/* Ergo, version 3.8.2, a program for linear scaling electronic structure
 * calculations.
 * Copyright (C) 2023 Elias Rudberg, Emanuel H. Rubensson, Pawel Salek,
 * and Anastasia Kruchinina.
 * 
 * This program is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 * 
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 * 
 * You should have received a copy of the GNU General Public License
 * along with this program.  If not, see <http://www.gnu.org/licenses/>.
 * 
 * Primary academic reference:
 * Ergo: An open-source program for linear-scaling electronic structure
 * calculations,
 * Elias Rudberg, Emanuel H. Rubensson, Pawel Salek, and Anastasia
 * Kruchinina,
 * SoftwareX 7, 107 (2018),
 * <http://dx.doi.org/10.1016/j.softx.2018.03.005>
 * 
 * For further information about Ergo, see <http://www.ergoscf.org>.
 */
 
 /* This file belongs to the template_lapack part of the Ergo source 
  * code. The source files in the template_lapack directory are modified
  * versions of files originally distributed as CLAPACK, see the
  * Copyright/license notice in the file template_lapack/COPYING.
  */
 

#ifndef TEMPLATE_LAPACK_STEQR_HEADER
#define TEMPLATE_LAPACK_STEQR_HEADER


template<class Treal>
int template_lapack_steqr(const char *compz, const integer *n, Treal *d__, 
	Treal *e, Treal *z__, const integer *ldz, Treal *work, 
	integer *info)
{
/*  -- LAPACK routine (version 3.0) --   
       Univ. of Tennessee, Univ. of California Berkeley, NAG Ltd.,   
       Courant Institute, Argonne National Lab, and Rice University   
       September 30, 1994   


    Purpose   
    =======   

    DSTEQR computes all eigenvalues and, optionally, eigenvectors of a   
    symmetric tridiagonal matrix using the implicit QL or QR method.   
    The eigenvectors of a full or band symmetric matrix can also be found   
    if DSYTRD or DSPTRD or DSBTRD has been used to reduce this matrix to   
    tridiagonal form.   

    Arguments   
    =========   

    COMPZ   (input) CHARACTER*1   
            = 'N':  Compute eigenvalues only.   
            = 'V':  Compute eigenvalues and eigenvectors of the original   
                    symmetric matrix.  On entry, Z must contain the   
                    orthogonal matrix used to reduce the original matrix   
                    to tridiagonal form.   
            = 'I':  Compute eigenvalues and eigenvectors of the   
                    tridiagonal matrix.  Z is initialized to the identity   
                    matrix.   

    N       (input) INTEGER   
            The order of the matrix.  N >= 0.   

    D       (input/output) DOUBLE PRECISION array, dimension (N)   
            On entry, the diagonal elements of the tridiagonal matrix.   
            On exit, if INFO = 0, the eigenvalues in ascending order.   

    E       (input/output) DOUBLE PRECISION array, dimension (N-1)   
            On entry, the (n-1) subdiagonal elements of the tridiagonal   
            matrix.   
            On exit, E has been destroyed.   

    Z       (input/output) DOUBLE PRECISION array, dimension (LDZ, N)   
            On entry, if  COMPZ = 'V', then Z contains the orthogonal   
            matrix used in the reduction to tridiagonal form.   
            On exit, if INFO = 0, then if  COMPZ = 'V', Z contains the   
            orthonormal eigenvectors of the original symmetric matrix,   
            and if COMPZ = 'I', Z contains the orthonormal eigenvectors   
            of the symmetric tridiagonal matrix.   
            If COMPZ = 'N', then Z is not referenced.   

    LDZ     (input) INTEGER   
            The leading dimension of the array Z.  LDZ >= 1, and if   
            eigenvectors are desired, then  LDZ >= max(1,N).   

    WORK    (workspace) DOUBLE PRECISION array, dimension (max(1,2*N-2))   
            If COMPZ = 'N', then WORK is not referenced.   

    INFO    (output) INTEGER   
            = 0:  successful exit   
            < 0:  if INFO = -i, the i-th argument had an illegal value   
            > 0:  the algorithm has failed to find all the eigenvalues in   
                  a total of 30*N iterations; if INFO = i, then i   
                  elements of E have not converged to zero; on exit, D   
                  and E contain the elements of a symmetric tridiagonal   
                  matrix which is orthogonally similar to the original   
                  matrix.   

    =====================================================================   


       Test the input parameters.   

