OpenBLAS/lapack-netlib/SRC/sgebal.c

#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;

/* > \brief \b SGEBAL */

/*  =========== DOCUMENTATION =========== */

/* Online html documentation available at */
/*            http://www.netlib.org/lapack/explore-html/ */

/* > \htmlonly */
/* > Download SGEBAL + dependencies */
/* > <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/sgebal.
f"> */
/* > [TGZ]</a> */
/* > <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/sgebal.
f"> */
/* > [ZIP]</a> */
/* > <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/sgebal.
f"> */
/* > [TXT]</a> */
/* > \endhtmlonly */

/*  Definition: */
/*  =========== */

/*       SUBROUTINE SGEBAL( JOB, N, A, LDA, ILO, IHI, SCALE, INFO ) */

/*       CHARACTER          JOB */
/*       INTEGER            IHI, ILO, INFO, LDA, N */
/*       REAL               A( LDA, * ), SCALE( * ) */


/* > \par Purpose: */
/*  ============= */
/* > */
/* > \verbatim */
/* > */
/* > SGEBAL balances a general real matrix A.  This involves, first, */
/* > permuting A by a similarity transformation to isolate eigenvalues */
/* > in the first 1 to ILO-1 and last IHI+1 to N elements on the */
/* > diagonal; and second, applying a diagonal similarity transformation */
/* > to rows and columns ILO to IHI to make the rows and columns as */
/* > close in norm as possible.  Both steps are optional. */
/* > */
/* > Balancing may reduce the 1-norm of the matrix, and improve the */
/* > accuracy of the computed eigenvalues and/or eigenvectors. */
/* > \endverbatim */

/*  Arguments: */
/*  ========== */

/* > \param[in] JOB */
/* > \verbatim */
/* >          JOB is CHARACTER*1 */
/* >          Specifies the operations to be performed on A: */
/* >          = 'N':  none:  simply set ILO = 1, IHI = N, SCALE(I) = 1.0 */
/* >                  for i = 1,...,N; */
/* >          = 'P':  permute only; */
/* >          = 'S':  scale only; */
/* >          = 'B':  both permute and scale. */
/* > \endverbatim */
/* > */
/* > \param[in] N */
/* > \verbatim */
/* >          N is INTEGER */
/* >          The order of the matrix A.  N >= 0. */
/* > \endverbatim */
/* > */
/* > \param[in,out] A */
/* > \verbatim */
/* >          A is REAL array, dimension (LDA,N) */
/* >          On entry, the input matrix A. */
/* >          On exit,  A is overwritten by the balanced matrix. */
/* >          If JOB = 'N', A is not referenced. */
/* >          See Further Details. */
/* > \endverbatim */
/* > */
/* > \param[in] LDA */
/* > \verbatim */
/* >          LDA is INTEGER */
/* >          The leading dimension of the array A.  LDA >= f2cmax(1,N). */
/* > \endverbatim */
/* > */
/* > \param[out] ILO */
/* > \verbatim */
/* >          ILO is INTEGER */
/* > \endverbatim */
/* > \param[out] IHI */
/* > \verbatim */
/* >          IHI is INTEGER */
/* >          ILO and IHI are set to integers such that on exit */
/* >          A(i,j) = 0 if i > j and j = 1,...,ILO-1 or I = IHI+1,...,N. */
/* >          If JOB = 'N' or 'S', ILO = 1 and IHI = N. */
/* > \endverbatim */
/* > */
/* > \param[out] SCALE */
/* > \verbatim */
/* >          SCALE is REAL array, dimension (N) */
/* >          Details of the permutations and scaling factors applied to */
/* >          A.  If P(j) is the index of the row and column interchanged */
/* >          with row and column j and D(j) is the scaling factor */
/* >          applied to row and column j, then */
/* >          SCALE(j) = P(j)    for j = 1,...,ILO-1 */
/* >                   = D(j)    for j = ILO,...,IHI */
/* >                   = P(j)    for j = IHI+1,...,N. */
/* >          The order in which the interchanges are made is N to IHI+1, */
/* >          then 1 to ILO-1. */
/* > \endverbatim */
/* > */
/* > \param[out] INFO */
/* > \verbatim */
/* >          INFO is INTEGER */
/* >          = 0:  successful exit. */
/* >          < 0:  if INFO = -i, the i-th argument had an illegal value. */
/* > \endverbatim */

/*  Authors: */
/*  ======== */

/* > \author Univ. of Tennessee */
/* > \author Univ. of California Berkeley */
/* > \author Univ. of Colorado Denver */
/* > \author NAG Ltd. */

