| 1 |
c********************************************************************\ |
| 2 |
c minos C interface |
| 3 |
c ASCEND |
| 4 |
c (C) Ben Allan, August 9, 1994 |
| 5 |
c $Revision: 1.1.1.1 $ |
| 6 |
c $Date: 1996/04/30 16:34:15 $ |
| 7 |
c |
| 8 |
c MINOS 5.4 is proprietary software sitelicensed to Carnegie Mellon. |
| 9 |
c Others who wish to use minos with ASCEND must get their own license |
| 10 |
c and MINOS 5.4 sources. We provide only interface code to feed problems |
| 11 |
c to MINOS 5.4. |
| 12 |
c********************************************************************/ |
| 13 |
|
| 14 |
|
| 15 |
c********************************************************************\ |
| 16 |
c Function to return values of various common block values. |
| 17 |
c major: number of major iterations by minos (int) |
| 18 |
c minor: number of minor iterations by minos (int) |
| 19 |
c********************************************************************/ |
| 20 |
subroutine get_minos_common(major,minor) |
| 21 |
|
| 22 |
implicit double precision (a-h,o-z) |
| 23 |
common /m8save/ vimax ,virel ,maxvi ,majits,minits,nssave |
| 24 |
integer major, minor |
| 25 |
major= majits |
| 26 |
minor= minits |
| 27 |
|
| 28 |
return |
| 29 |
end |
| 30 |
|
| 31 |
********************************************************************* |
| 32 |
* The following utilities are also used by minos. |
| 33 |
* They are defined in mi15blas.f in normal minos installations. |
| 34 |
* |
| 35 |
* dddiv ddscl dload dnorm1 |
| 36 |
* hcopy hload icopy iload iload1 |
| 37 |
* |
| 38 |
* These could be tuned to the machine being used. |
| 39 |
* dload is used the most. |
| 40 |
* |
| 41 |
*++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ |
| 42 |
|
| 43 |
|
| 44 |
subroutine dddiv ( n, d, incd, x, incx ) |
| 45 |
|
| 46 |
implicit double precision (a-h,o-z) |
| 47 |
double precision d(*), x(*) |
| 48 |
|
| 49 |
* dddiv performs the diagonal scaling x = x / d. |
| 50 |
|
| 51 |
integer i, id, ix |
| 52 |
external dscal |
| 53 |
intrinsic abs |
| 54 |
parameter ( one = 1.0d+0 ) |
| 55 |
|
| 56 |
if (n .gt. 0) then |
| 57 |
if (incd .eq. 0 .and. incx .ne. 0) then |
| 58 |
call dscal ( n, one/d(1), x, abs(incx) ) |
| 59 |
else if (incd .eq. incx .and. incd .gt. 0) then |
| 60 |
do 10 id = 1, 1 + (n - 1)*incd, incd |
| 61 |
x(id) = x(id) / d(id) |
| 62 |
10 continue |
| 63 |
else |
| 64 |
if (incx .ge. 0) then |
| 65 |
ix = 1 |
| 66 |
else |
| 67 |
ix = 1 - (n - 1)*incx |
| 68 |
end if |
| 69 |
if (incd .gt. 0) then |
| 70 |
do 20 id = 1, 1 + (n - 1)*incd, incd |
| 71 |
x(ix) = x(ix) / d(id) |
| 72 |
ix = ix + incx |
| 73 |
20 continue |
| 74 |
else |
| 75 |
id = 1 - (n - 1)*incd |
| 76 |
do 30 i = 1, n |
| 77 |
x(ix) = x(ix) / d(id) |
| 78 |
id = id + incd |
| 79 |
ix = ix + incx |
| 80 |
30 continue |
| 81 |
end if |
| 82 |
end if |
| 83 |
end if |
| 84 |
|
| 85 |
* end of dddiv |
| 86 |
end |
| 87 |
|
| 88 |
*++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ |
| 89 |
|
| 90 |
subroutine ddscl ( n, d, incd, x, incx ) |
| 91 |
|
| 92 |
integer incd, incx, n |
| 93 |
double precision d(*), x(*) |
| 94 |
|
| 95 |
* ddscl performs the diagonal scaling x = d * x. |
