OpenBLAS/lapack-netlib/SRC/sggsvp3.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 int logical;
typedef short int shortlogical;
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 pow_dd(ap, bp) ( pow(*(ap), *(bp)))
#define pow_si(B,E) spow_ui(*(B),*(E))
#define pow_ri(B,E) spow_ui(*(B),*(E))
#define pow_di(B,E) dpow_ui(*(B),*(E))
#define pow_zi(p, a, b) {pCd(p) = zpow_ui(Cd(a), *(b));}
#define pow_ci(p, a, b) {pCf(p) = cpow_ui(Cf(a), *(b));}
#define pow_zz(R,A,B) {pCd(R) = cpow(Cd(A),*(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);}
static char junk[] = "\n@(#)LIBF77 VERSION 19990503\n";
#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)}

/* procedure parameter types for -A and -C++ */

#define F2C_proc_par_types 1
#ifdef __cplusplus
typedef logical (*L_fp)(...);
#else
typedef logical (*L_fp)();
#endif

static float spow_ui(float x, integer n) {
	float pow=1.0; unsigned long int u;
	if(n != 0) {
		if(n < 0) n = -n, x = 1/x;
		for(u = n; ; ) {
			if(u & 01) pow *= x;
			if(u >>= 1) x *= x;
			else break;
		}
	}
	return pow;
}
static double dpow_ui(double x, integer n) {
	double pow=1.0; unsigned long int u;
	if(n != 0) {
		if(n < 0) n = -n, x = 1/x;
		for(u = n; ; ) {
			if(u & 01) pow *= x;
			if(u >>= 1) x *= x;
			else break;
		}
	}
	return pow;
}
#ifdef _MSC_VER
static _Fcomplex cpow_ui(complex x, integer n) {
	complex pow={1.0,0.0}; unsigned long int u;
		if(n != 0) {
		if(n < 0) n = -n, x.r = 1/x.r, x.i=1/x.i;
		for(u = n; ; ) {
			if(u & 01) pow.r *= x.r, pow.i *= x.i;
			if(u >>= 1) x.r *= x.r, x.i *= x.i;
			else break;
		}
	}
	_Fcomplex p={pow.r, pow.i};
	return p;
}
#else
static _Complex float cpow_ui(_Complex float x, integer n) {
	_Complex float pow=1.0; unsigned long int u;
	if(n != 0) {
		if(n < 0) n = -n, x = 1/x;
		for(u = n; ; ) {
			if(u & 01) pow *= x;
			if(u >>= 1) x *= x;
			else break;
		}
	}
	return pow;
}
#endif
#ifdef _MSC_VER
static _Dcomplex zpow_ui(_Dcomplex x, integer n) {
	_Dcomplex pow={1.0,0.0}; unsigned long int u;
	if(n != 0) {
		if(n < 0) n = -n, x._Val[0] = 1/x._Val[0], x._Val[1] =1/x._Val[1];
		for(u = n; ; ) {
			if(u & 01) pow._Val[0] *= x._Val[0], pow._Val[1] *= x._Val[1];
			if(u >>= 1) x._Val[0] *= x._Val[0], x._Val[1] *= x._Val[1];
			else break;
		}
	}
	_Dcomplex p = {pow._Val[0], pow._Val[1]};
	return p;
}
#else
static _Complex double zpow_ui(_Complex double x, integer n) {
	_Complex double pow=1.0; unsigned long int u;
	if(n != 0) {
		if(n < 0) n = -n, x = 1/x;
		for(u = n; ; ) {
			if(u & 01) pow *= x;
			if(u >>= 1) x *= x;
			else break;
		}
	}
	return pow;
}
#endif
static integer pow_ii(integer x, integer n) {
	integer pow; unsigned long int u;
	if (n <= 0) {
		if (n == 0 || x == 1) pow = 1;
		else if (x != -1) pow = x == 0 ? 1/x : 0;
		else n = -n;
	}
	if ((n > 0) || !(n == 0 || x == 1 || x != -1)) {
		u = n;
		for(pow = 1; ; ) {
			if(u & 01) pow *= x;
			if(u >>= 1) x *= x;
			else break;
		}
	}
	return pow;
}
static integer dmaxloc_(double *w, integer s, integer e, integer *n)
{
	double m; integer i, mi;
	for(m=w[s-1], mi=s, i=s+1; i<=e; i++)
		if (w[i-1]>m) mi=i ,m=w[i-1];
	return mi-s+1;
}
static integer smaxloc_(float *w, integer s, integer e, integer *n)
{
	float m; integer i, mi;
	for(m=w[s-1], mi=s, i=s+1; i<=e; i++)
		if (w[i-1]>m) mi=i ,m=w[i-1];
	return mi-s+1;
}
static inline void cdotc_(complex *z, integer *n_, complex *x, integer *incx_, complex *y, integer *incy_) {
	integer n = *n_, incx = *incx_, incy = *incy_, i;
#ifdef _MSC_VER
	_Fcomplex zdotc = {0.0, 0.0};
	if (incx == 1 && incy == 1) {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc._Val[0] += conjf(Cf(&x[i]))._Val[0] * Cf(&y[i])._Val[0];