/* > \date December 2016 */

/* > \ingroup realGEcomputational */

/* > \par Further Details: */
/*  ===================== */
/* > */
/* > \verbatim */
/* > */
/* >  The permutations consist of row and column interchanges which put */
/* >  the matrix in the form */
/* > */
/* >             ( T1   X   Y  ) */
/* >     P A P = (  0   B   Z  ) */
/* >             (  0   0   T2 ) */
/* > */
/* >  where T1 and T2 are upper triangular matrices whose eigenvalues lie */
/* >  along the diagonal.  The column indices ILO and IHI mark the starting */
/* >  and ending columns of the submatrix B. Balancing consists of applying */
/* >  a diagonal similarity transformation inv(D) * B * D to make the */
/* >  1-norms of each row of B and its corresponding column nearly equal. */
/* >  The output matrix is */
/* > */
/* >     ( T1     X*D          Y    ) */
/* >     (  0  inv(D)*B*D  inv(D)*Z ). */
/* >     (  0      0           T2   ) */
/* > */
/* >  Information about the permutations P and the diagonal matrix D is */
/* >  returned in the vector SCALE. */
/* > */
/* >  This subroutine is based on the EISPACK routine BALANC. */
/* > */
/* >  Modified by Tzu-Yi Chen, Computer Science Division, University of */
/* >    California at Berkeley, USA */
/* > \endverbatim */
/* > */
/*  ===================================================================== */
/* Subroutine */ void sgebal_(char *job, integer *n, real *a, integer *lda, 
	integer *ilo, integer *ihi, real *scale, integer *info)
{
    /* System generated locals */
    integer a_dim1, a_offset, i__1, i__2;
    real r__1, r__2;

    /* Local variables */
    integer iexc;
    extern real snrm2_(integer *, real *, integer *);
    real c__, f, g;
    integer i__, j, k, l, m;
    real r__, s;
    extern logical lsame_(char *, char *);
    extern /* Subroutine */ void sscal_(integer *, real *, real *, integer *), 
	    sswap_(integer *, real *, integer *, real *, integer *);
    real sfmin1, sfmin2, sfmax1, sfmax2, ca, ra;
    extern real slamch_(char *);
    extern /* Subroutine */ int xerbla_(char *, integer *,ftnlen);
    extern integer isamax_(integer *, real *, integer *);
    extern logical sisnan_(real *);
    logical noconv;
    integer ica, ira;


/*  -- 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 */


/*  ===================================================================== */


/*     Test the input parameters */

    /* Parameter adjustments */
    a_dim1 = *lda;
    a_offset = 1 + a_dim1 * 1;
    a -= a_offset;
    --scale;

    /* Function Body */
    *info = 0;
    if (! lsame_(job, "N") && ! lsame_(job, "P") && ! lsame_(job, "S") 
	    && ! lsame_(job, "B")) {
	*info = -1;
    } else if (*n < 0) {
	*info = -2;
    } else if (*lda < f2cmax(1,*n)) {
	*info = -4;
    }
    if (*info != 0) {
	i__1 = -(*info);
	xerbla_("SGEBAL", &i__1,(ftnlen)6);
	return;
    }

    k = 1;
    l = *n;

    if (*n == 0) {
	goto L210;
    }

    if (lsame_(job, "N")) {
	i__1 = *n;
	for (i__ = 1; i__ <= i__1; ++i__) {
	    scale[i__] = 1.f;
/* L10: */
	}
	goto L210;
    }

    if (lsame_(job, "S")) {
	goto L120;
    }

/*     Permutation to isolate eigenvalues if possible */

    goto L50;

/*     Row and column exchange. */

L20:
    scale[m] = (real) j;
    if (j == m) {
	goto L30;
    }

    sswap_(&l, &a[j * a_dim1 + 1], &c__1, &a[m * a_dim1 + 1], &c__1);
    i__1 = *n - k + 1;
    sswap_(&i__1, &a[j + k * a_dim1], lda, &a[m + k * a_dim1], lda);

L30:
    switch (iexc) {
	case 1:  goto L40;
	case 2:  goto L80;
    }

/*     Search for rows isolating an eigenvalue and push them down. */

L40:
    if (l == 1) {
	goto L210;
    }
    --l;

