| 1 |
subroutine elmhes(nm,n,low,igh,a,int) |
| 2 |
c |
| 3 |
integer i,j,m,n,la,nm,igh,kp1,low,mm1,mp1 |
| 4 |
double precision a(nm,n) |
| 5 |
double precision x,y |
| 6 |
integer int(igh) |
| 7 |
c |
| 8 |
c this subroutine is a translation of the algol procedure elmhes, |
| 9 |
c num. math. 12, 349-368(1968) by martin and wilkinson. |
| 10 |
c handbook for auto. comp., vol.ii-linear algebra, 339-358(1971). |
| 11 |
c |
| 12 |
c given a real general matrix, this subroutine |
| 13 |
c reduces a submatrix situated in rows and columns |
| 14 |
c low through igh to upper hessenberg form by |
| 15 |
c stabilized elementary similarity transformations. |
| 16 |
c |
| 17 |
c on input |
| 18 |
c |
| 19 |
c nm must be set to the row dimension of two-dimensional |
| 20 |
c array parameters as declared in the calling program |
| 21 |
c dimension statement. |
| 22 |
c |
| 23 |
c n is the order of the matrix. |
| 24 |
c |
| 25 |
c low and igh are integers determined by the balancing |
| 26 |
c subroutine balanc. if balanc has not been used, |
| 27 |
c set low=1, igh=n. |
| 28 |
c |
| 29 |
c a contains the input matrix. |
| 30 |
c |
| 31 |
c on output |
| 32 |
c |
| 33 |
c a contains the hessenberg matrix. the multipliers |
| 34 |
c which were used in the reduction are stored in the |
| 35 |
c remaining triangle under the hessenberg matrix. |
| 36 |
c |
| 37 |
c int contains information on the rows and columns |
| 38 |
c interchanged in the reduction. |
| 39 |
c only elements low through igh are used. |
| 40 |
c |
| 41 |
c questions and comments should be directed to burton s. garbow, |
| 42 |
c mathematics and computer science div, argonne national laboratory |
| 43 |
c |
| 44 |
c this version dated august 1983. |
| 45 |
c |
| 46 |
c ------------------------------------------------------------------ |
| 47 |
c |
| 48 |
la = igh - 1 |
| 49 |
kp1 = low + 1 |
| 50 |
if (la .lt. kp1) go to 200 |
| 51 |
c |
| 52 |
do 180 m = kp1, la |
| 53 |
mm1 = m - 1 |
| 54 |
x = 0.0d0 |
| 55 |
i = m |
| 56 |
c |
| 57 |
do 100 j = m, igh |
| 58 |
if (dabs(a(j,mm1)) .le. dabs(x)) go to 100 |
| 59 |
x = a(j,mm1) |
| 60 |
i = j |
| 61 |
100 continue |
| 62 |
c |
| 63 |
int(m) = i |
| 64 |
if (i .eq. m) go to 130 |
| 65 |
c .......... interchange rows and columns of a .......... |
| 66 |
do 110 j = mm1, n |
| 67 |
y = a(i,j) |
| 68 |
a(i,j) = a(m,j) |
| 69 |
a(m,j) = y |
| 70 |
110 continue |
| 71 |
c |
| 72 |
do 120 j = 1, igh |
| 73 |
y = a(j,i) |
| 74 |
a(j,i) = a(j,m) |
| 75 |
a(j,m) = y |
| 76 |
120 continue |
| 77 |
c .......... end interchange .......... |
| 78 |
130 if (x .eq. 0.0d0) go to 180 |
| 79 |
mp1 = m + 1 |
| 80 |
c |
| 81 |
do 160 i = mp1, igh |
| 82 |
y = a(i,mm1) |
| 83 |
if (y .eq. 0.0d0) go to 160 |
| 84 |
y = y / x |
| 85 |
a(i,mm1) = y |
| 86 |
c |
| 87 |
do 140 j = m, n |
| 88 |
140 a(i,j) = a(i,j) - y * a(m,j) |
| 89 |
c |
| 90 |
do 150 j = 1, igh |
| 91 |
150 a(j,m) = a(j,m) + y * a(j,i) |
| 92 |
c |
| 93 |
160 continue |
| 94 |
c |
| 95 |
180 continue |
| 96 |
c |
| 97 |
200 return |
| 98 |
end |