floraachy
/
OpenBLAS
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
OpenBLAS/lapack-netlib/SRC/sgbtrf.c

888 lines
24 KiB
C
Raw Permalink Blame History

#include <math.h>
#include <stdlib.h>
#include <string.h>
#include <stdio.h>
#include <complex.h>
#ifdef complex
#undef complex
#endif
#ifdef I
#undef I
#endif
#if defined(_WIN64)
typedef long long BLASLONG;
typedef unsigned long long BLASULONG;
#else
typedef long BLASLONG;
typedef unsigned long BLASULONG;
#endif
#ifdef LAPACK_ILP64
typedef BLASLONG blasint;
#if defined(_WIN64)
#define blasabs(x) llabs(x)
#else
#define blasabs(x) labs(x)
#endif
#else
typedef int blasint;
#define blasabs(x) abs(x)
#endif
typedef blasint integer;
typedef unsigned int uinteger;
typedef char *address;
typedef short int shortint;
typedef float real;
typedef double doublereal;
typedef struct { real r, i; } complex;
typedef struct { doublereal r, i; } doublecomplex;
#ifdef _MSC_VER
static inline _Fcomplex Cf(complex *z) {_Fcomplex zz={z->r , z->i}; return zz;}
static inline _Dcomplex Cd(doublecomplex *z) {_Dcomplex zz={z->r , z->i};return zz;}
static inline _Fcomplex * _pCf(complex *z) {return (_Fcomplex*)z;}
static inline _Dcomplex * _pCd(doublecomplex *z) {return (_Dcomplex*)z;}
#else
static inline _Complex float Cf(complex *z) {return z->r + z->i*_Complex_I;}
static inline _Complex double Cd(doublecomplex *z) {return z->r + z->i*_Complex_I;}
static inline _Complex float * _pCf(complex *z) {return (_Complex float*)z;}
static inline _Complex double * _pCd(doublecomplex *z) {return (_Complex double*)z;}
#endif
#define pCf(z) (*_pCf(z))
#define pCd(z) (*_pCd(z))
typedef blasint logical;
typedef char logical1;
typedef char integer1;
#define TRUE_ (1)
#define FALSE_ (0)
/* Extern is for use with -E */
#ifndef Extern
#define Extern extern
#endif
/* I/O stuff */
typedef int flag;
typedef int ftnlen;
typedef int ftnint;
/*external read, write*/
typedef struct
{ flag cierr;
ftnint ciunit;
flag ciend;
char *cifmt;
ftnint cirec;
} cilist;
/*internal read, write*/
typedef struct
{ flag icierr;
char *iciunit;
flag iciend;
char *icifmt;
ftnint icirlen;
ftnint icirnum;
} icilist;
/*open*/
typedef struct
{ flag oerr;
ftnint ounit;
char *ofnm;
ftnlen ofnmlen;
char *osta;
char *oacc;
char *ofm;
ftnint orl;
char *oblnk;
} olist;
/*close*/
typedef struct
{ flag cerr;
ftnint cunit;
char *csta;
} cllist;
/*rewind, backspace, endfile*/
typedef struct
{ flag aerr;
ftnint aunit;
} alist;
/* inquire */
typedef struct
{ flag inerr;
ftnint inunit;
char *infile;