       Parameter adjustments */
    /* Table of constant values */
     Treal c_b9 = 0.;
     Treal c_b10 = 1.;
     integer c__0 = 0;
     integer c__1 = 1;
     integer c__2 = 2;
    
    /* System generated locals */
    integer z_dim1, z_offset, i__1, i__2;
    Treal d__1, d__2;
    /* Local variables */
     integer lend, jtot;
     Treal b, c__, f, g;
     integer i__, j, k, l, m;
     Treal p, r__, s;
     Treal anorm;
     integer l1;
     integer lendm1, lendp1;
     integer ii;
     integer mm, iscale;
     Treal safmin;
     Treal safmax;
     integer lendsv;
     Treal ssfmin;
     integer nmaxit, icompz;
     Treal ssfmax;
     integer lm1, mm1, nm1;
     Treal rt1, rt2, eps;
     integer lsv;
     Treal tst, eps2;
#define z___ref(a_1,a_2) z__[(a_2)*z_dim1 + a_1]


    --d__;
    --e;
    z_dim1 = *ldz;
    z_offset = 1 + z_dim1 * 1;
    z__ -= z_offset;
    --work;

    /* Function Body */
    *info = 0;

    if (template_blas_lsame(compz, "N")) {
	icompz = 0;
    } else if (template_blas_lsame(compz, "V")) {
	icompz = 1;
    } else if (template_blas_lsame(compz, "I")) {
	icompz = 2;
    } else {
	icompz = -1;
    }
    if (icompz < 0) {
	*info = -1;
    } else if (*n < 0) {
	*info = -2;
    } else if (*ldz < 1 || (icompz > 0 && *ldz < maxMACRO(1,*n) ) ) {
	*info = -6;
    }
    if (*info != 0) {
	i__1 = -(*info);
	template_blas_erbla("STEQR ", &i__1);
	return 0;
    }

/*     Quick return if possible */

    if (*n == 0) {
	return 0;
    }

    if (*n == 1) {
	if (icompz == 2) {
	    z___ref(1, 1) = 1.;
	}
	return 0;
    }

/*     Determine the unit roundoff and over/underflow thresholds. */

    eps = template_lapack_lamch("E", (Treal)0);
/* Computing 2nd power */
    d__1 = eps;
    eps2 = d__1 * d__1;
    safmin = template_lapack_lamch("S", (Treal)0);
    safmax = 1. / safmin;
    ssfmax = template_blas_sqrt(safmax) / 3.;
    ssfmin = template_blas_sqrt(safmin) / eps2;

/*     Compute the eigenvalues and eigenvectors of the tridiagonal   
       matrix. */

    if (icompz == 2) {
	template_lapack_laset("Full", n, n, &c_b9, &c_b10, &z__[z_offset], ldz);
    }

    nmaxit = *n * 30;
    jtot = 0;

/*     Determine where the matrix splits and choose QL or QR iteration   
       for each block, according to whether top or bottom diagonal   
       element is smaller. */

    l1 = 1;
    nm1 = *n - 1;

L10:
    if (l1 > *n) {
	goto L160;
    }
    if (l1 > 1) {
	e[l1 - 1] = 0.;
    }
    if (l1 <= nm1) {
	i__1 = nm1;
	for (m = l1; m <= i__1; ++m) {
	    tst = (d__1 = e[m], absMACRO(d__1));
	    if (tst == 0.) {
		goto L30;
	    }
	    if (tst <= template_blas_sqrt((d__1 = d__[m], absMACRO(d__1))) * template_blas_sqrt((d__2 = d__[m 
		    + 1], absMACRO(d__2))) * eps) {
		e[m] = 0.;
		goto L30;
	    }
/* L20: */
	}
    }
    m = *n;

L30:
    l = l1;
    lsv = l;
    lend = m;
    lendsv = lend;
    l1 = m + 1;
    if (lend == l) {
	goto L10;
    }

/*     Scale submatrix in rows and columns L to LEND */

    i__1 = lend - l + 1;
    anorm = template_lapack_lanst("I", &i__1, &d__[l], &e[l]);
    iscale = 0;
    if (anorm == 0.) {
	goto L10;
    }
    if (anorm > ssfmax) {
	iscale = 1;
	i__1 = lend - l + 1;
	template_lapack_lascl("G", &c__0, &c__0, &anorm, &ssfmax, &i__1, &c__1, &d__[l], n, 
		info);
	i__1 = lend - l;
	template_lapack_lascl("G", &c__0, &c__0, &anorm, &ssfmax, &i__1, &c__1, &e[l], n, 
		info);
    } else if (anorm < ssfmin) {
	iscale = 2;
	i__1 = lend - l + 1;
	template_lapack_lascl("G", &c__0, &c__0, &anorm, &ssfmin, &i__1, &c__1, &d__[l], n, 
		info);
	i__1 = lend - l;
	template_lapack_lascl("G", &c__0, &c__0, &anorm, &ssfmin, &i__1, &c__1, &e[l], n, 
		info);
    }

