Debian Codegraph @

Modules/ThirdParty/VNL/src/vxl/v3p/netlib/eispack/tred1.f

raw
1 subroutine tred1(nm,n,a,d,e,e2)
2 c
3 integer i,j,k,l,n,ii,nm,jp1
4 double precision a(nm,n),d(n),e(n),e2(n)
5 double precision f,g,h,scale
6 c
7 c this subroutine is a translation of the algol procedure tred1,
8 c num. math. 11, 181-195(1968) by martin, reinsch, and wilkinson.
9 c handbook for auto. comp., vol.ii-linear algebra, 212-226(1971).
10 c
11 c this subroutine reduces a real symmetric matrix
12 c to a symmetric tridiagonal matrix using
13 c orthogonal similarity transformations.
14 c
15 c on input
16 c
17 c nm must be set to the row dimension of two-dimensional
18 c array parameters as declared in the calling program
19 c dimension statement.
20 c
21 c n is the order of the matrix.
22 c
23 c a contains the real symmetric input matrix. only the
24 c lower triangle of the matrix need be supplied.
25 c
26 c on output
27 c
28 c a contains information about the orthogonal trans-
29 c formations used in the reduction in its strict lower
30 c triangle. the full upper triangle of a is unaltered.
31 c
32 c d contains the diagonal elements of the tridiagonal matrix.
33 c
34 c e contains the subdiagonal elements of the tridiagonal
35 c matrix in its last n-1 positions. e(1) is set to zero.
36 c
37 c e2 contains the squares of the corresponding elements of e.
38 c e2 may coincide with e if the squares are not needed.
39 c
40 c questions and comments should be directed to burton s. garbow,
41 c mathematics and computer science div, argonne national laboratory
42 c
43 c this version dated august 1983.
44 c
45 c ------------------------------------------------------------------
46 c
47 do 100 i = 1, n
48 d(i) = a(n,i)
49 a(n,i) = a(i,i)
50 100 continue
51 c .......... for i=n step -1 until 1 do -- ..........
52 do 300 ii = 1, n
53 i = n + 1 - ii
54 l = i - 1
55 h = 0.0d0
56 scale = 0.0d0
57 if (l .lt. 1) go to 130
58 c .......... scale row (algol tol then not needed) ..........
59 do 120 k = 1, l
60 120 scale = scale + dabs(d(k))
61 c
62 if (scale .ne. 0.0d0) go to 140
63 c
64 do 125 j = 1, l
65 d(j) = a(l,j)
66 a(l,j) = a(i,j)
67 a(i,j) = 0.0d0
68 125 continue
69 c
70 130 e(i) = 0.0d0
71 e2(i) = 0.0d0
72 go to 300
73 c
74 140 do 150 k = 1, l
75 d(k) = d(k) / scale
76 h = h + d(k) * d(k)
77 150 continue
78 c
79 e2(i) = scale * scale * h
80 f = d(l)
81 g = -dsign(dsqrt(h),f)
82 e(i) = scale * g
83 h = h - f * g
84 d(l) = f - g
85 if (l .eq. 1) go to 285
86 c .......... form a*u ..........
87 do 170 j = 1, l
88 170 e(j) = 0.0d0
89 c
90 do 240 j = 1, l
91 f = d(j)
92 g = e(j) + a(j,j) * f
93 jp1 = j + 1
94 if (l .lt. jp1) go to 220
95 c
96 do 200 k = jp1, l
97 g = g + a(k,j) * d(k)
98 e(k) = e(k) + a(k,j) * f
99 200 continue
100 c
101 220 e(j) = g
102 240 continue
103 c .......... form p ..........
104 f = 0.0d0
105 c
106 do 245 j = 1, l
107 e(j) = e(j) / h
108 f = f + e(j) * d(j)
109 245 continue
110 c
111 h = f / (h + h)
112 c .......... form q ..........
113 do 250 j = 1, l
114 250 e(j) = e(j) - h * d(j)
115 c .......... form reduced a ..........
116 do 280 j = 1, l
117 f = d(j)
118 g = e(j)
119 c
120 do 260 k = j, l
121 260 a(k,j) = a(k,j) - f * e(k) - g * d(k)
122 c
123 280 continue
124 c
125 285 do 290 j = 1, l
126 f = d(j)
127 d(j) = a(l,j)
128 a(l,j) = a(i,j)
129 a(i,j) = f * scale
130 290 continue
131 c
132 300 continue
133 c
134 return
135 end