ruby-lapack 1.7 → 1.7.1

data/ext/slatbs.c

@@ -0,0 +1,130 @@
+#include "rb_lapack.h"
+
+extern VOID slatbs_(char* uplo, char* trans, char* diag, char* normin, integer* n, integer* kd, real* ab, integer* ldab, real* x, real* scale, real* cnorm, integer* info);
+
+
+static VALUE
+rblapack_slatbs(int argc, VALUE *argv, VALUE self){
+  VALUE rblapack_uplo;
+  char uplo; 
+  VALUE rblapack_trans;
+  char trans; 
+  VALUE rblapack_diag;
+  char diag; 
+  VALUE rblapack_normin;
+  char normin; 
+  VALUE rblapack_kd;
+  integer kd; 
+  VALUE rblapack_ab;
+  real *ab; 
+  VALUE rblapack_x;
+  real *x; 
+  VALUE rblapack_cnorm;
+  real *cnorm; 
+  VALUE rblapack_scale;
+  real scale; 
+  VALUE rblapack_info;
+  integer info; 
+  VALUE rblapack_x_out__;
+  real *x_out__;
+  VALUE rblapack_cnorm_out__;
+  real *cnorm_out__;
+
+  integer ldab;
+  integer n;
+
+  VALUE rblapack_options;
+  if (argc > 0 && TYPE(argv[argc-1]) == T_HASH) {
+    argc--;
+    rblapack_options = argv[argc];
+    if (rb_hash_aref(rblapack_options, sHelp) == Qtrue) {
+      printf("%s\n", "USAGE:\n  scale, info, x, cnorm = NumRu::Lapack.slatbs( uplo, trans, diag, normin, kd, ab, x, cnorm, [:usage => usage, :help => help])\n\n\nFORTRAN MANUAL\n      SUBROUTINE SLATBS( UPLO, TRANS, DIAG, NORMIN, N, KD, AB, LDAB, X, SCALE, CNORM, INFO )\n\n*  Purpose\n*  =======\n*\n*  SLATBS solves one of the triangular systems\n*\n*     A *x = s*b  or  A'*x = s*b\n*\n*  with scaling to prevent overflow, where A is an upper or lower\n*  triangular band matrix.  Here A' denotes the transpose of A, x and b\n*  are n-element vectors, and s is a scaling factor, usually less than\n*  or equal to 1, chosen so that the components of x will be less than\n*  the overflow threshold.  If the unscaled problem will not cause\n*  overflow, the Level 2 BLAS routine STBSV is called.  If the matrix A\n*  is singular (A(j,j) = 0 for some j), then s is set to 0 and a\n*  non-trivial solution to A*x = 0 is returned.\n*\n\n*  Arguments\n*  =========\n*\n*  UPLO    (input) CHARACTER*1\n*          Specifies whether the matrix A is upper or lower triangular.\n*          = 'U':  Upper triangular\n*          = 'L':  Lower triangular\n*\n*  TRANS   (input) CHARACTER*1\n*          Specifies the operation applied to A.\n*          = 'N':  Solve A * x = s*b  (No transpose)\n*          = 'T':  Solve A'* x = s*b  (Transpose)\n*          = 'C':  Solve A'* x = s*b  (Conjugate transpose = Transpose)\n*\n*  DIAG    (input) CHARACTER*1\n*          Specifies whether or not the matrix A is unit triangular.\n*          = 'N':  Non-unit triangular\n*          = 'U':  Unit triangular\n*\n*  NORMIN  (input) CHARACTER*1\n*          Specifies whether CNORM has been set or not.\n*          = 'Y':  CNORM contains the column norms on entry\n*          = 'N':  CNORM is not set on entry.  On exit, the norms will\n*                  be computed and stored in CNORM.\n*\n*  N       (input) INTEGER\n*          The order of the matrix A.  N >= 0.\n*\n*  KD      (input) INTEGER\n*          The number of subdiagonals or superdiagonals in the\n*          triangular matrix A.  KD >= 0.\n*\n*  AB      (input) REAL array, dimension (LDAB,N)\n*          The upper or lower triangular band matrix A, stored in the\n*          first KD+1 rows of the array. The j-th column of A is stored\n*          in the j-th column of the array AB as follows:\n*          if UPLO = 'U', AB(kd+1+i-j,j) = A(i,j) for max(1,j-kd)<=i<=j;\n*          if UPLO = 'L', AB(1+i-j,j)    = A(i,j) for j<=i<=min(n,j+kd).\n*\n*  LDAB    (input) INTEGER\n*          The leading dimension of the array AB.  LDAB >= KD+1.\n*\n*  X       (input/output) REAL array, dimension (N)\n*          On entry, the right hand side b of the triangular system.\n*          On exit, X is overwritten by the solution vector x.\n*\n*  SCALE   (output) REAL\n*          The scaling factor s for the triangular system\n*             A * x = s*b  or  A'* x = s*b.\n*          If SCALE = 0, the matrix A is singular or badly scaled, and\n*          the vector x is an exact or approximate solution to A*x = 0.\n*\n*  CNORM   (input or output) REAL array, dimension (N)\n*\n*          If NORMIN = 'Y', CNORM is an input argument and CNORM(j)\n*          contains the norm of the off-diagonal part of the j-th column\n*          of A.  If TRANS = 'N', CNORM(j) must be greater than or equal\n*          to the infinity-norm, and if TRANS = 'T' or 'C', CNORM(j)\n*          must be greater than or equal to the 1-norm.\n*\n*          If NORMIN = 'N', CNORM is an output argument and CNORM(j)\n*          returns the 1-norm of the offdiagonal part of the j-th column\n*          of A.