			zdotc._Val[1] += conjf(Cf(&x[i]))._Val[1] * Cf(&y[i])._Val[1];
		}
	} else {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc._Val[0] += conjf(Cf(&x[i*incx]))._Val[0] * Cf(&y[i*incy])._Val[0];
			zdotc._Val[1] += conjf(Cf(&x[i*incx]))._Val[1] * Cf(&y[i*incy])._Val[1];
		}
	}
	pCf(z) = zdotc;
}
#else
	_Complex float zdotc = 0.0;
	if (incx == 1 && incy == 1) {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc += conjf(Cf(&x[i])) * Cf(&y[i]);
		}
	} else {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc += conjf(Cf(&x[i*incx])) * Cf(&y[i*incy]);
		}
	}
	pCf(z) = zdotc;
}
#endif
static inline void zdotc_(doublecomplex *z, integer *n_, doublecomplex *x, integer *incx_, doublecomplex *y, integer *incy_) {
	integer n = *n_, incx = *incx_, incy = *incy_, i;
#ifdef _MSC_VER
	_Dcomplex zdotc = {0.0, 0.0};
	if (incx == 1 && incy == 1) {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc._Val[0] += conj(Cd(&x[i]))._Val[0] * Cd(&y[i])._Val[0];
			zdotc._Val[1] += conj(Cd(&x[i]))._Val[1] * Cd(&y[i])._Val[1];
		}
	} else {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc._Val[0] += conj(Cd(&x[i*incx]))._Val[0] * Cd(&y[i*incy])._Val[0];
			zdotc._Val[1] += conj(Cd(&x[i*incx]))._Val[1] * Cd(&y[i*incy])._Val[1];
		}
	}
	pCd(z) = zdotc;
}
#else
	_Complex double zdotc = 0.0;
	if (incx == 1 && incy == 1) {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc += conj(Cd(&x[i])) * Cd(&y[i]);
		}
	} else {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc += conj(Cd(&x[i*incx])) * Cd(&y[i*incy]);
		}
	}
	pCd(z) = zdotc;
}
#endif	
static inline void cdotu_(complex *z, integer *n_, complex *x, integer *incx_, complex *y, integer *incy_) {
	integer n = *n_, incx = *incx_, incy = *incy_, i;
#ifdef _MSC_VER
	_Fcomplex zdotc = {0.0, 0.0};
	if (incx == 1 && incy == 1) {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc._Val[0] += Cf(&x[i])._Val[0] * Cf(&y[i])._Val[0];
			zdotc._Val[1] += Cf(&x[i])._Val[1] * Cf(&y[i])._Val[1];
		}
	} else {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc._Val[0] += Cf(&x[i*incx])._Val[0] * Cf(&y[i*incy])._Val[0];
			zdotc._Val[1] += Cf(&x[i*incx])._Val[1] * Cf(&y[i*incy])._Val[1];
		}
	}
	pCf(z) = zdotc;
}
#else
	_Complex float zdotc = 0.0;
	if (incx == 1 && incy == 1) {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc += Cf(&x[i]) * Cf(&y[i]);
		}
	} else {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc += Cf(&x[i*incx]) * Cf(&y[i*incy]);
		}
	}
	pCf(z) = zdotc;
}
#endif
static inline void zdotu_(doublecomplex *z, integer *n_, doublecomplex *x, integer *incx_, doublecomplex *y, integer *incy_) {
	integer n = *n_, incx = *incx_, incy = *incy_, i;
#ifdef _MSC_VER
	_Dcomplex zdotc = {0.0, 0.0};
	if (incx == 1 && incy == 1) {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc._Val[0] += Cd(&x[i])._Val[0] * Cd(&y[i])._Val[0];
			zdotc._Val[1] += Cd(&x[i])._Val[1] * Cd(&y[i])._Val[1];
		}
	} else {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc._Val[0] += Cd(&x[i*incx])._Val[0] * Cd(&y[i*incy])._Val[0];
			zdotc._Val[1] += Cd(&x[i*incx])._Val[1] * Cd(&y[i*incy])._Val[1];
		}
	}
	pCd(z) = zdotc;
}
#else
	_Complex double zdotc = 0.0;
	if (incx == 1 && incy == 1) {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc += Cd(&x[i]) * Cd(&y[i]);
		}
	} else {
		for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
			zdotc += Cd(&x[i*incx]) * Cd(&y[i*incy]);
		}
	}
	pCd(z) = zdotc;
}
#endif
/*  -- 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_n1 = -1;
static real c_b14 = 0.f;
static real c_b24 = 1.f;