L50:
    for (j = l; j >= 1; --j) {

	i__1 = l;
	for (i__ = 1; i__ <= i__1; ++i__) {
	    if (i__ == j) {
		goto L60;
	    }
	    if (a[j + i__ * a_dim1] != 0.f) {
		goto L70;
	    }
L60:
	    ;
	}

	m = l;
	iexc = 1;
	goto L20;
L70:
	;
    }

    goto L90;

/*     Search for columns isolating an eigenvalue and push them left. */

L80:
    ++k;

L90:
    i__1 = l;
    for (j = k; j <= i__1; ++j) {

	i__2 = l;
	for (i__ = k; i__ <= i__2; ++i__) {
	    if (i__ == j) {
		goto L100;
	    }
	    if (a[i__ + j * a_dim1] != 0.f) {
		goto L110;
	    }
L100:
	    ;
	}

	m = k;
	iexc = 2;
	goto L20;
L110:
	;
    }

L120:
    i__1 = l;
    for (i__ = k; i__ <= i__1; ++i__) {
	scale[i__] = 1.f;
/* L130: */
    }

    if (lsame_(job, "P")) {
	goto L210;
    }

/*     Balance the submatrix in rows K to L. */

/*     Iterative loop for norm reduction */

    sfmin1 = slamch_("S") / slamch_("P");
    sfmax1 = 1.f / sfmin1;
    sfmin2 = sfmin1 * 2.f;
    sfmax2 = 1.f / sfmin2;
L140:
    noconv = FALSE_;

    i__1 = l;
    for (i__ = k; i__ <= i__1; ++i__) {

	i__2 = l - k + 1;
	c__ = snrm2_(&i__2, &a[k + i__ * a_dim1], &c__1);
	i__2 = l - k + 1;
	r__ = snrm2_(&i__2, &a[i__ + k * a_dim1], lda);
	ica = isamax_(&l, &a[i__ * a_dim1 + 1], &c__1);
	ca = (r__1 = a[ica + i__ * a_dim1], abs(r__1));
	i__2 = *n - k + 1;
	ira = isamax_(&i__2, &a[i__ + k * a_dim1], lda);
	ra = (r__1 = a[i__ + (ira + k - 1) * a_dim1], abs(r__1));

/*        Guard against zero C or R due to underflow. */

	if (c__ == 0.f || r__ == 0.f) {
	    goto L200;
	}
	g = r__ / 2.f;
	f = 1.f;
	s = c__ + r__;
L160:
/* Computing MAX */
	r__1 = f2cmax(f,c__);
/* Computing MIN */
	r__2 = f2cmin(r__,g);
	if (c__ >= g || f2cmax(r__1,ca) >= sfmax2 || f2cmin(r__2,ra) <= sfmin2) {
	    goto L170;
	}
	f *= 2.f;
	c__ *= 2.f;
	ca *= 2.f;
	r__ /= 2.f;
	g /= 2.f;
	ra /= 2.f;
	goto L160;

L170:
	g = c__ / 2.f;
L180:
/* Computing MIN */
	r__1 = f2cmin(f,c__), r__1 = f2cmin(r__1,g);
	if (g < r__ || f2cmax(r__,ra) >= sfmax2 || f2cmin(r__1,ca) <= sfmin2) {
	    goto L190;
	}
	r__1 = c__ + f + ca + r__ + g + ra;
	if (sisnan_(&r__1)) {

/*           Exit if NaN to avoid infinite loop */

	    *info = -3;
	    i__2 = -(*info);
	    xerbla_("SGEBAL", &i__2, (ftnlen)6);
	    return;
	}
	f /= 2.f;
	c__ /= 2.f;
	g /= 2.f;
	ca /= 2.f;
	r__ *= 2.f;
	ra *= 2.f;
	goto L180;

/*        Now balance. */

L190:
	if (c__ + r__ >= s * .95f) {
	    goto L200;
	}
	if (f < 1.f && scale[i__] < 1.f) {
	    if (f * scale[i__] <= sfmin1) {
		goto L200;
	    }
	}
	if (f > 1.f && scale[i__] > 1.f) {
	    if (scale[i__] >= sfmax1 / f) {
		goto L200;
	    }
	}
	g = 1.f / f;
	scale[i__] *= f;
	noconv = TRUE_;

	i__2 = *n - k + 1;
	sscal_(&i__2, &g, &a[i__ + k * a_dim1], lda);
	sscal_(&l, &f, &a[i__ * a_dim1 + 1], &c__1);

L200:
	;
    }

    if (noconv) {
	goto L140;
    }

L210:
    *ilo = k;
    *ihi = l;

    return;

/*     End of SGEBAL */

} /* sgebal_ */