\n*\n*  INFO    (output) INTEGER\n*          = 0:  successful exit\n*          < 0:  if INFO = -k, the k-th argument had an illegal value\n*\n\n*  Further Details\n*  ======= =======\n*\n*  A rough bound on x is computed; if that is less than overflow, STBSV\n*  is called, otherwise, specific code is used which checks for possible\n*  overflow or divide-by-zero at every operation.\n*\n*  A columnwise scheme is used for solving A*x = b.  The basic algorithm\n*  if A is lower triangular is\n*\n*       x[1:n] := b[1:n]\n*       for j = 1, ..., n\n*            x(j) := x(j) / A(j,j)\n*            x[j+1:n] := x[j+1:n] - x(j) * A[j+1:n,j]\n*       end\n*\n*  Define bounds on the components of x after j iterations of the loop:\n*     M(j) = bound on x[1:j]\n*     G(j) = bound on x[j+1:n]\n*  Initially, let M(0) = 0 and G(0) = max{x(i), i=1,...,n}.\n*\n*  Then for iteration j+1 we have\n*     M(j+1) <= G(j) / | A(j+1,j+1) |\n*     G(j+1) <= G(j) + M(j+1) * | A[j+2:n,j+1] |\n*            <= G(j) ( 1 + CNORM(j+1) / | A(j+1,j+1) | )\n*\n*  where CNORM(j+1) is greater than or equal to the infinity-norm of\n*  column j+1 of A, not counting the diagonal.  Hence\n*\n*     G(j) <= G(0) product ( 1 + CNORM(i) / | A(i,i) | )\n*                  1<=i<=j\n*  and\n*\n*     |x(j)| <= ( G(0) / |A(j,j)| ) product ( 1 + CNORM(i) / |A(i,i)| )\n*                                   1<=i< j\n*\n*  Since |x(j)| <= M(j), we use the Level 2 BLAS routine STBSV if the\n*  reciprocal of the largest M(j), j=1,..,n, is larger than\n*  max(underflow, 1/overflow).\n*\n*  The bound on x(j) is also used to determine when a step in the\n*  columnwise method can be performed without fear of overflow.  If\n*  the computed bound is greater than a large constant, x is scaled to\n*  prevent overflow, but if the bound overflows, x is set to 0, x(j) to\n*  1, and scale to 0, and a non-trivial solution to A*x = 0 is found.\n*\n*  Similarly, a row-wise scheme is used to solve A'*x = b.  The basic\n*  algorithm for A upper triangular is\n*\n*       for j = 1, ..., n\n*            x(j) := ( b(j) - A[1:j-1,j]' * x[1:j-1] ) / A(j,j)\n*       end\n*\n*  We simultaneously compute two bounds\n*       G(j) = bound on ( b(i) - A[1:i-1,i]' * x[1:i-1] ), 1<=i<=j\n*       M(j) = bound on x(i), 1<=i<=j\n*\n*  The initial values are G(0) = 0, M(0) = max{b(i), i=1,..,n}, and we\n*  add the constraint G(j) >= G(j-1) and M(j) >= M(j-1) for j >= 1.\n*  Then the bound on x(j) is\n*\n*       M(j) <= M(j-1) * ( 1 + CNORM(j) ) / | A(j,j) |\n*\n*            <= M(0) * product ( ( 1 + CNORM(i) ) / |A(i,i)| )\n*                      1<=i<=j\n*\n*  and we can safely call STBSV if 1/M(n) and 1/G(n) are both greater\n*  than max(underflow, 1/overflow).\n*\n*  =====================================================================\n*\n\n");
+      return Qnil;
+    }
+    if (rb_hash_aref(rblapack_options, sUsage) == Qtrue) {
+      printf("%s\n", "USAGE:\n  scale, info, x, cnorm = NumRu::Lapack.slatbs( uplo, trans, diag, normin, kd, ab, x, cnorm, [:usage => usage, :help => help])\n");
+      return Qnil;
+    } 
+  } else
+    rblapack_options = Qnil;
+  if (argc != 8 && argc != 8)
+    rb_raise(rb_eArgError,"wrong number of arguments (%d for 8)", argc);
+  rblapack_uplo = argv[0];
+  rblapack_trans = argv[1];
+  rblapack_diag = argv[2];
+  rblapack_normin = argv[3];
+  rblapack_kd = argv[4];
+  rblapack_ab = argv[5];
+  rblapack_x = argv[6];
+  rblapack_cnorm = argv[7];
+  if (argc == 8) {
+  } else if (rblapack_options != Qnil) {
+  } else {
+  }
+
+  uplo = StringValueCStr(rblapack_uplo)[0];
+  diag = StringValueCStr(rblapack_diag)[0];
+  kd = NUM2INT(rblapack_kd);
+  if (!NA_IsNArray(rblapack_x))
+    rb_raise(rb_eArgError, "x (7th argument) must be NArray");
+  if (NA_RANK(rblapack_x) != 1)
+    rb_raise(rb_eArgError, "rank of x (7th argument) must be %d", 1);
+  n = NA_SHAPE0(rblapack_x);
+  if (NA_TYPE(rblapack_x) != NA_SFLOAT)
+    rblapack_x = na_change_type(rblapack_x, NA_SFLOAT);
+  x = NA_PTR_TYPE(rblapack_x, real*);
+  trans = StringValueCStr(rblapack_trans)[0];
+  if (!NA_IsNArray(rblapack_ab))
+    rb_raise(rb_eArgError, "ab (6th argument) must be NArray");
+  if (NA_RANK(rblapack_ab) != 2)
+    rb_raise(rb_eArgError, "rank of ab (6th argument) must be %d", 2);
+  ldab = NA_SHAPE0(rblapack_ab);
+  if (NA_SHAPE1(rblapack_ab) != n)
+    rb_raise(rb_eRuntimeError, "shape 1 of ab must be the same as shape 0 of x");
+  if (NA_TYPE(rblapack_ab) != NA_SFLOAT)
+    rblapack_ab = na_change_type(rblapack_ab, NA_SFLOAT);
+  ab = NA_PTR_TYPE(rblapack_ab, real*);
+  normin = StringValueCStr(rblapack_normin)[0];
+  if (!NA_IsNArray(rblapack_cnorm))
+    rb_raise(rb_eArgError, "cnorm (8th argument) must be NArray");