ftnlen infilen;
ftnint *inex; /*parameters in standard's order*/
ftnint *inopen;
ftnint *innum;
ftnint *innamed;
char *inname;
ftnlen innamlen;
char *inacc;
ftnlen inacclen;
char *inseq;
ftnlen inseqlen;
char *indir;
ftnlen indirlen;
char *infmt;
ftnlen infmtlen;
char *inform;
ftnint informlen;
char *inunf;
ftnlen inunflen;
ftnint *inrecl;
ftnint *innrec;
char *inblank;
ftnlen inblanklen;
} inlist;
#define VOID void
union Multitype { /* for multiple entry points */
integer1 g;
shortint h;
integer i;
/* longint j; */
real r;
doublereal d;
complex c;
doublecomplex z;
};
typedef union Multitype Multitype;
struct Vardesc { /* for Namelist */
char *name;
char *addr;
ftnlen *dims;
int type;
};
typedef struct Vardesc Vardesc;
struct Namelist {
char *name;
Vardesc **vars;
int nvars;
};
typedef struct Namelist Namelist;
#define abs(x) ((x) >= 0 ? (x) : -(x))
#define dabs(x) (fabs(x))
#define f2cmin(a,b) ((a) <= (b) ? (a) : (b))
#define f2cmax(a,b) ((a) >= (b) ? (a) : (b))
#define dmin(a,b) (f2cmin(a,b))
#define dmax(a,b) (f2cmax(a,b))
#define bit_test(a,b) ((a) >> (b) & 1)
#define bit_clear(a,b) ((a) & ~((uinteger)1 << (b)))
#define bit_set(a,b) ((a) | ((uinteger)1 << (b)))
#define abort_() { sig_die("Fortran abort routine called", 1); }
#define c_abs(z) (cabsf(Cf(z)))
#define c_cos(R,Z) { pCf(R)=ccos(Cf(Z)); }
#ifdef _MSC_VER
#define c_div(c, a, b) {Cf(c)._Val[0] = (Cf(a)._Val[0]/Cf(b)._Val[0]); Cf(c)._Val[1]=(Cf(a)._Val[1]/Cf(b)._Val[1]);}
#define z_div(c, a, b) {Cd(c)._Val[0] = (Cd(a)._Val[0]/Cd(b)._Val[0]); Cd(c)._Val[1]=(Cd(a)._Val[1]/df(b)._Val[1]);}
#else
#define c_div(c, a, b) {pCf(c) = Cf(a)/Cf(b);}
#define z_div(c, a, b) {pCd(c) = Cd(a)/Cd(b);}
#endif
#define c_exp(R, Z) {pCf(R) = cexpf(Cf(Z));}
#define c_log(R, Z) {pCf(R) = clogf(Cf(Z));}
#define c_sin(R, Z) {pCf(R) = csinf(Cf(Z));}
//#define c_sqrt(R, Z) {*(R) = csqrtf(Cf(Z));}
#define c_sqrt(R, Z) {pCf(R) = csqrtf(Cf(Z));}
#define d_abs(x) (fabs(*(x)))
#define d_acos(x) (acos(*(x)))
#define d_asin(x) (asin(*(x)))
#define d_atan(x) (atan(*(x)))
#define d_atn2(x, y) (atan2(*(x),*(y)))
#define d_cnjg(R, Z) { pCd(R) = conj(Cd(Z)); }
#define r_cnjg(R, Z) { pCf(R) = conjf(Cf(Z)); }
#define d_cos(x) (cos(*(x)))
#define d_cosh(x) (cosh(*(x)))
#define d_dim(__a, __b) ( *(__a) > *(__b) ? *(__a) - *(__b) : 0.0 )
#define d_exp(x) (exp(*(x)))
#define d_imag(z) (cimag(Cd(z)))
#define r_imag(z) (cimagf(Cf(z)))
#define d_int(__x) (*(__x)>0 ? floor(*(__x)) : -floor(- *(__x)))
#define r_int(__x) (*(__x)>0 ? floor(*(__x)) : -floor(- *(__x)))
