ; k++) len [k] = Cp [k+1] - Cp [k] ;
    len [n] = 0 ;
    nzmax = C->nzmax ;
    Ci = C->i ;
    for (i = 0 ; i <= n ; i++)
    {
        head [i] = -1 ;                     /* degree list i is empty */
        last [i] = -1 ;
        next [i] = -1 ;
        hhead [i] = -1 ;                    /* hash list i is empty */
        nv [i] = 1 ;                        /* node i is just one node */
        w [i] = 1 ;                         /* node i is alive */
        elen [i] = 0 ;                      /* Ek of node i is empty */
        degree [i] = len [i] ;              /* degree of node i */
    }
    mark = cs_wclear (0, 0, w, n) ;         /* clear w */
    elen [n] = -2 ;                         /* n is a dead element */
    Cp [n] = -1 ;                           /* n is a root of assembly tree */
    w [n] = 0 ;                             /* n is a dead element */
    /* --- Initialize degree lists ------------------------------------------ */
    for (i = 0 ; i < n ; i++)
    {
        d = degree [i] ;
        if (d == 0)                         /* node i is empty */
        {
            elen [i] = -2 ;                 /* element i is dead */
            nel++ ;
            Cp [i] = -1 ;                   /* i is a root of assembly tree */
            w [i] = 0 ;
        }
        else if (d > dense)                 /* node i is dense */
        {
            nv [i] = 0 ;                    /* absorb i into element n */
            elen [i] = -1 ;                 /* node i is dead */
            nel++ ;
            Cp [i] = CS_FLIP (n) ;
            nv [n]++ ;
        }
        else
        {
            if (head [d] != -1) last [head [d]] = i ;
            next [i] = head [d] ;           /* put node i in degree list d */
            head [d] = i ;
        }
    }
    while (nel < n)                         /* while (selecting pivots) do */
    {
        /* --- Select node of minimum approximate degree -------------------- */
        for (k = -1 ; mindeg < n && (k = head [mindeg]) == -1 ; mindeg++) ;
        if (next [k] != -1) last [next [k]] = -1 ;
        head [mindeg] = next [k] ;          /* remove k from degree list */
        elenk = elen [k] ;                  /* elenk = |Ek| */
        nvk = nv [k] ;                      /* # of nodes k represents */
        nel += nvk ;                        /* nv[k] nodes of A eliminated */
        /* --- Garbage collection ------------------------------------------- */
        if (elenk > 0 && cnz + mindeg >= nzmax)
        {
            for (j = 0 ; j < n ; j++)
            {
                if ((p = Cp [j]) >= 0)      /* j is a live node or element */
                {
                    Cp [j] = Ci [p] ;       /* save first entry of object */
                    Ci [p] = CS_FLIP (j) ;  /* first entry is now CS_FLIP(j) */
                }
            }
            for (q = 0, p = 0 ; p < cnz ; ) /* scan all of memory */
            {
                if ((j = CS_FLIP (Ci [p++])) >= 0)  /* found object j */
                {
                    Ci [q] = Cp [j] ;       /* restore first entry of object */
                    Cp [j] = q++ ;          /* new pointer to object j */
                    for (k3 = 0 ; k3 < len [j]-1 ; k3++) Ci [q++] = Ci [p++] ;
                }
            }
            cnz = q ;                       /* Ci [cnz...nzmax-1] now free */
        }
        /* --- Construct new element ---------------------------------------- */
        dk = 0 ;
        nv [k] = -nvk ;                     /* flag k as in Lk */
        p = Cp [k] ;
        pk1 = (elenk == 0) ? p : cnz ;      /* do in place if elen[k] == 0 */
        pk2 = pk1 ;
        for (k1 = 1 ; k1 <= elenk + 1 ; k1++)
        {
            if (k1 > elenk)
            {
                e = k ;                     /* search the nodes in k */
                pj = p ;                    /* list of nodes starts at Ci[pj]*/
                ln = len [k] - elenk ;      /* length of list of nodes in k */
            }
            else
            {
                e = Ci [p++] ;              /* search the nodes in e */
                pj = Cp [e] ;
                ln = len [e] ;              /* length of list of nodes in e */
            }
            for (k2 = 1 ; k2 <= ln ; k2++)
            {
                i = Ci [pj++] ;
                if ((nvi = nv [i]) <= 0) continue ; /* node i dead, or seen */
                dk += nvi ;                 /* degree[Lk] += size of node i */
                nv [i] = -nvi ;             /* negate nv[i] to denote i in Lk*/