+  if (NA_RANK(rblapack_cnorm) != 1)
+    rb_raise(rb_eArgError, "rank of cnorm (8th argument) must be %d", 1);
+  if (NA_SHAPE0(rblapack_cnorm) != n)
+    rb_raise(rb_eRuntimeError, "shape 0 of cnorm must be the same as shape 0 of x");
+  if (NA_TYPE(rblapack_cnorm) != NA_SFLOAT)
+    rblapack_cnorm = na_change_type(rblapack_cnorm, NA_SFLOAT);
+  cnorm = NA_PTR_TYPE(rblapack_cnorm, real*);
+  {
+    na_shape_t shape[1];
+    shape[0] = n;
+    rblapack_x_out__ = na_make_object(NA_SFLOAT, 1, shape, cNArray);
+  }
+  x_out__ = NA_PTR_TYPE(rblapack_x_out__, real*);
+  MEMCPY(x_out__, x, real, NA_TOTAL(rblapack_x));
+  rblapack_x = rblapack_x_out__;
+  x = x_out__;
+  {
+    na_shape_t shape[1];
+    shape[0] = n;
+    rblapack_cnorm_out__ = na_make_object(NA_SFLOAT, 1, shape, cNArray);
+  }
+  cnorm_out__ = NA_PTR_TYPE(rblapack_cnorm_out__, real*);
+  MEMCPY(cnorm_out__, cnorm, real, NA_TOTAL(rblapack_cnorm));
+  rblapack_cnorm = rblapack_cnorm_out__;
+  cnorm = cnorm_out__;
+
+  slatbs_(&uplo, &trans, &diag, &normin, &n, &kd, ab, &ldab, x, &scale, cnorm, &info);
+
+  rblapack_scale = rb_float_new((double)scale);
+  rblapack_info = INT2NUM(info);
+  return rb_ary_new3(4, rblapack_scale, rblapack_info, rblapack_x, rblapack_cnorm);
+}
+
+void
+init_lapack_slatbs(VALUE mLapack, VALUE sH, VALUE sU, VALUE zero){
+  sHelp = sH;
+  sUsage = sU;
+  rblapack_ZERO = zero;
+
+  rb_define_module_function(mLapack, "slatbs", rblapack_slatbs, -1);
+}

data/ext/slatdf.c

@@ -0,0 +1,119 @@
+#include "rb_lapack.h"
+
+extern VOID slatdf_(integer* ijob, integer* n, real* z, integer* ldz, real* rhs, real* rdsum, real* rdscal, integer* ipiv, integer* jpiv);
+
+
+static VALUE
+rblapack_slatdf(int argc, VALUE *argv, VALUE self){
+  VALUE rblapack_ijob;
+  integer ijob; 
+  VALUE rblapack_z;
+  real *z; 
+  VALUE rblapack_rhs;
+  real *rhs; 
+  VALUE rblapack_rdsum;
+  real rdsum; 
+  VALUE rblapack_rdscal;
+  real rdscal; 
+  VALUE rblapack_ipiv;
+  integer *ipiv; 
+  VALUE rblapack_jpiv;
+  integer *jpiv; 
+  VALUE rblapack_rhs_out__;
+  real *rhs_out__;
+
+  integer ldz;
+  integer n;
+
+  VALUE rblapack_options;
+  if (argc > 0 && TYPE(argv[argc-1]) == T_HASH) {
+    argc--;
+    rblapack_options = argv[argc];
+    if (rb_hash_aref(rblapack_options, sHelp) == Qtrue) {
+      printf("%s\n", "USAGE:\n  rhs, rdsum, rdscal = NumRu::Lapack.slatdf( ijob, z, rhs, rdsum, rdscal, ipiv, jpiv, [:usage => usage, :help => help])\n\n\nFORTRAN MANUAL\n      SUBROUTINE SLATDF( IJOB, N, Z, LDZ, RHS, RDSUM, RDSCAL, IPIV, JPIV )\n\n*  Purpose\n*  =======\n*\n*  SLATDF uses the LU factorization of the n-by-n matrix Z computed by\n*  SGETC2 and computes a contribution to the reciprocal Dif-estimate\n*  by solving Z * x = b for x, and choosing the r.h.s. b such that\n*  the norm of x is as large as possible. On entry RHS = b holds the\n*  contribution from earlier solved sub-systems, and on return RHS = x.\n*\n*  The factorization of Z returned by SGETC2 has the form Z = P*L*U*Q,\n*  where P and Q are permutation matrices. L is lower triangular with\n*  unit diagonal elements and U is upper triangular.\n*\n\n*  Arguments\n*  =========\n*\n*  IJOB    (input) INTEGER\n*          IJOB = 2: First compute an approximative null-vector e\n*              of Z using SGECON, e is normalized and solve for\n*              Zx = +-e - f with the sign giving the greater value\n*              of 2-norm(x). About 5 times as expensive as Default.\n*          IJOB .ne. 2: Local look ahead strategy where all entries of\n*              the r.h.s. b is choosen as either +1 or -1 (Default).\n*\n*  N       (input) INTEGER\n*          The number of columns of the matrix Z.\n*\n*  Z       (input) REAL array, dimension (LDZ, N)\n*          On entry, the LU part of the factorization of the n-by-n\n*          matrix Z computed by SGETC2:  Z = P * L * U * Q\n*\n*  LDZ     (input) INTEGER\n*          The leading dimension of the array Z.  LDA >= max(1, N).\n*\n*  RHS     (input/output) REAL array, dimension N.\n*          On entry, RHS contains contributions from other subsystems.\n*          On exit, RHS contains the solution of the subsystem with\n*          entries acoording to the value of IJOB (see above).\n*\n*  RDSUM   (input/output) REAL\n*          On entry, the sum of squares of computed contributions to\n*          the Dif-estimate under computation by STGSYL, where the\n*          scaling factor RDSCAL (see below) has been factored out.\n*          On exit, the corresponding sum of squares updated with the\n*          contributions from the current sub-system.\n*          If TRANS = 'T' RDSUM is not touched.\n*          NOTE: RDSUM only makes sense when STGSY2 is called by STGSYL.