#define d_lg10(x) ( 0.43429448190325182765 * log(*(x)) )
#define r_lg10(x) ( 0.43429448190325182765 * log(*(x)) )
#define d_log(x) (log(*(x)))
#define d_mod(x, y) (fmod(*(x), *(y)))
#define u_nint(__x) ((__x)>=0 ? floor((__x) + .5) : -floor(.5 - (__x)))
#define d_nint(x) u_nint(*(x))
#define u_sign(__a,__b) ((__b) >= 0 ? ((__a) >= 0 ? (__a) : -(__a)) : -((__a) >= 0 ? (__a) : -(__a)))
#define d_sign(a,b) u_sign(*(a),*(b))
#define r_sign(a,b) u_sign(*(a),*(b))
#define d_sin(x) (sin(*(x)))
#define d_sinh(x) (sinh(*(x)))
#define d_sqrt(x) (sqrt(*(x)))
#define d_tan(x) (tan(*(x)))
#define d_tanh(x) (tanh(*(x)))
#define i_abs(x) abs(*(x))
#define i_dnnt(x) ((integer)u_nint(*(x)))
#define i_len(s, n) (n)
#define i_nint(x) ((integer)u_nint(*(x)))
#define i_sign(a,b) ((integer)u_sign((integer)*(a),(integer)*(b)))
#define s_cat(lpp, rpp, rnp, np, llp) { ftnlen i, nc, ll; char *f__rp, *lp; ll = (llp); lp = (lpp); for(i=0; i < (int)*(np); ++i) { nc = ll; if((rnp)[i] < nc) nc = (rnp)[i]; ll -= nc; f__rp = (rpp)[i]; while(--nc >= 0) *lp++ = *(f__rp)++; } while(--ll >= 0) *lp++ = ' '; }
#define s_cmp(a,b,c,d) ((integer)strncmp((a),(b),f2cmin((c),(d))))
#define s_copy(A,B,C,D) { int __i,__m; for (__i=0, __m=f2cmin((C),(D)); __i<__m && (B)[__i] != 0; ++__i) (A)[__i] = (B)[__i]; }
#define sig_die(s, kill) { exit(1); }
#define s_stop(s, n) {exit(0);}
#define z_abs(z) (cabs(Cd(z)))
#define z_exp(R, Z) {pCd(R) = cexp(Cd(Z));}
#define z_sqrt(R, Z) {pCd(R) = csqrt(Cd(Z));}
#define myexit_() break;
#define mycycle() continue;
#define myceiling(w) {ceil(w)}
#define myhuge(w) {HUGE_VAL}
//#define mymaxloc_(w,s,e,n) {if (sizeof(*(w)) == sizeof(double)) dmaxloc_((w),*(s),*(e),n); else dmaxloc_((w),*(s),*(e),n);}
#define mymaxloc(w,s,e,n) {dmaxloc_(w,*(s),*(e),n)}
/* -- translated by f2c (version 20000121).
You must link the resulting object file with the libraries:
-lf2c -lm (in that order)
*/
/* Table of constant values */
static integer c__1 = 1;
static integer c__65 = 65;
static real c_b18 = -1.f;
static real c_b31 = 1.f;
/* > \brief \b SGBTRF */
/* =========== DOCUMENTATION =========== */
/* Online html documentation available at */
/* http://www.netlib.org/lapack/explore-html/ */
/* > \htmlonly */
/* > Download SGBTRF + dependencies */
/* > <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/sgbtrf.
f"> */
/* > [TGZ]</a> */
/* > <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/sgbtrf.
f"> */
/* > [ZIP]</a> */
/* > <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/sgbtrf.