                Ci [pk2++] = i ;            /* place i in Lk */
                if (next [i] != -1) last [next [i]] = last [i] ;
                if (last [i] != -1)         /* remove i from degree list */
                {
                    next [last [i]] = next [i] ;
                }
                else
                {
                    head [degree [i]] = next [i] ;
                }
            }
            if (e != k)
            {
                Cp [e] = CS_FLIP (k) ;      /* absorb e into k */
                w [e] = 0 ;                 /* e is now a dead element */
            }
        }
        if (elenk != 0) cnz = pk2 ;         /* Ci [cnz...nzmax] is free */
        degree [k] = dk ;                   /* external degree of k - |Lk\i| */
        Cp [k] = pk1 ;                      /* element k is in Ci[pk1..pk2-1] */
        len [k] = pk2 - pk1 ;
        elen [k] = -2 ;                     /* k is now an element */
        /* --- Find set differences ----------------------------------------- */
        mark = cs_wclear (mark, lemax, w, n) ;  /* clear w if necessary */
        for (pk = pk1 ; pk < pk2 ; pk++)    /* scan 1: find |Le\Lk| */
        {
            i = Ci [pk] ;
            if ((eln = elen [i]) <= 0) continue ;/* skip if elen[i] empty */
            nvi = -nv [i] ;                      /* nv [i] was negated */
            wnvi = mark - nvi ;
            for (p = Cp [i] ; p <= Cp [i] + eln - 1 ; p++)  /* scan Ei */
            {
                e = Ci [p] ;
                if (w [e] >= mark)
                {
                    w [e] -= nvi ;          /* decrement |Le\Lk| */
                }
                else if (w [e] != 0)        /* ensure e is a live element */
                {
                    w [e] = degree [e] + wnvi ; /* 1st time e seen in scan 1 */
                }
            }
        }
        /* --- Degree update ------------------------------------------------ */
        for (pk = pk1 ; pk < pk2 ; pk++)    /* scan2: degree update */
        {
            i = Ci [pk] ;                   /* consider node i in Lk */
            p1 = Cp [i] ;
            p2 = p1 + elen [i] - 1 ;
            pn = p1 ;
            for (h = 0, d = 0, p = p1 ; p <= p2 ; p++)    /* scan Ei */
            {
                e = Ci [p] ;
                if (w [e] != 0)             /* e is an unabsorbed element */
                {
                    dext = w [e] - mark ;   /* dext = |Le\Lk| */
                    if (dext > 0)
                    {
                        d += dext ;         /* sum up the set differences */
                        Ci [pn++] = e ;     /* keep e in Ei */
                        h += e ;            /* compute the hash of node i */
                    }
                    else
                    {
                        Cp [e] = CS_FLIP (k) ;  /* aggressive absorb. e->k */
                        w [e] = 0 ;             /* e is a dead element */
                    }
                }
            }
            elen [i] = pn - p1 + 1 ;        /* elen[i] = |Ei| */
            p3 = pn ;
            p4 = p1 + len [i] ;
            for (p = p2 + 1 ; p < p4 ; p++) /* prune edges in Ai */
            {
                j = Ci [p] ;
                if ((nvj = nv [j]) <= 0) continue ; /* node j dead or in Lk */
                d += nvj ;                  /* degree(i) += |j| */
                Ci [pn++] = j ;             /* place j in node list of i */
                h += j ;                    /* compute hash for node i */
            }
            if (d == 0)                     /* check for mass elimination */
            {
                Cp [i] = CS_FLIP (k) ;      /* absorb i into k */
                nvi = -nv [i] ;
                dk -= nvi ;                 /* |Lk| -= |i| */
                nvk += nvi ;                /* |k| += nv[i] */
                nel += nvi ;
                nv [i] = 0 ;
                elen [i] = -1 ;             /* node i is dead */
            }
            else
            {
                degree [i] = CS_MIN (degree [i], d) ;   /* update degree(i) */
                Ci [pn] = Ci [p3] ;         /* move first node to end */
                Ci [p3] = Ci [p1] ;         /* move 1st el. to end of Ei */
                Ci [p1] = k ;               /* add k as 1st element in of Ei */
                len [i] = pn - p1 + 1 ;     /* new len of adj. list of node i */
                h %= n ;                    /* finalize hash of i */