\n*\n*  RDSCAL  (input/output) REAL\n*          On entry, scaling factor used to prevent overflow in RDSUM.\n*          On exit, RDSCAL is updated w.r.t. the current contributions\n*          in RDSUM.\n*          If TRANS = 'T', RDSCAL is not touched.\n*          NOTE: RDSCAL only makes sense when STGSY2 is called by\n*                STGSYL.\n*\n*  IPIV    (input) INTEGER array, dimension (N).\n*          The pivot indices; for 1 <= i <= N, row i of the\n*          matrix has been interchanged with row IPIV(i).\n*\n*  JPIV    (input) INTEGER array, dimension (N).\n*          The pivot indices; for 1 <= j <= N, column j of the\n*          matrix has been interchanged with column JPIV(j).\n*\n\n*  Further Details\n*  ===============\n*\n*  Based on contributions by\n*     Bo Kagstrom and Peter Poromaa, Department of Computing Science,\n*     Umea University, S-901 87 Umea, Sweden.\n*\n*  This routine is a further developed implementation of algorithm\n*  BSOLVE in [1] using complete pivoting in the LU factorization.\n*\n*  [1] Bo Kagstrom and Lars Westin,\n*      Generalized Schur Methods with Condition Estimators for\n*      Solving the Generalized Sylvester Equation, IEEE Transactions\n*      on Automatic Control, Vol. 34, No. 7, July 1989, pp 745-751.\n*\n*  [2] Peter Poromaa,\n*      On Efficient and Robust Estimators for the Separation\n*      between two Regular Matrix Pairs with Applications in\n*      Condition Estimation. Report IMINF-95.05, Departement of\n*      Computing Science, Umea University, S-901 87 Umea, Sweden, 1995.\n*\n*  =====================================================================\n*\n\n");
+      return Qnil;
+    }
+    if (rb_hash_aref(rblapack_options, sUsage) == Qtrue) {
+      printf("%s\n", "USAGE:\n  rhs, rdsum, rdscal = NumRu::Lapack.slatdf( ijob, z, rhs, rdsum, rdscal, ipiv, jpiv, [:usage => usage, :help => help])\n");
+      return Qnil;
+    } 
+  } else
+    rblapack_options = Qnil;
+  if (argc != 7 && argc != 7)
+    rb_raise(rb_eArgError,"wrong number of arguments (%d for 7)", argc);
+  rblapack_ijob = argv[0];
+  rblapack_z = argv[1];
+  rblapack_rhs = argv[2];
+  rblapack_rdsum = argv[3];
+  rblapack_rdscal = argv[4];
+  rblapack_ipiv = argv[5];
+  rblapack_jpiv = argv[6];
+  if (argc == 7) {
+  } else if (rblapack_options != Qnil) {
+  } else {
+  }
+
+  ijob = NUM2INT(rblapack_ijob);
+  if (!NA_IsNArray(rblapack_rhs))
+    rb_raise(rb_eArgError, "rhs (3th argument) must be NArray");
+  if (NA_RANK(rblapack_rhs) != 1)
+    rb_raise(rb_eArgError, "rank of rhs (3th argument) must be %d", 1);
+  n = NA_SHAPE0(rblapack_rhs);
+  if (NA_TYPE(rblapack_rhs) != NA_SFLOAT)
+    rblapack_rhs = na_change_type(rblapack_rhs, NA_SFLOAT);
+  rhs = NA_PTR_TYPE(rblapack_rhs, real*);
+  rdscal = (real)NUM2DBL(rblapack_rdscal);
+  if (!NA_IsNArray(rblapack_jpiv))
+    rb_raise(rb_eArgError, "jpiv (7th argument) must be NArray");
+  if (NA_RANK(rblapack_jpiv) != 1)
+    rb_raise(rb_eArgError, "rank of jpiv (7th argument) must be %d", 1);
+  if (NA_SHAPE0(rblapack_jpiv) != n)
+    rb_raise(rb_eRuntimeError, "shape 0 of jpiv must be the same as shape 0 of rhs");
+  if (NA_TYPE(rblapack_jpiv) != NA_LINT)
+    rblapack_jpiv = na_change_type(rblapack_jpiv, NA_LINT);
+  jpiv = NA_PTR_TYPE(rblapack_jpiv, integer*);
+  if (!NA_IsNArray(rblapack_z))
+    rb_raise(rb_eArgError, "z (2th argument) must be NArray");
+  if (NA_RANK(rblapack_z) != 2)
+    rb_raise(rb_eArgError, "rank of z (2th argument) must be %d", 2);
+  ldz = NA_SHAPE0(rblapack_z);
+  if (NA_SHAPE1(rblapack_z) != n)
+    rb_raise(rb_eRuntimeError, "shape 1 of z must be the same as shape 0 of rhs");
+  if (NA_TYPE(rblapack_z) != NA_SFLOAT)
+    rblapack_z = na_change_type(rblapack_z, NA_SFLOAT);
+  z = NA_PTR_TYPE(rblapack_z, real*);
+  if (!NA_IsNArray(rblapack_ipiv))
+    rb_raise(rb_eArgError, "ipiv (6th argument) must be NArray");
+  if (NA_RANK(rblapack_ipiv) != 1)
+    rb_raise(rb_eArgError, "rank of ipiv (6th argument) must be %d", 1);
+  if (NA_SHAPE0(rblapack_ipiv) != n)
+    rb_raise(rb_eRuntimeError, "shape 0 of ipiv must be the same as shape 0 of rhs");
+  if (NA_TYPE(rblapack_ipiv) != NA_LINT)
+    rblapack_ipiv = na_change_type(rblapack_ipiv, NA_LINT);
+  ipiv = NA_PTR_TYPE(rblapack_ipiv, integer*);
+  rdsum = (real)NUM2DBL(rblapack_rdsum);
+  {
+    na_shape_t shape[1];
+    shape[0] = n;
+    rblapack_rhs_out__ = na_make_object(NA_SFLOAT, 1, shape, cNArray);
+  }
+  rhs_out__ = NA_PTR_TYPE(rblapack_rhs_out__, real*);
+  MEMCPY(rhs_out__, rhs, real, NA_TOTAL(rblapack_rhs));
+  rblapack_rhs = rblapack_rhs_out__;
+  rhs = rhs_out__;
+
+  slatdf_(&ijob, &n, z, &ldz, rhs, &rdsum, &rdscal, ipiv, jpiv);
+