/*     Choose between QL and QR iteration */

    if ((d__1 = d__[lend], absMACRO(d__1)) < (d__2 = d__[l], absMACRO(d__2))) {
	lend = lsv;
	l = lendsv;
    }

    if (lend > l) {

/*        QL Iteration   

          Look for small subdiagonal element. */

L40:
	if (l != lend) {
	    lendm1 = lend - 1;
	    i__1 = lendm1;
	    for (m = l; m <= i__1; ++m) {
/* Computing 2nd power */
		d__2 = (d__1 = e[m], absMACRO(d__1));
		tst = d__2 * d__2;
		if (tst <= eps2 * (d__1 = d__[m], absMACRO(d__1)) * (d__2 = d__[m 
			+ 1], absMACRO(d__2)) + safmin) {
		    goto L60;
		}
/* L50: */
	    }
	}

	m = lend;

L60:
	if (m < lend) {
	    e[m] = 0.;
	}
	p = d__[l];
	if (m == l) {
	    goto L80;
	}

/*        If remaining matrix is 2-by-2, use DLAE2 or SLAEV2   
          to compute its eigensystem. */

	if (m == l + 1) {
	    if (icompz > 0) {
		template_lapack_laev2(&d__[l], &e[l], &d__[l + 1], &rt1, &rt2, &c__, &s);
		work[l] = c__;
		work[*n - 1 + l] = s;
		template_lapack_lasr("R", "V", "B", n, &c__2, &work[l], &work[*n - 1 + l], &
			z___ref(1, l), ldz);
	    } else {
		template_lapack_lae2(&d__[l], &e[l], &d__[l + 1], &rt1, &rt2);
	    }
	    d__[l] = rt1;
	    d__[l + 1] = rt2;
	    e[l] = 0.;
	    l += 2;
	    if (l <= lend) {
		goto L40;
	    }
	    goto L140;
	}

	if (jtot == nmaxit) {
	    goto L140;
	}
	++jtot;

/*        Form shift. */

	g = (d__[l + 1] - p) / (e[l] * 2.);
	r__ = template_lapack_lapy2(&g, &c_b10);
	g = d__[m] - p + e[l] / (g + template_lapack_d_sign(&r__, &g));

	s = 1.;
	c__ = 1.;
	p = 0.;

/*        Inner loop */

	mm1 = m - 1;
	i__1 = l;
	for (i__ = mm1; i__ >= i__1; --i__) {
	    f = s * e[i__];
	    b = c__ * e[i__];
	    template_lapack_lartg(&g, &f, &c__, &s, &r__);
	    if (i__ != m - 1) {
		e[i__ + 1] = r__;
	    }
	    g = d__[i__ + 1] - p;
	    r__ = (d__[i__] - g) * s + c__ * 2. * b;
	    p = s * r__;
	    d__[i__ + 1] = g + p;
	    g = c__ * r__ - b;

/*           If eigenvectors are desired, then save rotations. */

	    if (icompz > 0) {
		work[i__] = c__;
		work[*n - 1 + i__] = -s;
	    }

/* L70: */
	}

/*        If eigenvectors are desired, then apply saved rotations. */

	if (icompz > 0) {
	    mm = m - l + 1;
	    template_lapack_lasr("R", "V", "B", n, &mm, &work[l], &work[*n - 1 + l], &
		    z___ref(1, l), ldz);
	}

	d__[l] -= p;
	e[l] = g;
	goto L40;

/*        Eigenvalue found. */

L80:
	d__[l] = p;

	++l;
	if (l <= lend) {
	    goto L40;
	}
	goto L140;

    } else {

/*        QR Iteration   

          Look for small superdiagonal element. */

L90:
	if (l != lend) {
	    lendp1 = lend + 1;
	    i__1 = lendp1;
	    for (m = l; m >= i__1; --m) {
/* Computing 2nd power */
		d__2 = (d__1 = e[m - 1], absMACRO(d__1));
		tst = d__2 * d__2;
		if (tst <= eps2 * (d__1 = d__[m], absMACRO(d__1)) * (d__2 = d__[m 
			- 1], absMACRO(d__2)) + safmin) {
		    goto L110;
		}
/* L100: */
	    }
	}

	m = lend;

L110:
	if (m > lend) {
	    e[m - 1] = 0.;
	}
	p = d__[l];
	if (m == l) {
	    goto L130;
	}