                next [i] = hhead [h] ;      /* place i in hash bucket */
                hhead [h] = i ;
                last [i] = h ;              /* save hash of i in last[i] */
            }
        }                                   /* scan2 is done */
        degree [k] = dk ;                   /* finalize |Lk| */
        lemax = CS_MAX (lemax, dk) ;
        mark = cs_wclear (mark+lemax, lemax, w, n) ;    /* clear w */
        /* --- Supernode detection ------------------------------------------ */
        for (pk = pk1 ; pk < pk2 ; pk++)
        {
            i = Ci [pk] ;
            if (nv [i] >= 0) continue ;         /* skip if i is dead */
            h = last [i] ;                      /* scan hash bucket of node i */
            i = hhead [h] ;
            hhead [h] = -1 ;                    /* hash bucket will be empty */
            for ( ; i != -1 && next [i] != -1 ; i = next [i], mark++)
            {
                ln = len [i] ;
                eln = elen [i] ;
                for (p = Cp [i]+1 ; p <= Cp [i] + ln-1 ; p++) w [Ci [p]] = mark;
                jlast = i ;
                for (j = next [i] ; j != -1 ; ) /* compare i with all j */
                {
                    ok = (len [j] == ln) && (elen [j] == eln) ;
                    for (p = Cp [j] + 1 ; ok && p <= Cp [j] + ln - 1 ; p++)
                    {
                        if (w [Ci [p]] != mark) ok = 0 ;    /* compare i and j*/
                    }
                    if (ok)                     /* i and j are identical */
                    {
                        Cp [j] = CS_FLIP (i) ;  /* absorb j into i */
                        nv [i] += nv [j] ;
                        nv [j] = 0 ;
                        elen [j] = -1 ;         /* node j is dead */
                        j = next [j] ;          /* delete j from hash bucket */
                        next [jlast] = j ;
                    }
                    else
                    {
                        jlast = j ;             /* j and i are different */
                        j = next [j] ;
                    }
                }
            }
        }
        /* --- Finalize new element------------------------------------------ */
        for (p = pk1, pk = pk1 ; pk < pk2 ; pk++)   /* finalize Lk */
        {
            i = Ci [pk] ;
            if ((nvi = -nv [i]) <= 0) continue ;/* skip if i is dead */
            nv [i] = nvi ;                      /* restore nv[i] */
            d = degree [i] + dk - nvi ;         /* compute external degree(i) */
            d = CS_MIN (d, n - nel - nvi) ;
            if (head [d] != -1) last [head [d]] = i ;
            next [i] = head [d] ;               /* put i back in degree list */
            last [i] = -1 ;
            head [d] = i ;
            mindeg = CS_MIN (mindeg, d) ;       /* find new minimum degree */
            degree [i] = d ;
            Ci [p++] = i ;                      /* place i in Lk */
        }
        nv [k] = nvk ;                      /* # nodes absorbed into k */
        if ((len [k] = p-pk1) == 0)         /* length of adj list of element k*/
        {
            Cp [k] = -1 ;                   /* k is a root of the tree */
            w [k] = 0 ;                     /* k is now a dead element */
        }
        if (elenk != 0) cnz = p ;           /* free unused space in Lk */
    }
    /* --- Postordering ----------------------------------------------------- */
    for (i = 0 ; i < n ; i++) Cp [i] = CS_FLIP (Cp [i]) ;/* fix assembly tree */
    for (j = 0 ; j <= n ; j++) head [j] = -1 ;
    for (j = n ; j >= 0 ; j--)              /* place unordered nodes in lists */
    {
        if (nv [j] > 0) continue ;          /* skip if j is an element */
        next [j] = head [Cp [j]] ;          /* place j in list of its parent */
        head [Cp [j]] = j ;
    }
    for (e = n ; e >= 0 ; e--)              /* place elements in lists */
    {
        if (nv [e] <= 0) continue ;         /* skip unless e is an element */
        if (Cp [e] != -1)
        {
            next [e] = head [Cp [e]] ;      /* place e in list of its parent */
            head [Cp [e]] = e ;
        }
    }
    for (k = 0, i = 0 ; i <= n ; i++)       /* postorder the assembly tree */
    {
        if (Cp [i] == -1) k = cs_tdfs (i, k, head, next, P, w) ;
    }
    return (cs_idone (P, C, W, 1)) ;
}
Esempio n. 24
0
void
cs_join_post_mesh(const char            *mesh_name,
                  const cs_join_mesh_t  *join_mesh)
{
  if (_cs_join_post_initialized == true)
    return;