/* > \brief \b SGGSVP3 */

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

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

/* > \htmlonly */
/* > Download SGGSVP3 + dependencies */
/* > <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/sggsvp3
.f"> */
/* > [TGZ]</a> */
/* > <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/sggsvp3
.f"> */
/* > [ZIP]</a> */
/* > <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/sggsvp3
.f"> */
/* > [TXT]</a> */
/* > \endhtmlonly */

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

/*       SUBROUTINE SGGSVP3( JOBU, JOBV, JOBQ, M, P, N, A, LDA, B, LDB, */
/*                           TOLA, TOLB, K, L, U, LDU, V, LDV, Q, LDQ, */
/*                           IWORK, TAU, WORK, LWORK, INFO ) */

/*       CHARACTER          JOBQ, JOBU, JOBV */
/*       INTEGER            INFO, K, L, LDA, LDB, LDQ, LDU, LDV, M, N, P, LWORK */
/*       REAL               TOLA, TOLB */
/*       INTEGER            IWORK( * ) */
/*       REAL               A( LDA, * ), B( LDB, * ), Q( LDQ, * ), */
/*      $                   TAU( * ), U( LDU, * ), V( LDV, * ), WORK( * ) */


/* > \par Purpose: */
/*  ============= */
/* > */
/* > \verbatim */
/* > */
/* > SGGSVP3 computes orthogonal matrices U, V and Q such that */
/* > */
/* >                    N-K-L  K    L */
/* >  U**T*A*Q =     K ( 0    A12  A13 )  if M-K-L >= 0; */
/* >                 L ( 0     0   A23 ) */
/* >             M-K-L ( 0     0    0  ) */
/* > */
/* >                  N-K-L  K    L */
/* >         =     K ( 0    A12  A13 )  if M-K-L < 0; */
/* >             M-K ( 0     0   A23 ) */
/* > */
/* >                  N-K-L  K    L */
/* >  V**T*B*Q =   L ( 0     0   B13 ) */
/* >             P-L ( 0     0    0  ) */
/* > */
/* > where the K-by-K matrix A12 and L-by-L matrix B13 are nonsingular */
/* > upper triangular; A23 is L-by-L upper triangular if M-K-L >= 0, */
/* > otherwise A23 is (M-K)-by-L upper trapezoidal.  K+L = the effective */
/* > numerical rank of the (M+P)-by-N matrix (A**T,B**T)**T. */
/* > */
/* > This decomposition is the preprocessing step for computing the */
/* > Generalized Singular Value Decomposition (GSVD), see subroutine */
/* > SGGSVD3. */
/* > \endverbatim */