f"> */
/* > [TXT]</a> */
/* > \endhtmlonly */
/* Definition: */
/* =========== */
/* SUBROUTINE SGBTRF( M, N, KL, KU, AB, LDAB, IPIV, INFO ) */
/* INTEGER INFO, KL, KU, LDAB, M, N */
/* INTEGER IPIV( * ) */
/* REAL AB( LDAB, * ) */
/* > \par Purpose: */
/* ============= */
/* > */
/* > \verbatim */
/* > */
/* > SGBTRF computes an LU factorization of a real m-by-n band matrix A */
/* > using partial pivoting with row interchanges. */
/* > */
/* > This is the blocked version of the algorithm, calling Level 3 BLAS. */
/* > \endverbatim */
/* Arguments: */
/* ========== */
/* > \param[in] M */
/* > \verbatim */
/* > M is INTEGER */
/* > The number of rows of the matrix A. M >= 0. */
/* > \endverbatim */
/* > */
/* > \param[in] N */
/* > \verbatim */
/* > N is INTEGER */
/* > The number of columns of the matrix A. N >= 0. */
/* > \endverbatim */
/* > */
/* > \param[in] KL */
/* > \verbatim */
/* > KL is INTEGER */
/* > The number of subdiagonals within the band of A. KL >= 0. */
/* > \endverbatim */
/* > */
/* > \param[in] KU */
/* > \verbatim */
/* > KU is INTEGER */
/* > The number of superdiagonals within the band of A. KU >= 0. */
/* > \endverbatim */
/* > */
/* > \param[in,out] AB */
/* > \verbatim */
/* > AB is REAL array, dimension (LDAB,N) */
/* > On entry, the matrix A in band storage, in rows KL+1 to */
/* > 2*KL+KU+1; rows 1 to KL of the array need not be set. */
/* > The j-th column of A is stored in the j-th column of the */
/* > array AB as follows: */
/* > AB(kl+ku+1+i-j,j) = A(i,j) for f2cmax(1,j-ku)<=i<=f2cmin(m,j+kl) */
/* > */
/* > On exit, details of the factorization: U is stored as an */
/* > upper triangular band matrix with KL+KU superdiagonals in */
/* > rows 1 to KL+KU+1, and the multipliers used during the */
/* > factorization are stored in rows KL+KU+2 to 2*KL+KU+1. */
/* > See below for further details. */
/* > \endverbatim */
/* > */
/* > \param[in] LDAB */
/* > \verbatim */
/* > LDAB is INTEGER */
/* > The leading dimension of the array AB. LDAB >= 2*KL+KU+1. */
/* > \endverbatim */
/* > */
/* > \param[out] IPIV */
/* > \verbatim */
/* > IPIV is INTEGER array, dimension (f2cmin(M,N)) */
/* > The pivot indices; for 1 <= i <= f2cmin(M,N), row i of the */
/* > matrix was interchanged with row IPIV(i). */
/* > \endverbatim */
/* > */
/* > \param[out] INFO */
/* > \verbatim */
/* > INFO is INTEGER */
/* > = 0: successful exit */
/* > < 0: if INFO = -i, the i-th argument had an illegal value */
/* > > 0: if INFO = +i, U(i,i) is exactly zero. The factorization */
/* > has been completed, but the factor U is exactly */
/* > singular, and division by zero will occur if it is used */
/* > to solve a system of equations. */
/* > \endverbatim */
/* Authors: */
/* ======== */
/* > \author Univ. of Tennessee */
/* > \author Univ. of California Berkeley */
/* > \author Univ. of Colorado Denver */
/* > \author NAG Ltd. */
/* > \date December 2016 */