| 96 |
|
| 97 |
integer i, id, ix |
| 98 |
external dscal |
| 99 |
intrinsic abs |
| 100 |
|
| 101 |
if (n .gt. 0) then |
| 102 |
if (incd .eq. 0 .and. incx .ne. 0) then |
| 103 |
call dscal ( n, d(1), x, abs(incx) ) |
| 104 |
else if (incd .eq. incx .and. incd .gt. 0) then |
| 105 |
do 10 id = 1, 1 + (n - 1)*incd, incd |
| 106 |
x(id) = d(id)*x(id) |
| 107 |
10 continue |
| 108 |
else |
| 109 |
if (incx .ge. 0) then |
| 110 |
ix = 1 |
| 111 |
else |
| 112 |
ix = 1 - (n - 1)*incx |
| 113 |
end if |
| 114 |
if (incd .gt. 0) then |
| 115 |
do 20 id = 1, 1 + (n - 1)*incd, incd |
| 116 |
x(ix) = d(id)*x(ix) |
| 117 |
ix = ix + incx |
| 118 |
20 continue |
| 119 |
else |
| 120 |
id = 1 - (n - 1)*incd |
| 121 |
do 30 i = 1, n |
| 122 |
x(ix) = d(id)*x(ix) |
| 123 |
id = id + incd |
| 124 |
ix = ix + incx |
| 125 |
30 continue |
| 126 |
end if |
| 127 |
end if |
| 128 |
end if |
| 129 |
|
| 130 |
* end of ddscl |
| 131 |
end |
| 132 |
|
| 133 |
*++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ |
| 134 |
|
| 135 |
subroutine dload ( n, const, x, incx ) |
| 136 |
|
| 137 |
double precision const |
| 138 |
integer incx, n |
| 139 |
double precision x(*) |
| 140 |
|
| 141 |
* dload loads elements of x with const. |
| 142 |
|
| 143 |
double precision zero |
| 144 |
parameter ( zero = 0.0d+0 ) |
| 145 |
integer ix |
| 146 |
|
| 147 |
if (n .gt. 0) then |
| 148 |
if (incx .eq. 1 .and. const .eq. zero) then |
| 149 |
do 10 ix = 1, n |
| 150 |
x(ix) = zero |
| 151 |
10 continue |
| 152 |
else |
| 153 |
do 20 ix = 1, 1 + (n - 1)*incx, incx |
| 154 |
x(ix) = const |
| 155 |
20 continue |
| 156 |
end if |
| 157 |
end if |
| 158 |
|
| 159 |
* end of dload |
| 160 |
end |
| 161 |
|
| 162 |
*++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ |
| 163 |
|
| 164 |
function dnorm1( n, x, incx ) |
| 165 |
|
| 166 |
implicit double precision (a-h,o-z) |
| 167 |
double precision x(*) |
| 168 |
|
| 169 |
* dnorm1 returns the 1-norm of the vector x, scaled by root(n). |
| 170 |
* This approximates an "average" element of x with some allowance |
| 171 |
* for x being sparse. |
| 172 |
|
| 173 |
intrinsic sqrt |
| 174 |
external dasum |
| 175 |
|
| 176 |
d = n |
| 177 |
d = dasum ( n, x, incx ) / sqrt(d) |
| 178 |
dnorm1 = d |
| 179 |
|
| 180 |
* end of dnorm1 |
| 181 |
end |
| 182 |
|
| 183 |
*++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ |
| 184 |
|
| 185 |
subroutine hcopy ( n, hx, incx, hy, incy ) |
| 186 |
|
| 187 |
integer*4 hx(*), hy(*) |
| 188 |
integer incx, incy |
| 189 |
|
| 190 |
* hcopy is the half-integer version of dcopy. |
| 191 |
* In this version of MINOS we no longer use half integers. |
| 192 |
|
| 193 |
integer ix, iy |
| 194 |
|
| 195 |
if (n .gt. 0) then |
| 196 |
if (incx .eq. incy .and. incy .gt. 0) then |
| 197 |
do 10 iy = 1, 1 + (n - 1)*incy, incy |
| 198 |
hy(iy) = hx(iy) |
| 199 |
10 continue |
| 200 |
else |
| 201 |
if (incx .ge. 0) then |
| 202 |
ix = 1 |
| 203 |
else |
| 204 |
ix = 1 - (n - 1)*incx |
| 205 |
end if |
| 206 |
if (incy .gt. 0) then |
| 207 |
do 20 iy = 1, 1 + ( n - 1 )*incy, incy |
| 208 |
hy(iy) = hx(ix) |
| 209 |
ix = ix + incx |
| 210 |
20 continue |
| 211 |
else |
| 212 |