+  rblapack_rdsum = rb_float_new((double)rdsum);
+  rblapack_rdscal = rb_float_new((double)rdscal);
+  return rb_ary_new3(3, rblapack_rhs, rblapack_rdsum, rblapack_rdscal);
+}
+
+void
+init_lapack_slatdf(VALUE mLapack, VALUE sH, VALUE sU, VALUE zero){
+  sHelp = sH;
+  sUsage = sU;
+  rblapack_ZERO = zero;
+
+  rb_define_module_function(mLapack, "slatdf", rblapack_slatdf, -1);
+}

data/ext/slatps.c

@@ -0,0 +1,124 @@
+#include "rb_lapack.h"
+
+extern VOID slatps_(char* uplo, char* trans, char* diag, char* normin, integer* n, real* ap, real* x, real* scale, real* cnorm, integer* info);
+
+
+static VALUE
+rblapack_slatps(int argc, VALUE *argv, VALUE self){
+  VALUE rblapack_uplo;
+  char uplo; 
+  VALUE rblapack_trans;
+  char trans; 
+  VALUE rblapack_diag;
+  char diag; 
+  VALUE rblapack_normin;
+  char normin; 
+  VALUE rblapack_ap;
+  real *ap; 
+  VALUE rblapack_x;
+  real *x; 
+  VALUE rblapack_cnorm;
+  real *cnorm; 
+  VALUE rblapack_scale;
+  real scale; 
+  VALUE rblapack_info;
+  integer info; 
+  VALUE rblapack_x_out__;
+  real *x_out__;
+  VALUE rblapack_cnorm_out__;
+  real *cnorm_out__;
+
+  integer n;
+
+  VALUE rblapack_options;
+  if (argc > 0 && TYPE(argv[argc-1]) == T_HASH) {
+    argc--;
+    rblapack_options = argv[argc];
+    if (rb_hash_aref(rblapack_options, sHelp) == Qtrue) {
+      printf("%s\n", "USAGE:\n  scale, info, x, cnorm = NumRu::Lapack.slatps( uplo, trans, diag, normin, ap, x, cnorm, [:usage => usage, :help => help])\n\n\nFORTRAN MANUAL\n      SUBROUTINE SLATPS( UPLO, TRANS, DIAG, NORMIN, N, AP, X, SCALE, CNORM, INFO )\n\n*  Purpose\n*  =======\n*\n*  SLATPS solves one of the triangular systems\n*\n*     A *x = s*b  or  A'*x = s*b\n*\n*  with scaling to prevent overflow, where A is an upper or lower\n*  triangular matrix stored in packed form.  Here A' denotes the\n*  transpose of A, x and b are n-element vectors, and s is a scaling\n*  factor, usually less than or equal to 1, chosen so that the\n*  components of x will be less than the overflow threshold.  If the\n*  unscaled problem will not cause overflow, the Level 2 BLAS routine\n*  STPSV is called. If the matrix A is singular (A(j,j) = 0 for some j),\n*  then s is set to 0 and a non-trivial solution to A*x = 0 is returned.\n*\n\n*  Arguments\n*  =========\n*\n*  UPLO    (input) CHARACTER*1\n*          Specifies whether the matrix A is upper or lower triangular.\n*          = 'U':  Upper triangular\n*          = 'L':  Lower triangular\n*\n*  TRANS   (input) CHARACTER*1\n*          Specifies the operation applied to A.\n*          = 'N':  Solve A * x = s*b  (No transpose)\n*          = 'T':  Solve A'* x = s*b  (Transpose)\n*          = 'C':  Solve A'* x = s*b  (Conjugate transpose = Transpose)\n*\n*  DIAG    (input) CHARACTER*1\n*          Specifies whether or not the matrix A is unit triangular.\n*          = 'N':  Non-unit triangular\n*          = 'U':  Unit triangular\n*\n*  NORMIN  (input) CHARACTER*1\n*          Specifies whether CNORM has been set or not.\n*          = 'Y':  CNORM contains the column norms on entry\n*          = 'N':  CNORM is not set on entry.  On exit, the norms will\n*                  be computed and stored in CNORM.\n*\n*  N       (input) INTEGER\n*          The order of the matrix A.  N >= 0.\n*\n*  AP      (input) REAL array, dimension (N*(N+1)/2)\n*          The upper or lower triangular matrix A, packed columnwise in\n*          a linear array.  The j-th column of A is stored in the array\n*          AP as follows:\n*          if UPLO = 'U', AP(i + (j-1)*j/2) = A(i,j) for 1<=i<=j;\n*          if UPLO = 'L', AP(i + (j-1)*(2n-j)/2) = A(i,j) for j<=i<=n.\n*\n*  X       (input/output) REAL array, dimension (N)\n*          On entry, the right hand side b of the triangular system.\n*          On exit, X is overwritten by the solution vector x.\n*\n*  SCALE   (output) REAL\n*          The scaling factor s for the triangular system\n*             A * x = s*b  or  A'* x = s*b.\n*          If SCALE = 0, the matrix A is singular or badly scaled, and\n*          the vector x is an exact or approximate solution to A*x = 0.\n*\n*  CNORM   (input or output) REAL array, dimension (N)\n*\n*          If NORMIN = 'Y', CNORM is an input argument and CNORM(j)\n*          contains the norm of the off-diagonal part of the j-th column\n*          of A.  If TRANS = 'N', CNORM(j) must be greater than or equal\n*          to the infinity-norm, and if TRANS = 'T' or 'C', CNORM(j)\n*          must be greater than or equal to the 1-norm.\n*\n*          If NORMIN = 'N', CNORM is an output argument and CNORM(j)\n*          returns the 1-norm of the offdiagonal part of the j-th column\n*          of A.