/* > \ingroup realGBcomputational */
/* > \par Further Details: */
/* ===================== */
/* > */
/* > \verbatim */
/* > */
/* > The band storage scheme is illustrated by the following example, when */
/* > M = N = 6, KL = 2, KU = 1: */
/* > */
/* > On entry: On exit: */
/* > */
/* > * * * + + + * * * u14 u25 u36 */
/* > * * + + + + * * u13 u24 u35 u46 */
/* > * a12 a23 a34 a45 a56 * u12 u23 u34 u45 u56 */
/* > a11 a22 a33 a44 a55 a66 u11 u22 u33 u44 u55 u66 */
/* > a21 a32 a43 a54 a65 * m21 m32 m43 m54 m65 * */
/* > a31 a42 a53 a64 * * m31 m42 m53 m64 * * */
/* > */
/* > Array elements marked * are not used by the routine; elements marked */
/* > + need not be set on entry, but are required by the routine to store */
/* > elements of U because of fill-in resulting from the row interchanges. */
/* > \endverbatim */
/* > */
/* ===================================================================== */
/* Subroutine */ void sgbtrf_(integer *m, integer *n, integer *kl, integer *ku,
real *ab, integer *ldab, integer *ipiv, integer *info)
{
/* System generated locals */
integer ab_dim1, ab_offset, i__1, i__2, i__3, i__4, i__5, i__6;
real r__1;
/* Local variables */
extern /* Subroutine */ void sger_(integer *, integer *, real *, real *,
integer *, real *, integer *, real *, integer *);
real temp;
integer i__, j;
extern /* Subroutine */ void sscal_(integer *, real *, real *, integer *),
sgemm_(char *, char *, integer *, integer *, integer *, real *,
real *, integer *, real *, integer *, real *, real *, integer *);
real work13[4160] /* was [65][64] */, work31[4160] /* was [65][
64] */;
integer i2, i3, j2, j3, k2;
extern /* Subroutine */ void scopy_(integer *, real *, integer *, real *,
integer *), sswap_(integer *, real *, integer *, real *, integer *
), strsm_(char *, char *, char *, char *, integer *, integer *,
real *, real *, integer *, real *, integer *), sgbtf2_(integer *, integer *, integer *, integer
*, real *, integer *, integer *, integer *);
integer jb, nb, ii, jj, jm, ip, jp, km, ju, kv, nw;
extern /* Subroutine */ int xerbla_(char *, integer *, ftnlen);
extern integer ilaenv_(integer *, char *, char *, integer *, integer *,
integer *, integer *, ftnlen, ftnlen), isamax_(integer *, real *,
integer *);
extern /* Subroutine */ int slaswp_(integer *, real *, integer *, integer
*, integer *, integer *, integer *);
/* -- LAPACK computational routine (version 3.7.0) -- */
/* -- LAPACK is a software package provided by Univ. of Tennessee, -- */
/* -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..-- */
/* December 2016 */
/* ===================================================================== */
/* KV is the number of superdiagonals in the factor U, allowing for */
/* fill-in */
/* Parameter adjustments */
ab_dim1 = *ldab;
ab_offset = 1 + ab_dim1 * 1;
ab -= ab_offset;
--ipiv;
/* Function Body */
kv = *ku + *kl;
/* Test the input parameters. */
*info = 0;
if (*m < 0) {
*info = -1;