  int  i, j;
  cs_lnum_t  n_vertices;

  const char *name = NULL;
  int  *ifield = NULL;
  double  *dfield = NULL;
  cs_gnum_t  *vertex_gnum = NULL;
  cs_real_t  *vertex_coord = NULL;
  cs_lnum_t  *parent_vtx_num = NULL;
  fvm_nodal_t  *post_mesh = NULL;
  fvm_writer_t  *writer = _cs_join_post_param.writer;

  const int  local_rank = CS_MAX(cs_glob_rank_id, 0);
  const cs_lnum_t  face_list_shift[2] = {0, join_mesh->n_faces};
  const cs_lnum_t  *face_vertex_idx[1] = {join_mesh->face_vtx_idx};
  const cs_lnum_t  *face_vertex_lst[1] = {join_mesh->face_vtx_lst};

  /* Define an fvm_nodal_mesh_t structure from a cs_join_mesh_t structure */

  /* Create an empty fvm_nodal_t structure. */

  if (mesh_name == NULL)
    name = join_mesh->name;
  else
    name = mesh_name;

  post_mesh = fvm_nodal_create(name, 3);

  /* Define fvm_nodal_t structure */

  fvm_nodal_from_desc_add_faces(post_mesh,
                                join_mesh->n_faces,
                                NULL,
                                1,
                                face_list_shift,
                                face_vertex_idx,
                                face_vertex_lst,
                                NULL,
                                NULL);

  /* Define vertex_coord for fvm_nodal_set_shared_vertices() */

  BFT_MALLOC(vertex_coord, 3*join_mesh->n_vertices, cs_real_t);

  for (i = 0; i < join_mesh->n_vertices; i++)
    for (j = 0; j < 3; j++)
      vertex_coord[3*i+j] = (join_mesh->vertices[i]).coord[j];

  fvm_nodal_set_shared_vertices(post_mesh, vertex_coord);

  /* Order faces by increasing global number */

  fvm_nodal_order_faces(post_mesh, join_mesh->face_gnum);
  fvm_nodal_init_io_num(post_mesh, join_mesh->face_gnum, 2);

  /* Order vertices by increasing global number */

  BFT_MALLOC(vertex_gnum, join_mesh->n_vertices, cs_gnum_t);

  for (i = 0; i < join_mesh->n_vertices; i++)
    vertex_gnum[i] = (join_mesh->vertices[i]).gnum;

  fvm_nodal_order_vertices(post_mesh, vertex_gnum);
  fvm_nodal_init_io_num(post_mesh, vertex_gnum, 0);

  /* Write current mesh */

  fvm_writer_export_nodal(writer, post_mesh);

  BFT_FREE(vertex_gnum);
  BFT_FREE(vertex_coord);

  /* Write rank associated to each face */

  BFT_MALLOC(ifield, join_mesh->n_faces, int);

  for (i = 0; i < join_mesh->n_faces; i++)
    ifield[i] = local_rank;

  _post_elt_ifield(post_mesh, _("Rank"), 1, ifield);

  BFT_FREE(ifield);

  /* Write vertex tolerance */

  n_vertices = fvm_nodal_get_n_entities(post_mesh, 0);