\n*\n*  INFO    (output) INTEGER\n*          = 0:  successful exit\n*          < 0:  if INFO = -k, the k-th argument had an illegal value\n*\n\n*  Further Details\n*  ======= =======\n*\n*  A rough bound on x is computed; if that is less than overflow, STPSV\n*  is called, otherwise, specific code is used which checks for possible\n*  overflow or divide-by-zero at every operation.\n*\n*  A columnwise scheme is used for solving A*x = b.  The basic algorithm\n*  if A is lower triangular is\n*\n*       x[1:n] := b[1:n]\n*       for j = 1, ..., n\n*            x(j) := x(j) / A(j,j)\n*            x[j+1:n] := x[j+1:n] - x(j) * A[j+1:n,j]\n*       end\n*\n*  Define bounds on the components of x after j iterations of the loop:\n*     M(j) = bound on x[1:j]\n*     G(j) = bound on x[j+1:n]\n*  Initially, let M(0) = 0 and G(0) = max{x(i), i=1,...,n}.\n*\n*  Then for iteration j+1 we have\n*     M(j+1) <= G(j) / | A(j+1,j+1) |\n*     G(j+1) <= G(j) + M(j+1) * | A[j+2:n,j+1] |\n*            <= G(j) ( 1 + CNORM(j+1) / | A(j+1,j+1) | )\n*\n*  where CNORM(j+1) is greater than or equal to the infinity-norm of\n*  column j+1 of A, not counting the diagonal.  Hence\n*\n*     G(j) <= G(0) product ( 1 + CNORM(i) / | A(i,i) | )\n*                  1<=i<=j\n*  and\n*\n*     |x(j)| <= ( G(0) / |A(j,j)| ) product ( 1 + CNORM(i) / |A(i,i)| )\n*                                   1<=i< j\n*\n*  Since |x(j)| <= M(j), we use the Level 2 BLAS routine STPSV if the\n*  reciprocal of the largest M(j), j=1,..,n, is larger than\n*  max(underflow, 1/overflow).\n*\n*  The bound on x(j) is also used to determine when a step in the\n*  columnwise method can be performed without fear of overflow.  If\n*  the computed bound is greater than a large constant, x is scaled to\n*  prevent overflow, but if the bound overflows, x is set to 0, x(j) to\n*  1, and scale to 0, and a non-trivial solution to A*x = 0 is found.\n*\n*  Similarly, a row-wise scheme is used to solve A'*x = b.  The basic\n*  algorithm for A upper triangular is\n*\n*       for j = 1, ..., n\n*            x(j) := ( b(j) - A[1:j-1,j]' * x[1:j-1] ) / A(j,j)\n*       end\n*\n*  We simultaneously compute two bounds\n*       G(j) = bound on ( b(i) - A[1:i-1,i]' * x[1:i-1] ), 1<=i<=j\n*       M(j) = bound on x(i), 1<=i<=j\n*\n*  The initial values are G(0) = 0, M(0) = max{b(i), i=1,..,n}, and we\n*  add the constraint G(j) >= G(j-1) and M(j) >= M(j-1) for j >= 1.\n*  Then the bound on x(j) is\n*\n*       M(j) <= M(j-1) * ( 1 + CNORM(j) ) / | A(j,j) |\n*\n*            <= M(0) * product ( ( 1 + CNORM(i) ) / |A(i,i)| )\n*                      1<=i<=j\n*\n*  and we can safely call STPSV if 1/M(n) and 1/G(n) are both greater\n*  than max(underflow, 1/overflow).\n*\n*  =====================================================================\n*\n\n");
+      return Qnil;
+    }
+    if (rb_hash_aref(rblapack_options, sUsage) == Qtrue) {
+      printf("%s\n", "USAGE:\n  scale, info, x, cnorm = NumRu::Lapack.slatps( uplo, trans, diag, normin, ap, x, cnorm, [:usage => usage, :help => help])\n");
+      return Qnil;
+    } 
+  } else
+    rblapack_options = Qnil;
+  if (argc != 7 && argc != 7)
+    rb_raise(rb_eArgError,"wrong number of arguments (%d for 7)", argc);
+  rblapack_uplo = argv[0];
+  rblapack_trans = argv[1];
+  rblapack_diag = argv[2];
+  rblapack_normin = argv[3];
+  rblapack_ap = argv[4];
+  rblapack_x = argv[5];
+  rblapack_cnorm = argv[6];
+  if (argc == 7) {
+  } else if (rblapack_options != Qnil) {
+  } else {
+  }
+
+  uplo = StringValueCStr(rblapack_uplo)[0];
+  diag = StringValueCStr(rblapack_diag)[0];
+  if (!NA_IsNArray(rblapack_x))
+    rb_raise(rb_eArgError, "x (6th argument) must be NArray");
+  if (NA_RANK(rblapack_x) != 1)
+    rb_raise(rb_eArgError, "rank of x (6th argument) must be %d", 1);
+  n = NA_SHAPE0(rblapack_x);
+  if (NA_TYPE(rblapack_x) != NA_SFLOAT)
+    rblapack_x = na_change_type(rblapack_x, NA_SFLOAT);
+  x = NA_PTR_TYPE(rblapack_x, real*);
+  trans = StringValueCStr(rblapack_trans)[0];
+  if (!NA_IsNArray(rblapack_cnorm))
+    rb_raise(rb_eArgError, "cnorm (7th argument) must be NArray");
+  if (NA_RANK(rblapack_cnorm) != 1)
+    rb_raise(rb_eArgError, "rank of cnorm (7th argument) must be %d", 1);
+  if (NA_SHAPE0(rblapack_cnorm) != n)
+    rb_raise(rb_eRuntimeError, "shape 0 of cnorm must be the same as shape 0 of x");
+  if (NA_TYPE(rblapack_cnorm) != NA_SFLOAT)
+    rblapack_cnorm = na_change_type(rblapack_cnorm, NA_SFLOAT);
+  cnorm = NA_PTR_TYPE(rblapack_cnorm, real*);
+  normin = StringValueCStr(rblapack_normin)[0];
+  if (!NA_IsNArray(rblapack_ap))
+    rb_raise(rb_eArgError, "ap (5th argument) must be NArray");
+  if (NA_RANK(rblapack_ap) != 1)
+    rb_raise(rb_eArgError, "rank of ap (5th argument) must be %d", 1);
