#include "blaswrap.h" #include "f2c.h" integer slaneg_(integer *n, real *d__, real *lld, real *sigma, real *pivmin, integer *r__) { /* -- LAPACK auxiliary routine (version 3.1) -- Univ. of Tennessee, Univ. of California Berkeley and NAG Ltd.. November 2006 Purpose ======= SLANEG computes the Sturm count, the number of negative pivots encountered while factoring tridiagonal T - sigma I = L D L^T. This implementation works directly on the factors without forming the tridiagonal matrix T. The Sturm count is also the number of eigenvalues of T less than sigma. This routine is called from SLARRB. The current routine does not use the PIVMIN parameter but rather requires IEEE-754 propagation of Infinities and NaNs. This routine also has no input range restrictions but does require default exception handling such that x/0 produces Inf when x is non-zero, and Inf/Inf produces NaN. For more information, see: Marques, Riedy, and Voemel, "Benefits of IEEE-754 Features in Modern Symmetric Tridiagonal Eigensolvers," SIAM Journal on Scientific Computing, v28, n5, 2006. DOI 10.1137/050641624 (Tech report version in LAWN 172 with the same title.) Arguments ========= N (input) INTEGER The order of the matrix. D (input) REAL array, dimension (N) The N diagonal elements of the diagonal matrix D. LLD (input) REAL array, dimension (N-1) The (N-1) elements L(i)*L(i)*D(i). SIGMA (input) REAL Shift amount in T - sigma I = L D L^T. PIVMIN (input) REAL The minimum pivot in the Sturm sequence. May be used when zero pivots are encountered on non-IEEE-754 architectures. R (input) INTEGER The twist index for the twisted factorization that is used for the negcount. Further Details =============== Based on contributions by Osni Marques, LBNL/NERSC, USA Christof Voemel, University of California, Berkeley, USA Jason Riedy, University of California, Berkeley, USA ===================================================================== Some architectures propagate Infinities and NaNs very slowly, so the code computes counts in BLKLEN chunks. Then a NaN can propagate at most BLKLEN columns before being detected. This is not a general tuning parameter; it needs only to be just large enough that the overhead is tiny in common cases. Parameter adjustments */ /* System generated locals */ integer ret_val, i__1, i__2, i__3, i__4; /* Local variables */ static integer j; static real p, t; static integer bj; static real tmp; static integer neg1, neg2; static real bsav, gamma, dplus; static integer negcnt; static logical sawnan; extern logical sisnan_(real *); static real dminus; --lld; --d__; /* Function Body */ negcnt = 0; /* I) upper part: L D L^T - SIGMA I = L+ D+ L+^T */ t = -(*sigma); i__1 = *r__ - 1; for (bj = 1; bj <= i__1; bj += 128) { neg1 = 0; bsav = t; /* Computing MIN */ i__3 = bj + 127, i__4 = *r__ - 1; i__2 = min(i__3,i__4); for (j = bj; j <= i__2; ++j) { dplus = d__[j] + t; if (dplus < 0.f) { ++neg1; } tmp = t / dplus; t = tmp * lld[j] - *sigma; /* L21: */ } sawnan = sisnan_(&t); /* Run a slower version of the above loop if a NaN is detected. A NaN should occur only with a zero pivot after an infinite pivot. In that case, substituting 1 for T/DPLUS is the correct limit. */ if (sawnan) { neg1 = 0; t = bsav; /* Computing MIN */ i__3 = bj + 127, i__4 = *r__ - 1; i__2 = min(i__3,i__4); for (j = bj; j <= i__2; ++j) { dplus = d__[j] + t; if (dplus < 0.f) { ++neg1; } tmp = t / dplus; if (sisnan_(&tmp)) { tmp = 1.f; } t = tmp * lld[j] - *sigma; /* L22: */ } } negcnt += neg1; /* L210: */ } /* II) lower part: L D L^T - SIGMA I = U- D- U-^T */ p = d__[*n] - *sigma; i__1 = *r__; for (bj = *n - 1; bj >= i__1; bj += -128) { neg2 = 0; bsav = p; /* Computing MAX */ i__3 = bj - 127; i__2 = max(i__3,*r__); for (j = bj; j >= i__2; --j) { dminus = lld[j] + p; if (dminus < 0.f) { ++neg2; } tmp = p / dminus; p = tmp * d__[j] - *sigma; /* L23: */ } sawnan = sisnan_(&p); /* As above, run a slower version that substitutes 1 for Inf/Inf. */ if (sawnan) { neg2 = 0; p = bsav; /* Computing MAX */ i__3 = bj - 127; i__2 = max(i__3,*r__); for (j = bj; j >= i__2; --j) { dminus = lld[j] + p; if (dminus < 0.f) { ++neg2; } tmp = p / dminus; if (sisnan_(&tmp)) { tmp = 1.f; } p = tmp * d__[j] - *sigma; /* L24: */ } } negcnt += neg2; /* L230: */ } /* III) Twist index T was shifted by SIGMA initially. */ gamma = t + *sigma + p; if (gamma < 0.f) { ++negcnt; } ret_val = negcnt; return ret_val; } /* slaneg_ */