  BFT_MALLOC(parent_vtx_num, n_vertices, cs_lnum_t);
  BFT_MALLOC(dfield, n_vertices, double);

  fvm_nodal_get_parent_num(post_mesh, 0, parent_vtx_num);

  for (i = 0; i < n_vertices; i++) {

    cs_join_vertex_t  data = join_mesh->vertices[parent_vtx_num[i]-1];

    dfield[i] = data.tolerance;
  }

  _post_vtx_dfield(post_mesh, _("VtxTolerance"), 1, dfield);

  BFT_FREE(parent_vtx_num);
  BFT_FREE(dfield);

  post_mesh = fvm_nodal_destroy(post_mesh);
}
Esempio n. 25
0
void *cs_calloc(int n, size_t size) {

	return (calloc(CS_MAX (n,1), size));
}
Esempio n. 26
0
void *cs_malloc(int n, size_t size) {

	return (malloc(CS_MAX (n,1) * size));
}
Esempio n. 27
0
static void
_get_global_tolerance(cs_mesh_t            *mesh,
                      cs_real_t            *vtx_tolerance)
{
  cs_int_t  i, rank, vtx_id, block_size, shift;
  cs_gnum_t  first_vtx_gnum;

  cs_lnum_t n_vertices = mesh->n_vertices;
  double  *g_vtx_tolerance = NULL, *send_list = NULL, *recv_list = NULL;
  cs_int_t  *send_count = NULL, *recv_count = NULL;
  cs_int_t  *send_shift = NULL, *recv_shift = NULL;
  cs_gnum_t  *send_glist = NULL, *recv_glist = NULL;
  cs_gnum_t  n_g_vertices = mesh->n_g_vertices;
  const cs_gnum_t  *io_gnum = mesh->global_vtx_num;

  MPI_Comm  mpi_comm = cs_glob_mpi_comm;
  const int  local_rank = CS_MAX(cs_glob_rank_id, 0);
  const int  n_ranks = cs_glob_n_ranks;

  /* Define a fvm_io_num_t structure on vertices */

  block_size = n_g_vertices / n_ranks;
  if (n_g_vertices % n_ranks > 0)
    block_size += 1;

  /* Count the number of vertices to send to each rank */
  /* ------------------------------------------------- */

  BFT_MALLOC(send_count, n_ranks, int);
  BFT_MALLOC(recv_count, n_ranks, int);
  BFT_MALLOC(send_shift, n_ranks + 1, int);
  BFT_MALLOC(recv_shift, n_ranks + 1, int);

  send_shift[0] = 0;
  recv_shift[0] = 0;

  for (rank = 0; rank < n_ranks; rank++)
    send_count[rank] = 0;

  for (i = 0; i < n_vertices; i++) {
    rank = (io_gnum[i] - 1)/block_size;
    send_count[rank] += 1;
  }

  MPI_Alltoall(send_count, 1, MPI_INT, recv_count, 1, MPI_INT, mpi_comm);

  for (rank = 0; rank < n_ranks; rank++) {
    send_shift[rank + 1] = send_shift[rank] + send_count[rank];
    recv_shift[rank + 1] = recv_shift[rank] + recv_count[rank];
  }

  assert(send_shift[n_ranks] == n_vertices);

  /* Send the global numbering for each vertex */
  /* ----------------------------------------- */

  BFT_MALLOC(send_glist, n_vertices, cs_gnum_t);
  BFT_MALLOC(recv_glist, recv_shift[n_ranks], cs_gnum_t);

  for (rank = 0; rank < n_ranks; rank++)
    send_count[rank] = 0;

  for (i = 0; i < n_vertices; i++) {
    rank = (io_gnum[i] - 1)/block_size;
    shift = send_shift[rank] + send_count[rank];
    send_count[rank] += 1;
    send_glist[shift] = io_gnum[i];
  }

  MPI_Alltoallv(send_glist, send_count, send_shift, CS_MPI_GNUM,
                recv_glist, recv_count, recv_shift, CS_MPI_GNUM, mpi_comm);

  /* Send the vertex tolerance for each vertex */
  /* ----------------------------------------- */

  BFT_MALLOC(send_list, n_vertices, double);
  BFT_MALLOC(recv_list, recv_shift[n_ranks], double);

  for (rank = 0; rank < n_ranks; rank++)
    send_count[rank] = 0;

  for (i = 0; i < n_vertices; i++) {
    rank = (io_gnum[i] - 1)/block_size;
    shift = send_shift[rank] + send_count[rank];
    send_count[rank] += 1;
    send_list[shift] = vtx_tolerance[i];
  }