iy = 1 - (n - 1)*incy |
| 213 |
do 30 i = 1, n |
| 214 |
hy(iy) = hx(ix) |
| 215 |
iy = iy + incy |
| 216 |
ix = ix + incx |
| 217 |
30 continue |
| 218 |
end if |
| 219 |
end if |
| 220 |
end if |
| 221 |
|
| 222 |
* end of hcopy |
| 223 |
end |
| 224 |
|
| 225 |
*++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ |
| 226 |
|
| 227 |
subroutine hload ( n, const, hx, incx ) |
| 228 |
|
| 229 |
integer incx, n |
| 230 |
integer const |
| 231 |
integer*4 hx(*) |
| 232 |
|
| 233 |
* hload loads elements of hx with const. |
| 234 |
* Beware that const is INTEGER, not half integer. |
| 235 |
|
| 236 |
integer ix |
| 237 |
|
| 238 |
if (n .gt. 0) then |
| 239 |
if (incx .eq. 1 .and. const .eq. 0) then |
| 240 |
do 10 ix = 1, n |
| 241 |
hx(ix) = 0 |
| 242 |
10 continue |
| 243 |
else |
| 244 |
do 20 ix = 1, 1 + (n - 1)*incx, incx |
| 245 |
hx(ix) = const |
| 246 |
20 continue |
| 247 |
end if |
| 248 |
end if |
| 249 |
|
| 250 |
* end of hload |
| 251 |
end |
| 252 |
|
| 253 |
*++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ |
| 254 |
|
| 255 |
subroutine icopy ( n, x, incx, y, incy ) |
| 256 |
|
| 257 |
integer x(*), y(*) |
| 258 |
integer incx, incy |
| 259 |
|
| 260 |
* icopy is the integer version of dcopy. |
| 261 |
|
| 262 |
integer ix, iy |
| 263 |
|
| 264 |
if (n .gt. 0) then |
| 265 |
if (incx .eq. incy .and. incy .gt. 0) then |
| 266 |
do 10 iy = 1, 1 + (n - 1)*incy, incy |
| 267 |
y(iy) = x(iy) |
| 268 |
10 continue |
| 269 |
else |
| 270 |
if (incx .ge. 0) then |
| 271 |
ix = 1 |
| 272 |
else |
| 273 |
ix = 1 - (n - 1)*incx |
| 274 |
end if |
| 275 |
if (incy .gt. 0) then |
| 276 |
do 20 iy = 1, 1 + ( n - 1 )*incy, incy |
| 277 |
y(iy) = x(ix) |
| 278 |
ix = ix + incx |
| 279 |
20 continue |
| 280 |
else |
| 281 |
iy = 1 - (n - 1)*incy |
| 282 |
do 30 i = 1, n |
| 283 |
y(iy) = x(ix) |
| 284 |
iy = iy + incy |
| 285 |
ix = ix + incx |
| 286 |
30 continue |
| 287 |
end if |
| 288 |
end if |
| 289 |
end if |
| 290 |
|
| 291 |
* end of icopy |
| 292 |
end |
| 293 |
|
| 294 |
*++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ |
| 295 |
|
| 296 |
subroutine iload ( n, const, x, incx ) |
| 297 |
|
| 298 |
integer incx, n |
| 299 |
integer const |
| 300 |
integer x(*) |
| 301 |
|
| 302 |
* iload loads elements of x with const. |
| 303 |
|
| 304 |
integer ix |
| 305 |
|
| 306 |
if (n .gt. 0) then |
| 307 |
if (incx .eq. 1 .and. const .eq. 0) then |
| 308 |
do 10 ix = 1, n |
| 309 |
x(ix) = 0 |
| 310 |
10 continue |
| 311 |
else |
| 312 |
do 20 ix = 1, 1 + (n - 1)*incx, incx |
| 313 |
x(ix) = const |
| 314 |
20 continue |
| 315 |
end if |
| 316 |
end if |
| 317 |
|
| 318 |
* end of iload |
| 319 |
end |
| 320 |
|
| 321 |
*++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ |
| 322 |
|
| 323 |
subroutine iload1( n, const, x, incx ) |
| 324 |
|
| 325 |
integer incx, n |
| 326 |
integer const |
| 327 |
integer x(*) |
| 328 |
|
| 329 |
* iload1 loads elements of x with const, by calling iload. |
| 330 |
* iload1 is needed in MINOS because iload is a file number. |
| 331 |
|
| 332 |
call iload ( n, const, x, incx ) |
| 333 |
|
| 334 |
* end of iload1 |
| 335 |
end |
Parent Directory
| 
Revision Log