} else if (*n < 0) {
*info = -2;
} else if (*kl < 0) {
*info = -3;
} else if (*ku < 0) {
*info = -4;
} else if (*ldab < *kl + kv + 1) {
*info = -6;
}
if (*info != 0) {
i__1 = -(*info);
xerbla_("SGBTRF", &i__1, (ftnlen)6);
return;
}
/* Quick return if possible */
if (*m == 0 || *n == 0) {
return;
}
/* Determine the block size for this environment */
nb = ilaenv_(&c__1, "SGBTRF", " ", m, n, kl, ku, (ftnlen)6, (ftnlen)1);
/* The block size must not exceed the limit set by the size of the */
/* local arrays WORK13 and WORK31. */
nb = f2cmin(nb,64);
if (nb <= 1 || nb > *kl) {
/* Use unblocked code */
sgbtf2_(m, n, kl, ku, &ab[ab_offset], ldab, &ipiv[1], info);
} else {
/* Use blocked code */
/* Zero the superdiagonal elements of the work array WORK13 */
i__1 = nb;
for (j = 1; j <= i__1; ++j) {
i__2 = j - 1;
for (i__ = 1; i__ <= i__2; ++i__) {
work13[i__ + j * 65 - 66] = 0.f;
/* L10: */
}
/* L20: */
}
/* Zero the subdiagonal elements of the work array WORK31 */
i__1 = nb;
for (j = 1; j <= i__1; ++j) {
i__2 = nb;
for (i__ = j + 1; i__ <= i__2; ++i__) {
work31[i__ + j * 65 - 66] = 0.f;
/* L30: */
}
/* L40: */
}
/* Gaussian elimination with partial pivoting */
/* Set fill-in elements in columns KU+2 to KV to zero */
i__1 = f2cmin(kv,*n);
for (j = *ku + 2; j <= i__1; ++j) {
i__2 = *kl;
for (i__ = kv - j + 2; i__ <= i__2; ++i__) {
ab[i__ + j * ab_dim1] = 0.f;
/* L50: */
}
/* L60: */
}
/* JU is the index of the last column affected by the current */
/* stage of the factorization */
ju = 1;
i__1 = f2cmin(*m,*n);
i__2 = nb;
for (j = 1; i__2 < 0 ? j >= i__1 : j <= i__1; j += i__2) {
/* Computing MIN */
i__3 = nb, i__4 = f2cmin(*m,*n) - j + 1;
jb = f2cmin(i__3,i__4);
/* The active part of the matrix is partitioned */
/* A11 A12 A13 */
/* A21 A22 A23 */
/* A31 A32 A33 */
/* Here A11, A21 and A31 denote the current block of JB columns */
/* which is about to be factorized. The number of rows in the */
/* partitioning are JB, I2, I3 respectively, and the numbers */
/* of columns are JB, J2, J3. The superdiagonal elements of A13 */
/* and the subdiagonal elements of A31 lie outside the band. */
/* Computing MIN */
i__3 = *kl - jb, i__4 = *m - j - jb + 1;
i2 = f2cmin(i__3,i__4);
/* Computing MIN */
i__3 = jb, i__4 = *m - j - *kl + 1;
i3 = f2cmin(i__3,i__4);
/* J2 and J3 are computed after JU has been updated. */
/* Factorize the current block of JB columns */
i__3 = j + jb - 1;
for (jj = j; jj <= i__3; ++jj) {
/* Set fill-in elements in column JJ+KV to zero */
if (jj + kv <= *n) {
i__4 = *kl;
for (i__ = 1; i__ <= i__4; ++i__) {
ab[i__ + (jj + kv) * ab_dim1] = 0.f;
/* L70: */
}
}
/* Find pivot and test for singularity. KM is the number of */
/* subdiagonal elements in the current column. */
/* Computing MIN */
i__4 = *kl, i__5 = *m - jj;
km = f2cmin(i__4,i__5);
i__4 = km + 1;
jp = isamax_(&i__4, &ab[kv + 1 + jj * ab_dim1], &c__1);
ipiv[jj] = jp + jj - j;
if (ab[kv + jp + jj * ab_dim1] != 0.f) {