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

/* > \param[in] JOBU */
/* > \verbatim */
/* >          JOBU is CHARACTER*1 */
/* >          = 'U':  Orthogonal matrix U is computed; */
/* >          = 'N':  U is not computed. */
/* > \endverbatim */
/* > */
/* > \param[in] JOBV */
/* > \verbatim */
/* >          JOBV is CHARACTER*1 */
/* >          = 'V':  Orthogonal matrix V is computed; */
/* >          = 'N':  V is not computed. */
/* > \endverbatim */
/* > */
/* > \param[in] JOBQ */
/* > \verbatim */
/* >          JOBQ is CHARACTER*1 */
/* >          = 'Q':  Orthogonal matrix Q is computed; */
/* >          = 'N':  Q is not computed. */
/* > \endverbatim */
/* > */
/* > \param[in] M */
/* > \verbatim */
/* >          M is INTEGER */
/* >          The number of rows of the matrix A.  M >= 0. */
/* > \endverbatim */
/* > */
/* > \param[in] P */
/* > \verbatim */
/* >          P is INTEGER */
/* >          The number of rows of the matrix B.  P >= 0. */
/* > \endverbatim */
/* > */
/* > \param[in] N */
/* > \verbatim */
/* >          N is INTEGER */
/* >          The number of columns of the matrices A and B.  N >= 0. */
/* > \endverbatim */
/* > */
/* > \param[in,out] A */
/* > \verbatim */
/* >          A is REAL array, dimension (LDA,N) */
/* >          On entry, the M-by-N matrix A. */
/* >          On exit, A contains the triangular (or trapezoidal) matrix */
/* >          described in the Purpose section. */
/* > \endverbatim */
/* > */
/* > \param[in] LDA */
/* > \verbatim */
/* >          LDA is INTEGER */
/* >          The leading dimension of the array A. LDA >= f2cmax(1,M). */
/* > \endverbatim */
/* > */
/* > \param[in,out] B */
/* > \verbatim */
/* >          B is REAL array, dimension (LDB,N) */
/* >          On entry, the P-by-N matrix B. */
/* >          On exit, B contains the triangular matrix described in */
/* >          the Purpose section. */
/* > \endverbatim */
/* > */
/* > \param[in] LDB */
/* > \verbatim */
/* >          LDB is INTEGER */
/* >          The leading dimension of the array B. LDB >= f2cmax(1,P). */
/* > \endverbatim */
/* > */
/* > \param[in] TOLA */
/* > \verbatim */
/* >          TOLA is REAL */
/* > \endverbatim */
/* > */
/* > \param[in] TOLB */
/* > \verbatim */
/* >          TOLB is REAL */
/* > */
/* >          TOLA and TOLB are the thresholds to determine the effective */
/* >          numerical rank of matrix B and a subblock of A. Generally, */
/* >          they are set to */
/* >             TOLA = MAX(M,N)*norm(A)*MACHEPS, */
/* >             TOLB = MAX(P,N)*norm(B)*MACHEPS. */
/* >          The size of TOLA and TOLB may affect the size of backward */
/* >          errors of the decomposition. */
/* > \endverbatim */
/* > */
/* > \param[out] K */
/* > \verbatim */
/* >          K is INTEGER */
/* > \endverbatim */
/* > */
/* > \param[out] L */
/* > \verbatim */
/* >          L is INTEGER */
/* > */
/* >          On exit, K and L specify the dimension of the subblocks */
/* >          described in Purpose section. */
/* >          K + L = effective numerical rank of (A**T,B**T)**T. */
/* > \endverbatim */
/* > */
/* > \param[out] U */
/* > \verbatim */
/* >          U is REAL array, dimension (LDU,M) */
/* >          If JOBU = 'U', U contains the orthogonal matrix U. */
/* >          If JOBU = 'N', U is not referenced. */
/* > \endverbatim */
/* > */
/* > \param[in] LDU */
/* > \verbatim */
/* >          LDU is INTEGER */
/* >          The leading dimension of the array U. LDU >= f2cmax(1,M) if */
/* >          JOBU = 'U'; LDU >= 1 otherwise. */
/* > \endverbatim */
/* > */
/* > \param[out] V */
/* > \verbatim */
/* >          V is REAL array, dimension (LDV,P) */
/* >          If JOBV = 'V', V contains the orthogonal matrix V. */
/* >          If JOBV = 'N', V is not referenced. */
/* > \endverbatim */
/* > */
/* > \param[in] LDV */
/* > \verbatim */
/* >          LDV is INTEGER */
/* >          The leading dimension of the array V. LDV >= f2cmax(1,P) if */
/* >          JOBV = 'V'; LDV >= 1 otherwise. */
/* > \endverbatim */
/* > */
/* > \param[out] Q */
/* > \verbatim */
/* >          Q is REAL array, dimension (LDQ,N) */
/* >          If JOBQ = 'Q', Q contains the orthogonal matrix Q. */
/* >          If JOBQ = 'N', Q is not referenced. */
/* > \endverbatim */
/* > */
/* > \param[in] LDQ */
/* > \verbatim */
/* >          LDQ is INTEGER */
/* >          The leading dimension of the array Q. LDQ >= f2cmax(1,N) if */
/* >          JOBQ = 'Q'; LDQ >= 1 otherwise. */
/* > \endverbatim */
/* > */
/* > \param[out] IWORK */
/* > \verbatim */
/* >          IWORK is INTEGER array, dimension (N) */
/* > \endverbatim */
/* > */
/* > \param[out] TAU */
/* > \verbatim */
/* >          TAU is REAL array, dimension (N) */
/* > \endverbatim */
/* > */
/* > \param[out] WORK */
/* > \verbatim */
/* >          WORK is REAL array, dimension (MAX(1,LWORK)) */
/* >          On exit, if INFO = 0, WORK(1) returns the optimal LWORK. */
/* > \endverbatim */
/* > */
/* > \param[in] LWORK */
/* > \verbatim */
/* >          LWORK is INTEGER */
/* >          The dimension of the array WORK. */
/* > */
/* >          If LWORK = -1, then a workspace query is assumed; the routine */
/* >          only calculates the optimal size of the WORK array, returns */
/* >          this value as the first entry of the WORK array, and no error */
/* >          message related to LWORK is issued by XERBLA. */
/* > \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 August 2015 */