  MPI_Alltoallv(send_list, send_count, send_shift, MPI_DOUBLE,
                recv_list, recv_count, recv_shift, MPI_DOUBLE, mpi_comm);

  /* Define the global tolerance array */

  BFT_MALLOC(g_vtx_tolerance, block_size, double);

  for (i = 0; i < block_size; i++)
    g_vtx_tolerance[i] = DBL_MAX;

  first_vtx_gnum = block_size * local_rank + 1;

  for (i = 0; i < recv_shift[n_ranks]; i++) {
    vtx_id = recv_glist[i] - first_vtx_gnum;
    g_vtx_tolerance[vtx_id] = CS_MIN(g_vtx_tolerance[vtx_id], recv_list[i]);
  }

  /* Replace local vertex tolerance by the new computed global tolerance */

  for (i = 0; i < recv_shift[n_ranks]; i++) {
    vtx_id = recv_glist[i] - first_vtx_gnum;
    recv_list[i] = g_vtx_tolerance[vtx_id];
  }

  MPI_Alltoallv(recv_list, recv_count, recv_shift, MPI_DOUBLE,
                send_list, send_count, send_shift, MPI_DOUBLE, mpi_comm);

  for (rank = 0; rank < n_ranks; rank++)
    send_count[rank] = 0;

  for (i = 0; i < n_vertices; i++) {
    rank = (io_gnum[i] - 1)/block_size;
    shift = send_shift[rank] + send_count[rank];
    send_count[rank] += 1;
    vtx_tolerance[i] = send_list[shift];
  }

  /* Free memory */

  BFT_FREE(recv_glist);
  BFT_FREE(send_glist);
  BFT_FREE(send_list);
  BFT_FREE(recv_list);
  BFT_FREE(recv_count);
  BFT_FREE(send_count);
  BFT_FREE(recv_shift);
  BFT_FREE(send_shift);
  BFT_FREE(g_vtx_tolerance);
}
Esempio n. 28
0
static void wd_config_changed (struct cs_fsm* fsm, int32_t event, void * data)
{
	int res;
	size_t len;
	char *state;
	objdb_value_types_t type;
	char *str;
	uint64_t tmp_value;
	uint64_t next_timeout;
	struct resource *ref = (struct resource*)data;
	char str_copy[256];

	next_timeout = ref->check_timeout;

	res = api->object_key_get_typed (ref->handle,
			"poll_period",
			(void**)&str, &len,
			&type);
	if (res == 0) {
		memcpy(str_copy, str, len);
		str_copy[len] = '\0';
		if (str_to_uint64_t(str_copy, &tmp_value, WD_MIN_TIMEOUT_MS, WD_MAX_TIMEOUT_MS) == CS_OK) {
			log_printf (LOGSYS_LEVEL_DEBUG,
				"poll_period changing from:%"PRIu64" to %"PRIu64".",
				ref->check_timeout, tmp_value);
			/*
			 * To easy in the transition between poll_period's we are going
			 * to make the first timeout the bigger of the new and old value.
			 * This is to give the monitoring system time to adjust.
			 */
			next_timeout = CS_MAX(tmp_value, ref->check_timeout);
			ref->check_timeout = tmp_value;
		} else {
			log_printf (LOGSYS_LEVEL_WARNING,
				"Could NOT use poll_period:%s ms for resource %s",
				str, ref->name);
		}
	}

	res = api->object_key_get_typed (ref->handle,
		"recovery", (void*)&ref->recovery, &len, &type);
	if (res != 0) {
		/* key does not exist.
		 */
		log_printf (LOGSYS_LEVEL_WARNING,
			"resource %s missing a recovery key.", ref->name);
		cs_fsm_state_set(&ref->fsm, WD_S_STOPPED, ref);
		return;
	}
	res = api->object_key_get_typed (ref->handle,
		"state", (void*)&state, &len, &type);
	if (res != 0) {
		/* key does not exist.
		*/
		log_printf (LOGSYS_LEVEL_WARNING,
			"resource %s missing a state key.", ref->name);
		cs_fsm_state_set(&ref->fsm, WD_S_STOPPED, ref);
		return;
	}
	if (ref->check_timer) {
		api->timer_delete(ref->check_timer);
		ref->check_timer = NULL;
	}