/* Computing MAX */
/* Computing MIN */
i__6 = jj + *ku + jp - 1;
i__4 = ju, i__5 = f2cmin(i__6,*n);
ju = f2cmax(i__4,i__5);
if (jp != 1) {
/* Apply interchange to columns J to J+JB-1 */
if (jp + jj - 1 < j + *kl) {
i__4 = *ldab - 1;
i__5 = *ldab - 1;
sswap_(&jb, &ab[kv + 1 + jj - j + j * ab_dim1], &
i__4, &ab[kv + jp + jj - j + j * ab_dim1],
&i__5);
} else {
/* The interchange affects columns J to JJ-1 of A31 */
/* which are stored in the work array WORK31 */
i__4 = jj - j;
i__5 = *ldab - 1;
sswap_(&i__4, &ab[kv + 1 + jj - j + j * ab_dim1],
&i__5, &work31[jp + jj - j - *kl - 1], &
c__65);
i__4 = j + jb - jj;
i__5 = *ldab - 1;
i__6 = *ldab - 1;
sswap_(&i__4, &ab[kv + 1 + jj * ab_dim1], &i__5, &
ab[kv + jp + jj * ab_dim1], &i__6);
}
}
/* Compute multipliers */
r__1 = 1.f / ab[kv + 1 + jj * ab_dim1];
sscal_(&km, &r__1, &ab[kv + 2 + jj * ab_dim1], &c__1);
/* Update trailing submatrix within the band and within */
/* the current block. JM is the index of the last column */
/* which needs to be updated. */
/* Computing MIN */
i__4 = ju, i__5 = j + jb - 1;
jm = f2cmin(i__4,i__5);
if (jm > jj) {
i__4 = jm - jj;
i__5 = *ldab - 1;
i__6 = *ldab - 1;
sger_(&km, &i__4, &c_b18, &ab[kv + 2 + jj * ab_dim1],
&c__1, &ab[kv + (jj + 1) * ab_dim1], &i__5, &
ab[kv + 1 + (jj + 1) * ab_dim1], &i__6);
}
} else {
/* If pivot is zero, set INFO to the index of the pivot */
/* unless a zero pivot has already been found. */
if (*info == 0) {
*info = jj;
}
}
/* Copy current column of A31 into the work array WORK31 */
/* Computing MIN */
i__4 = jj - j + 1;
nw = f2cmin(i__4,i3);
if (nw > 0) {
scopy_(&nw, &ab[kv + *kl + 1 - jj + j + jj * ab_dim1], &
c__1, &work31[(jj - j + 1) * 65 - 65], &c__1);
}
/* L80: */
}
if (j + jb <= *n) {
/* Apply the row interchanges to the other blocks. */
/* Computing MIN */
i__3 = ju - j + 1;
j2 = f2cmin(i__3,kv) - jb;
/* Computing MAX */
i__3 = 0, i__4 = ju - j - kv + 1;
j3 = f2cmax(i__3,i__4);
/* Use SLASWP to apply the row interchanges to A12, A22, and */
/* A32. */
i__3 = *ldab - 1;
slaswp_(&j2, &ab[kv + 1 - jb + (j + jb) * ab_dim1], &i__3, &
c__1, &jb, &ipiv[j], &c__1);
/* Adjust the pivot indices. */
i__3 = j + jb - 1;
for (i__ = j; i__ <= i__3; ++i__) {
ipiv[i__] = ipiv[i__] + j - 1;
/* L90: */
}
/* Apply the row interchanges to A13, A23, and A33 */
/* columnwise. */
k2 = j - 1 + jb + j2;
i__3 = j3;
for (i__ = 1; i__ <= i__3; ++i__) {
jj = k2 + i__;
i__4 = j + jb - 1;
for (ii = j + i__ - 1; ii <= i__4; ++ii) {
ip = ipiv[ii];
if (ip != ii) {
temp = ab[kv + 1 + ii - jj + jj * ab_dim1];
ab[kv + 1 + ii - jj + jj * ab_dim1] = ab[kv + 1 +
ip - jj + jj * ab_dim1];
ab[kv + 1 + ip - jj + jj * ab_dim1] = temp;
}
/* L100: */
}
/* L110: */
}
/* Update the relevant part of the trailing submatrix */
if (j2 > 0) {
/* Update A12 */
i__3 = *ldab - 1;
i__4 = *ldab - 1;
strsm_("Left", "Lower", "No transpose", "Unit", &jb, &j2,
&c_b31, &ab[kv + 1 + j * ab_dim1], &i__3, &ab[kv