/* > \ingroup realOTHERcomputational */

/* > \par Further Details: */
/*  ===================== */
/* > */
/* > \verbatim */
/* > */
/* >  The subroutine uses LAPACK subroutine SGEQP3 for the QR factorization */
/* >  with column pivoting to detect the effective numerical rank of the */
/* >  a matrix. It may be replaced by a better rank determination strategy. */
/* > */
/* >  SGGSVP3 replaces the deprecated subroutine SGGSVP. */
/* > */
/* > \endverbatim */
/* > */
/*  ===================================================================== */
/* Subroutine */ int sggsvp3_(char *jobu, char *jobv, char *jobq, integer *m, 
	integer *p, integer *n, real *a, integer *lda, real *b, integer *ldb, 
	real *tola, real *tolb, integer *k, integer *l, real *u, integer *ldu,
	 real *v, integer *ldv, real *q, integer *ldq, integer *iwork, real *
	tau, real *work, integer *lwork, integer *info)
{
    /* System generated locals */
    integer a_dim1, a_offset, b_dim1, b_offset, q_dim1, q_offset, u_dim1, 
	    u_offset, v_dim1, v_offset, i__1, i__2, i__3;
    real r__1;

    /* Local variables */
    integer i__, j;
    extern logical lsame_(char *, char *);
    logical wantq, wantu, wantv;
    extern /* Subroutine */ int sgeqp3_(integer *, integer *, real *, integer 
	    *, integer *, real *, real *, integer *, integer *), sgeqr2_(
	    integer *, integer *, real *, integer *, real *, real *, integer *
	    ), sgerq2_(integer *, integer *, real *, integer *, real *, real *
	    , integer *), sorg2r_(integer *, integer *, integer *, real *, 
	    integer *, real *, real *, integer *), sorm2r_(char *, char *, 
	    integer *, integer *, integer *, real *, integer *, real *, real *
	    , integer *, real *, integer *), sormr2_(char *, 
	    char *, integer *, integer *, integer *, real *, integer *, real *
	    , real *, integer *, real *, integer *), xerbla_(
	    char *, integer *, ftnlen), slacpy_(char *, integer *, integer *, 
	    real *, integer *, real *, integer *), slaset_(char *, 
	    integer *, integer *, real *, real *, real *, integer *), 
	    slapmt_(logical *, integer *, integer *, real *, integer *, 
	    integer *);
    logical forwrd;
    integer lwkopt;
    logical lquery;