	if (strcmp(wd_stopped_str, state) == 0) {
		cs_fsm_state_set(&ref->fsm, WD_S_STOPPED, ref);
	} else {
		api->timer_add_duration(next_timeout * MILLI_2_NANO_SECONDS,
			ref, wd_resource_check_fn, &ref->check_timer);
		cs_fsm_state_set(&ref->fsm, WD_S_RUNNING, ref);
	}
}
static double toc (double t) { double s = tic () ; return (CS_MAX (0, s-t)) ; }
Esempio n. 30
0
/*
 * return 0   - fully configured
 * return -1  - partially configured
 */
static int32_t wd_resource_create (hdb_handle_t resource_obj)
{
	int res;
	size_t len;
	char *state;
	objdb_value_types_t type;
	char period_str[32];
	char str_copy[256];
	char *str;
	uint64_t tmp_value;
	struct resource *ref = malloc (sizeof (struct resource));

	ref->handle = resource_obj;
	ref->check_timeout = WD_DEFAULT_TIMEOUT_MS;
	ref->check_timer = NULL;
	api->object_name_get (resource_obj,
		ref->name,
		&len);
	ref->name[len] = '\0';
	ref->fsm.name = ref->name;
	ref->fsm.table = wd_fsm_table;
	ref->fsm.entries = sizeof(wd_fsm_table) / sizeof(struct cs_fsm_entry);
	ref->fsm.curr_entry = 0;
	ref->fsm.curr_state = WD_S_STOPPED;
	ref->fsm.state_to_str = wd_res_state_to_str;
	ref->fsm.event_to_str = wd_res_event_to_str;
	api->object_priv_set (resource_obj, NULL);

	res = api->object_key_get_typed (resource_obj,
			"poll_period",
			(void**)&str, &len,
			&type);
	if (res != 0) {
		len = snprintf (period_str, 32, "%"PRIu64"", ref->check_timeout);
		api->object_key_create_typed (resource_obj,
			"poll_period", &period_str,
			len,
			OBJDB_VALUETYPE_STRING);
	}
	else {
		memcpy(str_copy, str, len);
		str_copy[len] = '\0';
		if (str_to_uint64_t(str_copy, &tmp_value, WD_MIN_TIMEOUT_MS, WD_MAX_TIMEOUT_MS) == CS_OK) {
			ref->check_timeout = tmp_value;
		} else {
			log_printf (LOGSYS_LEVEL_WARNING,
				"Could NOT use poll_period:%s ms for resource %s",
				str, ref->name);
		}
	}

	api->object_track_start (resource_obj, OBJECT_TRACK_DEPTH_RECURSIVE,
			wd_key_changed, NULL, wd_object_destroyed,
			NULL, ref);

	res = api->object_key_get_typed (resource_obj,
		"recovery", (void*)&ref->recovery, &len, &type);
	if (res != 0) {
		/* key does not exist.
		 */
		log_printf (LOGSYS_LEVEL_WARNING,
			"resource %s missing a recovery key.", ref->name);
		return -1;
	}
	res = api->object_key_get_typed (resource_obj,
		"state", (void*)&state, &len, &type);
	if (res != 0) {
		/* key does not exist.
		*/
		log_printf (LOGSYS_LEVEL_WARNING,
			"resource %s missing a state key.", ref->name);
		return -1;
	}

	res = api->object_key_get_typed (resource_obj,
		"last_updated", (void*)&ref->last_updated, &len, &type);
	if (res != 0) {
		/* key does not exist.
		 */
		ref->last_updated = 0;
	}

	/*
	 * delay the first check to give the monitor time to start working.
	 */
	tmp_value = CS_MAX(ref->check_timeout * 2, WD_DEFAULT_TIMEOUT_MS);
	api->timer_add_duration(tmp_value * MILLI_2_NANO_SECONDS,
		ref,
		wd_resource_check_fn, &ref->check_timer);

	cs_fsm_state_set(&ref->fsm, WD_S_RUNNING, ref);
	return 0;
}