/*        If remaining matrix is 2-by-2, use DLAE2 or SLAEV2   
          to compute its eigensystem. */

	if (m == l - 1) {
	    if (icompz > 0) {
		template_lapack_laev2(&d__[l - 1], &e[l - 1], &d__[l], &rt1, &rt2, &c__, &s)
			;
		work[m] = c__;
		work[*n - 1 + m] = s;
		template_lapack_lasr("R", "V", "F", n, &c__2, &work[m], &work[*n - 1 + m], &
			z___ref(1, l - 1), ldz);
	    } else {
		template_lapack_lae2(&d__[l - 1], &e[l - 1], &d__[l], &rt1, &rt2);
	    }
	    d__[l - 1] = rt1;
	    d__[l] = rt2;
	    e[l - 1] = 0.;
	    l += -2;
	    if (l >= lend) {
		goto L90;
	    }
	    goto L140;
	}

	if (jtot == nmaxit) {
	    goto L140;
	}
	++jtot;

/*        Form shift. */

	g = (d__[l - 1] - p) / (e[l - 1] * 2.);
	r__ = template_lapack_lapy2(&g, &c_b10);
	g = d__[m] - p + e[l - 1] / (g + template_lapack_d_sign(&r__, &g));

	s = 1.;
	c__ = 1.;
	p = 0.;

/*        Inner loop */

	lm1 = l - 1;
	i__1 = lm1;
	for (i__ = m; i__ <= i__1; ++i__) {
	    f = s * e[i__];
	    b = c__ * e[i__];
	    template_lapack_lartg(&g, &f, &c__, &s, &r__);
	    if (i__ != m) {
		e[i__ - 1] = r__;
	    }
	    g = d__[i__] - p;
	    r__ = (d__[i__ + 1] - g) * s + c__ * 2. * b;
	    p = s * r__;
	    d__[i__] = g + p;
	    g = c__ * r__ - b;

/*           If eigenvectors are desired, then save rotations. */

	    if (icompz > 0) {
		work[i__] = c__;
		work[*n - 1 + i__] = s;
	    }

/* L120: */
	}

/*        If eigenvectors are desired, then apply saved rotations. */

	if (icompz > 0) {
	    mm = l - m + 1;
	    template_lapack_lasr("R", "V", "F", n, &mm, &work[m], &work[*n - 1 + m], &
		    z___ref(1, m), ldz);
	}

	d__[l] -= p;
	e[lm1] = g;
	goto L90;

/*        Eigenvalue found. */

L130:
	d__[l] = p;

	--l;
	if (l >= lend) {
	    goto L90;
	}
	goto L140;

    }

/*     Undo scaling if necessary */

L140:
    if (iscale == 1) {
	i__1 = lendsv - lsv + 1;
	template_lapack_lascl("G", &c__0, &c__0, &ssfmax, &anorm, &i__1, &c__1, &d__[lsv], 
		n, info);
	i__1 = lendsv - lsv;
	template_lapack_lascl("G", &c__0, &c__0, &ssfmax, &anorm, &i__1, &c__1, &e[lsv], n, 
		info);
    } else if (iscale == 2) {
	i__1 = lendsv - lsv + 1;
	template_lapack_lascl("G", &c__0, &c__0, &ssfmin, &anorm, &i__1, &c__1, &d__[lsv], 
		n, info);
	i__1 = lendsv - lsv;
	template_lapack_lascl("G", &c__0, &c__0, &ssfmin, &anorm, &i__1, &c__1, &e[lsv], n, 
		info);
    }

/*     Check for no convergence to an eigenvalue after a total   
       of N*MAXIT iterations. */

    if (jtot < nmaxit) {
	goto L10;
    }
    i__1 = *n - 1;
    for (i__ = 1; i__ <= i__1; ++i__) {
	if (e[i__] != 0.) {
	    ++(*info);
	}
/* L150: */
    }
    goto L190;

/*     Order eigenvalues and eigenvectors. */

L160:
    if (icompz == 0) {

/*        Use Quick Sort */

	template_lapack_lasrt("I", n, &d__[1], info);

    } else {

/*        Use Selection Sort to minimize swaps of eigenvectors */

	i__1 = *n;
	for (ii = 2; ii <= i__1; ++ii) {
	    i__ = ii - 1;
	    k = i__;
	    p = d__[i__];
	    i__2 = *n;
	    for (j = ii; j <= i__2; ++j) {
		if (d__[j] < p) {
		    k = j;
		    p = d__[j];
		}
/* L170: */
	    }
	    if (k != i__) {
		d__[k] = d__[i__];
		d__[i__] = p;
		template_blas_swap(n, &z___ref(1, i__), &c__1, &z___ref(1, k), &c__1);
	    }
/* L180: */
	}
    }

L190:
    return 0;

/*     End of DSTEQR */

} /* dsteqr_ */

#undef z___ref


#endif