/*  -- 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..-- */
/*     August 2015 */



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


/*     Test the input parameters */

    /* Parameter adjustments */
    a_dim1 = *lda;
    a_offset = 1 + a_dim1 * 1;
    a -= a_offset;
    b_dim1 = *ldb;
    b_offset = 1 + b_dim1 * 1;
    b -= b_offset;
    u_dim1 = *ldu;
    u_offset = 1 + u_dim1 * 1;
    u -= u_offset;
    v_dim1 = *ldv;
    v_offset = 1 + v_dim1 * 1;
    v -= v_offset;
    q_dim1 = *ldq;
    q_offset = 1 + q_dim1 * 1;
    q -= q_offset;
    --iwork;
    --tau;
    --work;

    /* Function Body */
    wantu = lsame_(jobu, "U");
    wantv = lsame_(jobv, "V");
    wantq = lsame_(jobq, "Q");
    forwrd = TRUE_;
    lquery = *lwork == -1;
    lwkopt = 1;

/*     Test the input arguments */

    *info = 0;
    if (! (wantu || lsame_(jobu, "N"))) {
	*info = -1;
    } else if (! (wantv || lsame_(jobv, "N"))) {
	*info = -2;
    } else if (! (wantq || lsame_(jobq, "N"))) {
	*info = -3;
    } else if (*m < 0) {
	*info = -4;
    } else if (*p < 0) {
	*info = -5;
    } else if (*n < 0) {
	*info = -6;
    } else if (*lda < f2cmax(1,*m)) {
	*info = -8;
    } else if (*ldb < f2cmax(1,*p)) {
	*info = -10;
    } else if (*ldu < 1 || wantu && *ldu < *m) {
	*info = -16;
    } else if (*ldv < 1 || wantv && *ldv < *p) {
	*info = -18;
    } else if (*ldq < 1 || wantq && *ldq < *n) {
	*info = -20;
    } else if (*lwork < 1 && ! lquery) {
	*info = -24;
    }

/*     Compute workspace */

    if (*info == 0) {
	sgeqp3_(p, n, &b[b_offset], ldb, &iwork[1], &tau[1], &work[1], &c_n1, 
		info);
	lwkopt = (integer) work[1];
	if (wantv) {
	    lwkopt = f2cmax(lwkopt,*p);
	}
/* Computing MAX */
	i__1 = lwkopt, i__2 = f2cmin(*n,*p);
	lwkopt = f2cmax(i__1,i__2);
	lwkopt = f2cmax(lwkopt,*m);
	if (wantq) {
	    lwkopt = f2cmax(lwkopt,*n);
	}
	sgeqp3_(m, n, &a[a_offset], lda, &iwork[1], &tau[1], &work[1], &c_n1, 
		info);
/* Computing MAX */
	i__1 = lwkopt, i__2 = (integer) work[1];
	lwkopt = f2cmax(i__1,i__2);
	lwkopt = f2cmax(1,lwkopt);
	work[1] = (real) lwkopt;
    }

    if (*info != 0) {
	i__1 = -(*info);
	xerbla_("SGGSVP3", &i__1, (ftnlen)7);
	return 0;
    }
    if (lquery) {
	return 0;
    }

/*     QR with column pivoting of B: B*P = V*( S11 S12 ) */
/*                                           (  0   0  ) */

    i__1 = *n;
    for (i__ = 1; i__ <= i__1; ++i__) {
	iwork[i__] = 0;
/* L10: */
    }
    sgeqp3_(p, n, &b[b_offset], ldb, &iwork[1], &tau[1], &work[1], lwork, 
	    info);

/*     Update A := A*P */

    slapmt_(&forwrd, m, n, &a[a_offset], lda, &iwork[1]);

