root / src / lapack / double / dlabad.f @ 2
Historique | Voir | Annoter | Télécharger (1,84 ko)
| 1 | 1 | equemene | SUBROUTINE DLABAD( SMALL, LARGE ) |
|---|---|---|---|
| 2 | 1 | equemene | * |
| 3 | 1 | equemene | * -- LAPACK auxiliary routine (version 3.2) -- |
| 4 | 1 | equemene | * -- LAPACK is a software package provided by Univ. of Tennessee, -- |
| 5 | 1 | equemene | * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..-- |
| 6 | 1 | equemene | * November 2006 |
| 7 | 1 | equemene | * |
| 8 | 1 | equemene | * .. Scalar Arguments .. |
| 9 | 1 | equemene | DOUBLE PRECISION LARGE, SMALL |
| 10 | 1 | equemene | * .. |
| 11 | 1 | equemene | * |
| 12 | 1 | equemene | * Purpose |
| 13 | 1 | equemene | * ======= |
| 14 | 1 | equemene | * |
| 15 | 1 | equemene | * DLABAD takes as input the values computed by DLAMCH for underflow and |
| 16 | 1 | equemene | * overflow, and returns the square root of each of these values if the |
| 17 | 1 | equemene | * log of LARGE is sufficiently large. This subroutine is intended to |
| 18 | 1 | equemene | * identify machines with a large exponent range, such as the Crays, and |
| 19 | 1 | equemene | * redefine the underflow and overflow limits to be the square roots of |
| 20 | 1 | equemene | * the values computed by DLAMCH. This subroutine is needed because |
| 21 | 1 | equemene | * DLAMCH does not compensate for poor arithmetic in the upper half of |
| 22 | 1 | equemene | * the exponent range, as is found on a Cray. |
| 23 | 1 | equemene | * |
| 24 | 1 | equemene | * Arguments |
| 25 | 1 | equemene | * ========= |
| 26 | 1 | equemene | * |
| 27 | 1 | equemene | * SMALL (input/output) DOUBLE PRECISION |
| 28 | 1 | equemene | * On entry, the underflow threshold as computed by DLAMCH. |
| 29 | 1 | equemene | * On exit, if LOG10(LARGE) is sufficiently large, the square |
| 30 | 1 | equemene | * root of SMALL, otherwise unchanged. |
| 31 | 1 | equemene | * |
| 32 | 1 | equemene | * LARGE (input/output) DOUBLE PRECISION |
| 33 | 1 | equemene | * On entry, the overflow threshold as computed by DLAMCH. |
| 34 | 1 | equemene | * On exit, if LOG10(LARGE) is sufficiently large, the square |
| 35 | 1 | equemene | * root of LARGE, otherwise unchanged. |
| 36 | 1 | equemene | * |
| 37 | 1 | equemene | * ===================================================================== |
| 38 | 1 | equemene | * |
| 39 | 1 | equemene | * .. Intrinsic Functions .. |
| 40 | 1 | equemene | INTRINSIC LOG10, SQRT |
| 41 | 1 | equemene | * .. |
| 42 | 1 | equemene | * .. Executable Statements .. |
| 43 | 1 | equemene | * |
| 44 | 1 | equemene | * If it looks like we're on a Cray, take the square root of |
| 45 | 1 | equemene | * SMALL and LARGE to avoid overflow and underflow problems. |
| 46 | 1 | equemene | * |
| 47 | 1 | equemene | IF( LOG10( LARGE ).GT.2000.D0 ) THEN |
| 48 | 1 | equemene | SMALL = SQRT( SMALL ) |
| 49 | 1 | equemene | LARGE = SQRT( LARGE ) |
| 50 | 1 | equemene | END IF |
| 51 | 1 | equemene | * |
| 52 | 1 | equemene | RETURN |
| 53 | 1 | equemene | * |
| 54 | 1 | equemene | * End of DLABAD |
| 55 | 1 | equemene | * |
| 56 | 1 | equemene | END |