1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
|
! Copyright (C) 2006 Imperial College London and others.
!
! Please see the AUTHORS file in the main source directory for a full list
! of copyright holders.
!
! Prof. C Pain
! Applied Modelling and Computation Group
! Department of Earth Science and Engineering
! Imperial College London
!
! amcgsoftware@imperial.ac.uk
!
! This library is free software; you can redistribute it and/or
! modify it under the terms of the GNU Lesser General Public
! License as published by the Free Software Foundation,
! version 2.1 of the License.
!
! This library is distributed in the hope that it will be useful,
! but WITHOUT ANY WARRANTY; without even the implied warranty of
! MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
! Lesser General Public License for more details.
!
! You should have received a copy of the GNU Lesser General Public
! License along with this library; if not, write to the Free Software
! Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
! USA
#include "fdebug.h"
module ann_bp
use spud
implicit none
! sample_num: number of samples. here equals to timesteps .In : NO. of input variables of each sample.
!Out :NO. of output variables of each sample.
! Neuron: NO.of neuron，Train
! integer, save :: sample_num = 200 !Data
integer, save :: in_num=2
integer, save :: out_num=1
integer, save :: Neuron =45
integer, save :: Train_num=200000
real, save :: A=0.2
real, save :: B=0.4
real, save :: aa=0.2
real, save :: bb=0.3
integer :: timestep,nsvd,sample_num, coef_num
! real, dimension(:,:),allocatable ::d_in, d_out,w,v,dv,dw
! real, dimension(:), allocatable :: OutputData,o !Maxin,Minin,Maxout,Minout
! double precision e;
! real Maxin(2),Minin(2),Maxout(1),Minout(1);
real, dimension(:,:),allocatable :: d_in_allcoef ! store coef_pod_all
real d_in(1600,2),d_out(1600,1);
real w(45,2),o(45),v(1,45);
real Maxin(2),Minin(2),Maxout(1),Minout(1);
real OutputData(1);
real dv(1,45),dw(45,2);
real e;
real, dimension(:,:), allocatable ::pod_coef_all_diff, pod_coef_all_ann,pod_coef_all_diff_obj
real :: w_all(360,45,2),o_all(360,45),v_all(360,1,45)
contains
subroutine ann_bp_main(total_timestep,timestep,output,l)
integer, intent(in) :: total_timestep,timestep,l
real, intent(inout) ::output
integer :: i, k,m,n
real, dimension(:),allocatable ::s
real ::ss
real :: coef, coefdiff
integer :: ctrl_timestep, ctrl_timestep1
call get_option(&
'/reduced_model/pod_basis_formation/pod_basis_count', nsvd)
coef_num=3*nsvd
allocate(s(3*nsvd))
! sample_num=820!timestep
sample_num=timestep-3
allocate(pod_coef_all_diff(total_timestep,coef_num))
allocate(pod_coef_all_diff_obj(total_timestep,coef_num))
allocate(pod_coef_all_ann(total_timestep,coef_num))
pod_coef_all_ann=0
pod_coef_all_diff=0
pod_coef_all_diff_obj=0
print *, 'timestep',timestep
open(1,file='coef_pod_all_obv')
read(1,*)((pod_coef_all_ann(m,n),n=1,coef_num),m=1,timestep) !initialisation
close(1)
pod_coef_all_diff=pod_coef_all_ann
do m=1, timestep-1
pod_coef_all_diff(m,:)= pod_coef_all_ann(m+1,:)-pod_coef_all_ann(m,:)
enddo
do m=1, timestep-2