/*     Determine the effective rank of matrix B. */

    *l = 0;
    i__1 = f2cmin(*p,*n);
    for (i__ = 1; i__ <= i__1; ++i__) {
	if ((r__1 = b[i__ + i__ * b_dim1], abs(r__1)) > *tolb) {
	    ++(*l);
	}
/* L20: */
    }

    if (wantv) {

/*        Copy the details of V, and form V. */

	slaset_("Full", p, p, &c_b14, &c_b14, &v[v_offset], ldv);
	if (*p > 1) {
	    i__1 = *p - 1;
	    slacpy_("Lower", &i__1, n, &b[b_dim1 + 2], ldb, &v[v_dim1 + 2], 
		    ldv);
	}
	i__1 = f2cmin(*p,*n);
	sorg2r_(p, p, &i__1, &v[v_offset], ldv, &tau[1], &work[1], info);
    }

/*     Clean up B */

    i__1 = *l - 1;
    for (j = 1; j <= i__1; ++j) {
	i__2 = *l;
	for (i__ = j + 1; i__ <= i__2; ++i__) {
	    b[i__ + j * b_dim1] = 0.f;
/* L30: */
	}
/* L40: */
    }
    if (*p > *l) {
	i__1 = *p - *l;
	slaset_("Full", &i__1, n, &c_b14, &c_b14, &b[*l + 1 + b_dim1], ldb);
    }

    if (wantq) {

/*        Set Q = I and Update Q := Q*P */

	slaset_("Full", n, n, &c_b14, &c_b24, &q[q_offset], ldq);
	slapmt_(&forwrd, n, n, &q[q_offset], ldq, &iwork[1]);
    }

    if (*p >= *l && *n != *l) {

/*        RQ factorization of (S11 S12): ( S11 S12 ) = ( 0 S12 )*Z */

	sgerq2_(l, n, &b[b_offset], ldb, &tau[1], &work[1], info);

/*        Update A := A*Z**T */

	sormr2_("Right", "Transpose", m, n, l, &b[b_offset], ldb, &tau[1], &a[
		a_offset], lda, &work[1], info);

	if (wantq) {

/*           Update Q := Q*Z**T */

	    sormr2_("Right", "Transpose", n, n, l, &b[b_offset], ldb, &tau[1],
		     &q[q_offset], ldq, &work[1], info);
	}

/*        Clean up B */

	i__1 = *n - *l;
	slaset_("Full", l, &i__1, &c_b14, &c_b14, &b[b_offset], ldb);
	i__1 = *n;
	for (j = *n - *l + 1; j <= i__1; ++j) {
	    i__2 = *l;
	    for (i__ = j - *n + *l + 1; i__ <= i__2; ++i__) {
		b[i__ + j * b_dim1] = 0.f;
/* L50: */
	    }
/* L60: */
	}

    }

/*     Let              N-L     L */
/*                A = ( A11    A12 ) M, */

/*     then the following does the complete QR decomposition of A11: */

/*              A11 = U*(  0  T12 )*P1**T */
/*                      (  0   0  ) */

    i__1 = *n - *l;
    for (i__ = 1; i__ <= i__1; ++i__) {
	iwork[i__] = 0;
/* L70: */
    }
    i__1 = *n - *l;
    sgeqp3_(m, &i__1, &a[a_offset], lda, &iwork[1], &tau[1], &work[1], lwork, 
	    info);

/*     Determine the effective rank of A11 */

    *k = 0;
/* Computing MIN */
    i__2 = *m, i__3 = *n - *l;
    i__1 = f2cmin(i__2,i__3);
    for (i__ = 1; i__ <= i__1; ++i__) {
	if ((r__1 = a[i__ + i__ * a_dim1], abs(r__1)) > *tola) {
	    ++(*k);
	}
/* L80: */
    }

