#include "blaswrap.h" #include "f2c.h" /* Subroutine */ int zlaesy_(doublecomplex *a, doublecomplex *b, doublecomplex *c__, doublecomplex *rt1, doublecomplex *rt2, doublecomplex *evscal, doublecomplex *cs1, doublecomplex *sn1) { /* -- LAPACK auxiliary routine (version 3.1) -- Univ. of Tennessee, Univ. of California Berkeley and NAG Ltd.. November 2006 Purpose ======= ZLAESY computes the eigendecomposition of a 2-by-2 symmetric matrix ( ( A, B );( B, C ) ) provided the norm of the matrix of eigenvectors is larger than some threshold value. RT1 is the eigenvalue of larger absolute value, and RT2 of smaller absolute value. If the eigenvectors are computed, then on return ( CS1, SN1 ) is the unit eigenvector for RT1, hence [ CS1 SN1 ] . [ A B ] . [ CS1 -SN1 ] = [ RT1 0 ] [ -SN1 CS1 ] [ B C ] [ SN1 CS1 ] [ 0 RT2 ] Arguments ========= A (input) COMPLEX*16 The ( 1, 1 ) element of input matrix. B (input) COMPLEX*16 The ( 1, 2 ) element of input matrix. The ( 2, 1 ) element is also given by B, since the 2-by-2 matrix is symmetric. C (input) COMPLEX*16 The ( 2, 2 ) element of input matrix. RT1 (output) COMPLEX*16 The eigenvalue of larger modulus. RT2 (output) COMPLEX*16 The eigenvalue of smaller modulus. EVSCAL (output) COMPLEX*16 The complex value by which the eigenvector matrix was scaled to make it orthonormal. If EVSCAL is zero, the eigenvectors were not computed. This means one of two things: the 2-by-2 matrix could not be diagonalized, or the norm of the matrix of eigenvectors before scaling was larger than the threshold value THRESH (set below). CS1 (output) COMPLEX*16 SN1 (output) COMPLEX*16 If EVSCAL .NE. 0, ( CS1, SN1 ) is the unit right eigenvector for RT1. ===================================================================== Special case: The matrix is actually diagonal. To avoid divide by zero later, we treat this case separately. */ /* Table of constant values */ static doublecomplex c_b1 = {1.,0.}; static integer c__2 = 2; /* System generated locals */ doublereal d__1, d__2; doublecomplex z__1, z__2, z__3, z__4, z__5, z__6, z__7; /* Builtin functions */ double z_abs(doublecomplex *); void pow_zi(doublecomplex *, doublecomplex *, integer *), z_sqrt( doublecomplex *, doublecomplex *), z_div(doublecomplex *, doublecomplex *, doublecomplex *); /* Local variables */ static doublecomplex s, t; static doublereal z__; static doublecomplex tmp; static doublereal babs, tabs, evnorm; if (z_abs(b) == 0.) { rt1->r = a->r, rt1->i = a->i; rt2->r = c__->r, rt2->i = c__->i; if (z_abs(rt1) < z_abs(rt2)) { tmp.r = rt1->r, tmp.i = rt1->i; rt1->r = rt2->r, rt1->i = rt2->i; rt2->r = tmp.r, rt2->i = tmp.i; cs1->r = 0., cs1->i = 0.; sn1->r = 1., sn1->i = 0.; } else { cs1->r = 1., cs1->i = 0.; sn1->r = 0., sn1->i = 0.; } } else { /* Compute the eigenvalues and eigenvectors. The characteristic equation is lambda **2 - (A+C) lambda + (A*C - B*B) and we solve it using the quadratic formula. */ z__2.r = a->r + c__->r, z__2.i = a->i + c__->i; z__1.r = z__2.r * .5, z__1.i = z__2.i * .5; s.r = z__1.r, s.i = z__1.i; z__2.r = a->r - c__->r, z__2.i = a->i - c__->i; z__1.r = z__2.r * .5, z__1.i = z__2.i * .5; t.r = z__1.r, t.i = z__1.i; /* Take the square root carefully to avoid over/under flow. */ babs = z_abs(b); tabs = z_abs(&t); z__ = max(babs,tabs); if (z__ > 0.) { z__5.r = t.r / z__, z__5.i = t.i / z__; pow_zi(&z__4, &z__5, &c__2); z__7.r = b->r / z__, z__7.i = b->i / z__; pow_zi(&z__6, &z__7, &c__2); z__3.r = z__4.r + z__6.r, z__3.i = z__4.i + z__6.i; z_sqrt(&z__2, &z__3); z__1.r = z__ * z__2.r, z__1.i = z__ * z__2.i; t.r = z__1.r, t.i = z__1.i; } /* Compute the two eigenvalues. RT1 and RT2 are exchanged if necessary so that RT1 will have the greater magnitude. */ z__1.r = s.r + t.r, z__1.i = s.i + t.i; rt1->r = z__1.r, rt1->i = z__1.i; z__1.r = s.r - t.r, z__1.i = s.i - t.i; rt2->r = z__1.r, rt2->i = z__1.i; if (z_abs(rt1) < z_abs(rt2)) { tmp.r = rt1->r, tmp.i = rt1->i; rt1->r = rt2->r, rt1->i = rt2->i; rt2->r = tmp.r, rt2->i = tmp.i; } /* Choose CS1 = 1 and SN1 to satisfy the first equation, then scale the components of this eigenvector so that the matrix of eigenvectors X satisfies X * X' = I . (No scaling is done if the norm of the eigenvalue matrix is less than THRESH.) */ z__2.r = rt1->r - a->r, z__2.i = rt1->i - a->i; z_div(&z__1, &z__2, b); sn1->r = z__1.r, sn1->i = z__1.i; tabs = z_abs(sn1); if (tabs > 1.) { /* Computing 2nd power */ d__2 = 1. / tabs; d__1 = d__2 * d__2; z__5.r = sn1->r / tabs, z__5.i = sn1->i / tabs; pow_zi(&z__4, &z__5, &c__2); z__3.r = d__1 + z__4.r, z__3.i = z__4.i; z_sqrt(&z__2, &z__3); z__1.r = tabs * z__2.r, z__1.i = tabs * z__2.i; t.r = z__1.r, t.i = z__1.i; } else { z__3.r = sn1->r * sn1->r - sn1->i * sn1->i, z__3.i = sn1->r * sn1->i + sn1->i * sn1->r; z__2.r = z__3.r + 1., z__2.i = z__3.i + 0.; z_sqrt(&z__1, &z__2); t.r = z__1.r, t.i = z__1.i; } evnorm = z_abs(&t); if (evnorm >= .1) { z_div(&z__1, &c_b1, &t); evscal->r = z__1.r, evscal->i = z__1.i; cs1->r = evscal->r, cs1->i = evscal->i; z__1.r = sn1->r * evscal->r - sn1->i * evscal->i, z__1.i = sn1->r * evscal->i + sn1->i * evscal->r; sn1->r = z__1.r, sn1->i = z__1.i; } else { evscal->r = 0., evscal->i = 0.; } } return 0; /* End of ZLAESY */ } /* zlaesy_ */