+ 1 - jb + (j + jb) * ab_dim1], &i__4);
if (i2 > 0) {
/* Update A22 */
i__3 = *ldab - 1;
i__4 = *ldab - 1;
i__5 = *ldab - 1;
sgemm_("No transpose", "No transpose", &i2, &j2, &jb,
&c_b18, &ab[kv + 1 + jb + j * ab_dim1], &i__3,
&ab[kv + 1 - jb + (j + jb) * ab_dim1], &i__4,
&c_b31, &ab[kv + 1 + (j + jb) * ab_dim1], &
i__5);
}
if (i3 > 0) {
/* Update A32 */
i__3 = *ldab - 1;
i__4 = *ldab - 1;
sgemm_("No transpose", "No transpose", &i3, &j2, &jb,
&c_b18, work31, &c__65, &ab[kv + 1 - jb + (j
+ jb) * ab_dim1], &i__3, &c_b31, &ab[kv + *kl
+ 1 - jb + (j + jb) * ab_dim1], &i__4);
}
}
if (j3 > 0) {
/* Copy the lower triangle of A13 into the work array */
/* WORK13 */
i__3 = j3;
for (jj = 1; jj <= i__3; ++jj) {
i__4 = jb;
for (ii = jj; ii <= i__4; ++ii) {
work13[ii + jj * 65 - 66] = ab[ii - jj + 1 + (jj
+ j + kv - 1) * ab_dim1];
/* L120: */
}
/* L130: */
}
/* Update A13 in the work array */
i__3 = *ldab - 1;
strsm_("Left", "Lower", "No transpose", "Unit", &jb, &j3,
&c_b31, &ab[kv + 1 + j * ab_dim1], &i__3, work13,
&c__65);
if (i2 > 0) {
/* Update A23 */
i__3 = *ldab - 1;
i__4 = *ldab - 1;
sgemm_("No transpose", "No transpose", &i2, &j3, &jb,
&c_b18, &ab[kv + 1 + jb + j * ab_dim1], &i__3,
work13, &c__65, &c_b31, &ab[jb + 1 + (j + kv)
* ab_dim1], &i__4);
}
if (i3 > 0) {
/* Update A33 */
i__3 = *ldab - 1;
sgemm_("No transpose", "No transpose", &i3, &j3, &jb,
&c_b18, work31, &c__65, work13, &c__65, &
c_b31, &ab[*kl + 1 + (j + kv) * ab_dim1], &
i__3);
}
/* Copy the lower triangle of A13 back into place */
i__3 = j3;
for (jj = 1; jj <= i__3; ++jj) {
i__4 = jb;
for (ii = jj; ii <= i__4; ++ii) {
ab[ii - jj + 1 + (jj + j + kv - 1) * ab_dim1] =
work13[ii + jj * 65 - 66];
/* L140: */
}
/* L150: */
}
}
} else {
/* Adjust the pivot indices. */
i__3 = j + jb - 1;
for (i__ = j; i__ <= i__3; ++i__) {
ipiv[i__] = ipiv[i__] + j - 1;
/* L160: */
}
}
/* Partially undo the interchanges in the current block to */
/* restore the upper triangular form of A31 and copy the upper */
/* triangle of A31 back into place */
i__3 = j;
for (jj = j + jb - 1; jj >= i__3; --jj) {
jp = ipiv[jj] - jj + 1;
if (jp != 1) {
/* Apply interchange to columns J to JJ-1 */
if (jp + jj - 1 < j + *kl) {
/* The interchange does not affect A31 */
i__4 = jj - j;
i__5 = *ldab - 1;
i__6 = *ldab - 1;
sswap_(&i__4, &ab[kv + 1 + jj - j + j * ab_dim1], &
i__5, &ab[kv + jp + jj - j + j * ab_dim1], &
i__6);
} else {
/* The interchange does affect A31 */
i__4 = jj - j;
i__5 = *ldab - 1;
sswap_(&i__4, &ab[kv + 1 + jj - j + j * ab_dim1], &
i__5, &work31[jp + jj - j - *kl - 1], &c__65);
}
}
/* Copy the current column of A31 back into place */
/* Computing MIN */
i__4 = i3, i__5 = jj - j + 1;
nw = f2cmin(i__4,i__5);
if (nw > 0) {
scopy_(&nw, &work31[(jj - j + 1) * 65 - 65], &c__1, &ab[
kv + *kl + 1 - jj + j + jj * ab_dim1], &c__1);
}
/* L170: */
}
/* L180: */
}
}
return;
/* End of SGBTRF */
} /* sgbtrf_ */
Reference in New Issue View Git Blame Copy Permalink