/*     Update A12 := U**T*A12, where A12 = A( 1:M, N-L+1:N ) */

/* Computing MIN */
    i__2 = *m, i__3 = *n - *l;
    i__1 = f2cmin(i__2,i__3);
    sorm2r_("Left", "Transpose", m, l, &i__1, &a[a_offset], lda, &tau[1], &a[(
	    *n - *l + 1) * a_dim1 + 1], lda, &work[1], info);

    if (wantu) {

/*        Copy the details of U, and form U */

	slaset_("Full", m, m, &c_b14, &c_b14, &u[u_offset], ldu);
	if (*m > 1) {
	    i__1 = *m - 1;
	    i__2 = *n - *l;
	    slacpy_("Lower", &i__1, &i__2, &a[a_dim1 + 2], lda, &u[u_dim1 + 2]
		    , ldu);
	}
/* Computing MIN */
	i__2 = *m, i__3 = *n - *l;
	i__1 = f2cmin(i__2,i__3);
	sorg2r_(m, m, &i__1, &u[u_offset], ldu, &tau[1], &work[1], info);
    }

    if (wantq) {

/*        Update Q( 1:N, 1:N-L )  = Q( 1:N, 1:N-L )*P1 */

	i__1 = *n - *l;
	slapmt_(&forwrd, n, &i__1, &q[q_offset], ldq, &iwork[1]);
    }

/*     Clean up A: set the strictly lower triangular part of */
/*     A(1:K, 1:K) = 0, and A( K+1:M, 1:N-L ) = 0. */

    i__1 = *k - 1;
    for (j = 1; j <= i__1; ++j) {
	i__2 = *k;
	for (i__ = j + 1; i__ <= i__2; ++i__) {
	    a[i__ + j * a_dim1] = 0.f;
/* L90: */
	}
/* L100: */
    }
    if (*m > *k) {
	i__1 = *m - *k;
	i__2 = *n - *l;
	slaset_("Full", &i__1, &i__2, &c_b14, &c_b14, &a[*k + 1 + a_dim1], 
		lda);
    }

    if (*n - *l > *k) {

/*        RQ factorization of ( T11 T12 ) = ( 0 T12 )*Z1 */

	i__1 = *n - *l;
	sgerq2_(k, &i__1, &a[a_offset], lda, &tau[1], &work[1], info);

	if (wantq) {

/*           Update Q( 1:N,1:N-L ) = Q( 1:N,1:N-L )*Z1**T */

	    i__1 = *n - *l;
	    sormr2_("Right", "Transpose", n, &i__1, k, &a[a_offset], lda, &
		    tau[1], &q[q_offset], ldq, &work[1], info);
	}

/*        Clean up A */

	i__1 = *n - *l - *k;
	slaset_("Full", k, &i__1, &c_b14, &c_b14, &a[a_offset], lda);
	i__1 = *n - *l;
	for (j = *n - *l - *k + 1; j <= i__1; ++j) {
	    i__2 = *k;
	    for (i__ = j - *n + *l + *k + 1; i__ <= i__2; ++i__) {
		a[i__ + j * a_dim1] = 0.f;
/* L110: */
	    }
/* L120: */
	}

    }

    if (*m > *k) {

/*        QR factorization of A( K+1:M,N-L+1:N ) */

	i__1 = *m - *k;
	sgeqr2_(&i__1, l, &a[*k + 1 + (*n - *l + 1) * a_dim1], lda, &tau[1], &
		work[1], info);

	if (wantu) {

/*           Update U(:,K+1:M) := U(:,K+1:M)*U1 */

	    i__1 = *m - *k;
/* Computing MIN */
	    i__3 = *m - *k;
	    i__2 = f2cmin(i__3,*l);
	    sorm2r_("Right", "No transpose", m, &i__1, &i__2, &a[*k + 1 + (*n 
		    - *l + 1) * a_dim1], lda, &tau[1], &u[(*k + 1) * u_dim1 + 
		    1], ldu, &work[1], info);
	}

/*        Clean up */

	i__1 = *n;
	for (j = *n - *l + 1; j <= i__1; ++j) {
	    i__2 = *m;
	    for (i__ = j - *n + *k + *l + 1; i__ <= i__2; ++i__) {
		a[i__ + j * a_dim1] = 0.f;
/* L130: */
	    }
/* L140: */
	}

    }

    work[1] = (real) lwkopt;
    return 0;

/*     End of SGGSVP3 */

} /* sggsvp3_ */