+  if (NA_SHAPE0(rblapack_ap) != (n*(n+1)/2))
+    rb_raise(rb_eRuntimeError, "shape 0 of ap must be %d", n*(n+1)/2);
+  if (NA_TYPE(rblapack_ap) != NA_SFLOAT)
+    rblapack_ap = na_change_type(rblapack_ap, NA_SFLOAT);
+  ap = NA_PTR_TYPE(rblapack_ap, real*);
+  {
+    na_shape_t shape[1];
+    shape[0] = n;
+    rblapack_x_out__ = na_make_object(NA_SFLOAT, 1, shape, cNArray);
+  }
+  x_out__ = NA_PTR_TYPE(rblapack_x_out__, real*);
+  MEMCPY(x_out__, x, real, NA_TOTAL(rblapack_x));
+  rblapack_x = rblapack_x_out__;
+  x = x_out__;
+  {
+    na_shape_t shape[1];
+    shape[0] = n;
+    rblapack_cnorm_out__ = na_make_object(NA_SFLOAT, 1, shape, cNArray);
+  }
+  cnorm_out__ = NA_PTR_TYPE(rblapack_cnorm_out__, real*);
+  MEMCPY(cnorm_out__, cnorm, real, NA_TOTAL(rblapack_cnorm));
+  rblapack_cnorm = rblapack_cnorm_out__;
+  cnorm = cnorm_out__;
+
+  slatps_(&uplo, &trans, &diag, &normin, &n, ap, x, &scale, cnorm, &info);
+
+  rblapack_scale = rb_float_new((double)scale);
+  rblapack_info = INT2NUM(info);
+  return rb_ary_new3(4, rblapack_scale, rblapack_info, rblapack_x, rblapack_cnorm);
+}
+
+void
+init_lapack_slatps(VALUE mLapack, VALUE sH, VALUE sU, VALUE zero){
+  sHelp = sH;
+  sUsage = sU;
+  rblapack_ZERO = zero;
+
+  rb_define_module_function(mLapack, "slatps", rblapack_slatps, -1);
+}

data/ext/slatrd.c

@@ -0,0 +1,105 @@
+#include "rb_lapack.h"
+
+extern VOID slatrd_(char* uplo, integer* n, integer* nb, real* a, integer* lda, real* e, real* tau, real* w, integer* ldw);
+
+
+static VALUE
+rblapack_slatrd(int argc, VALUE *argv, VALUE self){
+  VALUE rblapack_uplo;
+  char uplo; 
+  VALUE rblapack_nb;
+  integer nb; 
+  VALUE rblapack_a;
+  real *a; 
+  VALUE rblapack_e;
+  real *e; 
+  VALUE rblapack_tau;
+  real *tau; 
+  VALUE rblapack_w;
+  real *w; 
+  VALUE rblapack_a_out__;
+  real *a_out__;
+
+  integer lda;
+  integer n;
+  integer ldw;
+
+  VALUE rblapack_options;
+  if (argc > 0 && TYPE(argv[argc-1]) == T_HASH) {
+    argc--;
+    rblapack_options = argv[argc];
+    if (rb_hash_aref(rblapack_options, sHelp) == Qtrue) {
+      printf("%s\n", "USAGE:\n  e, tau, w, a = NumRu::Lapack.slatrd( uplo, nb, a, [:usage => usage, :help => help])\n\n\nFORTRAN MANUAL\n      SUBROUTINE SLATRD( UPLO, N, NB, A, LDA, E, TAU, W, LDW )\n\n*  Purpose\n*  =======\n*\n*  SLATRD reduces NB rows and columns of a real symmetric matrix A to\n*  symmetric tridiagonal form by an orthogonal similarity\n*  transformation Q' * A * Q, and returns the matrices V and W which are\n*  needed to apply the transformation to the unreduced part of A.\n*\n*  If UPLO = 'U', SLATRD reduces the last NB rows and columns of a\n*  matrix, of which the upper triangle is supplied;\n*  if UPLO = 'L', SLATRD reduces the first NB rows and columns of a\n*  matrix, of which the lower triangle is supplied.\n*\n*  This is an auxiliary routine called by SSYTRD.\n*\n\n*  Arguments\n*  =========\n*\n*  UPLO    (input) CHARACTER*1\n*          Specifies whether the upper or lower triangular part of the\n*          symmetric matrix A is stored:\n*          = 'U': Upper triangular\n*          = 'L': Lower triangular\n*\n*  N       (input) INTEGER\n*          The order of the matrix A.\n*\n*  NB      (input) INTEGER\n*          The number of rows and columns to be reduced.\n*\n*  A       (input/output) REAL array, dimension (LDA,N)\n*          On entry, the symmetric matrix A.  If UPLO = 'U', the leading\n*          n-by-n upper triangular part of A contains the upper\n*          triangular part of the matrix A, and the strictly lower\n*          triangular part of A is not referenced.  If UPLO = 'L', the\n*          leading n-by-n lower triangular part of A contains the lower\n*          triangular part of the matrix A, and the strictly upper\n*          triangular part of A is not referenced.\n*          On exit:\n*          if UPLO = 'U', the last NB columns have been reduced to\n*            tridiagonal form, with the diagonal elements overwriting\n*            the diagonal elements of A; the elements above the diagonal\n*            with the array TAU, represent the orthogonal matrix Q as a\n*            product of elementary reflectors;\n*          if UPLO = 'L', the first NB columns have been reduced to\n*            tridiagonal form, with the diagonal elements overwriting\n*            the diagonal elements of A; the elements below the diagonal\n*            with the array TAU, represent the  orthogonal matrix Q as a\n*            product of elementary reflectors.\n*          See Further Details.\n*\n*  LDA     (input) INTEGER\n*          The leading dimension of the array A.  LDA >= (1,N).\n*\n*  E       (output) REAL array, dimension (N-1)\n*          If UPLO = 'U', E(n-nb:n-1) contains the superdiagonal\n*          elements of the last NB columns of the reduced matrix;\n*          if UPLO = 'L', E(1:nb) contains the subdiagonal elements of\n*          the first NB columns of the reduced matrix.\n*\n*  TAU     (output) REAL array, dimension (N-1)\n*          The scalar factors of the elementary reflectors, stored in\n*          TAU(n-nb:n-1) if UPLO = 'U', and in TAU(1:nb) if UPLO = 'L'.\n*          See Further Details.\n*\n*  W       (output) REAL array, dimension (LDW,NB)\n*          The n-by-nb matrix W required to update the unreduced part\n*          of A.\n*\n*  LDW     (input) INTEGER\n*          The leading dimension of the array W. LDW >= max(1,N).\n*\n\n*  Further Details\n*  ===============\n*\n*  If UPLO = 'U', the matrix Q is represented as a product of elementary\n*  reflectors\n*\n*     Q = H(n) H(n-1) . . . H(n-nb+1).\n*\n*  Each H(i) has the form\n*\n*     H(i) = I - tau * v * v'\n*\n*  where tau is a real scalar, and v is a real vector with\n*  v(i:n) = 0 and v(i-1) = 1; v(1:i-1) is stored on exit in A(1:i-1,i),\n*  and tau in TAU(i-1).\n*\n*  If UPLO = 'L', the matrix Q is represented as a product of elementary\n*  reflectors\n*\n*     Q = H(1) H(2) . . . H(nb).\n*\n*  Each H(i) has the form\n*\n*     H(i) = I - tau * v * v'\n*\n*  where tau is a real scalar, and v is a real vector with\n*  v(1:i) = 0 and v(i+1) = 1; v(i+1:n) is stored on exit in A(i+1:n,i),\n*  and tau in TAU(i).\n*\n*  The elements of the vectors v together form the n-by-nb matrix V\n*  which is needed, with W, to apply the transformation to the unreduced\n*  part of the matrix, using a symmetric rank-2k update of the form:\n*  A := A - V*W' - W*V'.\n*\n*  The contents of A on exit are illustrated by the following examples\n*  with n = 5 and nb = 2:\n*\n*  if UPLO = 'U':                       if UPLO = 'L':\n*\n*    (  a   a   a   v4  v5 )              (  d                  )\n*    (      a   a   v4  v5 )              (  1   d              )\n*    (          a   1   v5 )              (  v1  1   a          )\n*    (              d   1  )              (  v1  v2  a   a      )\n*    (                  d  )              (  v1  v2  a   a   a  )\n*\n*  where d denotes a diagonal element of the reduced matrix, a denotes\n*  an element of the original matrix that is unchanged, and vi denotes\n*  an element of the vector defining H(i).\n*\n*  =====================================================================\n*\n\n");
+      return Qnil;
+    }
+    if (rb_hash_aref(rblapack_options, sUsage) == Qtrue) {
+      printf("%s\n", "USAGE:\n  e, tau, w, a = NumRu::Lapack.slatrd( uplo, nb, a, [:usage => usage, :help => help])\n");
+      return Qnil;
+    } 
+  } else
+    rblapack_options = Qnil;
+  if (argc != 3 && argc != 3)
+    rb_raise(rb_eArgError,"wrong number of arguments (%d for 3)", argc);
+  rblapack_uplo = argv[0];
+  rblapack_nb = argv[1];
+  rblapack_a = argv[2];
+  if (argc == 3) {
+  } else if (rblapack_options != Qnil) {
+  } else {
+  }
+
+  uplo = StringValueCStr(rblapack_uplo)[0];
+  if (!NA_IsNArray(rblapack_a))
+    rb_raise(rb_eArgError, "a (3th argument) must be NArray");
+  if (NA_RANK(rblapack_a) != 2)
+    rb_raise(rb_eArgError, "rank of a (3th argument) must be %d", 2);
+  lda = NA_SHAPE0(rblapack_a);
+  n = NA_SHAPE1(rblapack_a);
+  if (NA_TYPE(rblapack_a) != NA_SFLOAT)
+    rblapack_a = na_change_type(rblapack_a, NA_SFLOAT);
+  a = NA_PTR_TYPE(rblapack_a, real*);
+  nb = NUM2INT(rblapack_nb);
+  ldw = MAX(1,n);
+  {
+    na_shape_t shape[1];
+    shape[0] = n-1;
+    rblapack_e = na_make_object(NA_SFLOAT, 1, shape, cNArray);
+  }
+  e = NA_PTR_TYPE(rblapack_e, real*);
+  {
+    na_shape_t shape[1];
+    shape[0] = n-1;
+    rblapack_tau = na_make_object(NA_SFLOAT, 1, shape, cNArray);
+  }
+  tau = NA_PTR_TYPE(rblapack_tau, real*);
+  {
+    na_shape_t shape[2];
+    shape[0] = ldw;
+    shape[1] = MAX(n,nb);
+    rblapack_w = na_make_object(NA_SFLOAT, 2, shape, cNArray);
+  }
+  w = NA_PTR_TYPE(rblapack_w, real*);
+  {
+    na_shape_t shape[2];
+    shape[0] = lda;
+    shape[1] = n;
+    rblapack_a_out__ = na_make_object(NA_SFLOAT, 2, shape, cNArray);
+  }
+  a_out__ = NA_PTR_TYPE(rblapack_a_out__, real*);
+  MEMCPY(a_out__, a, real, NA_TOTAL(rblapack_a));
+  rblapack_a = rblapack_a_out__;
+  a = a_out__;
+
+  slatrd_(&uplo, &n, &nb, a, &lda, e, tau, w, &ldw);
+
+  return rb_ary_new3(4, rblapack_e, rblapack_tau, rblapack_w, rblapack_a);
+}
+
+void
+init_lapack_slatrd(VALUE mLapack, VALUE sH, VALUE sU, VALUE zero){
+  sHelp = sH;
+  sUsage = sU;
+  rblapack_ZERO = zero;
+
+  rb_define_module_function(mLapack, "slatrd", rblapack_slatrd, -1);
+}