pod_coef_all_diff_obj(m,:)= pod_coef_all_ann(m+2,:)-pod_coef_all_ann(m+1,:)
enddo
open(21,file='coef_pod_all_diff')
write(21,*)((pod_coef_all_diff(m,n),n=1,coef_num),m=1,timestep)
close(21)
open(21,file='coef_pod_all_diff_obj')
write(21,*)((pod_coef_all_diff_obj(m,n),n=1,coef_num),m=1,timestep)
close(21)
! stop 93
if (have_option("/reduced_model/training")) then
ctrl_timestep=1500
ctrl_timestep1=1500
else
ctrl_timestep=555555
ctrl_timestep1=19
endif
if(timestep .eq. ctrl_timestep) then ! if run then eq.5555
! call readData(l)
call readtestData2(l)
! call readtestData()
call initBPNetwork()
call trainNetwork()
call writeweight(l)
coef= pod_coef_all_ann(timestep-2,l)
! coefdiff= pod_coef_all_diff(timestep-2,1)
coefdiff= pod_coef_all_ann(timestep-1,l)
call bpresultok(coef, coefdiff,ss)
output=ss
! output=ss+pod_coef_all_ann(timestep-1,1)
print*, 'coef+coefdiff=', output
elseif(timestep >ctrl_timestep1) then
if(timestep .eq. ctrl_timestep1+1) then
call readallweight()
call readweight(l)
else
call readweight(l)
endif
coef= pod_coef_all_ann(timestep-2,l)
coefdiff= pod_coef_all_ann(timestep-1,l)
print *, 'coef,coefdiff',coef,coefdiff
call bpresultok(coef, coefdiff,ss)
print *, 'coef+coefdiff=', ss
! output=ss+pod_coef_all_ann(timestep-1,1)
output=ss
! print*, 'coef+coefdiff=', output
! stop 93
endif
! open(40,file='pod_coef')
! write(40,*)(s(i),i=1,3*nsvd)
! close(40)
deallocate(pod_coef_all_diff)
deallocate(pod_coef_all_ann)
deallocate(pod_coef_all_diff_obj)
end subroutine ann_bp_main
subroutine readtestData()
integer :: k,j
integer :: i, timestep
open(1,file='in')
! do i=1, 820
! do j=1,2
read(1,*)((d_in(i,j),j=1,2),i=1,820)
! enddo
! enddo
close(1)
open(1,file='out')
read(1,*)(d_out(i,1),i=1,820)
close(1)
do i=1,820
print *, 'out' ,d_in(i,1), d_in(i,2), d_out(i,1)
enddo
!stop 153
end subroutine readtestData
subroutine readData(l)
integer :: k,j
!integer, intent(in) :: i, timestep
double precision :: d_in_tmp(200), d_in_diff(200)
integer, intent(in) ::l
DO k=1, sample_num
d_in(k,1)=pod_coef_all_ann(k,l)
ENDDO
DO k=1, sample_num
d_in(k,2)=pod_coef_all_diff(k,l)
d_out(k,1)=pod_coef_all_diff_obj(k,l) !test output
ENDDO
end subroutine readData
subroutine readtestData2(l)
integer :: k,j
integer, intent(in) ::l
DO k=1, sample_num
d_in(k,1)=pod_coef_all_ann(k,l)
ENDDO
DO k=1, sample_num
d_in(k,2)=pod_coef_all_ann(k+1,l)
d_out(k,1)=pod_coef_all_ann(k+2,l) !test output
ENDDO
! print *,'d_ind_out', d_in,d_out
end subroutine readtestData2
subroutine initBPNetwork() ! normalisation, between [0,1]
integer :: i,j
real::x
Do i=1,in_num ! cannot use minval because some of the vector is null.
Minin(i)=d_in(1,i)
Maxin(i)=d_in(1,i)
do j=1,sample_num
if(Minin(i)>d_in(j,i)) then
Minin(i)=d_in(j,i)
endif
if(Maxin(i)<d_in(j,i)) then
Maxin(i)=d_in(j,i)
endif
enddo
ENDDO
Do i=1,out_num
Minout(i)=d_out(1,i)
Maxout(i)=d_out(1,i)
do j=1,sample_num
if(Minout(i)>d_out(j,i)) then
Minout(i)=d_out(j,i)
endif
if(Maxout(i)<d_out(j,i)) then
Maxout(i)=d_out(j,i)
endif
enddo
! print *, 'Minin(1),Maxin(1),Minin(2),Maxin(2)',Minin(1),Maxin(1),Minin(2),Maxin(2)
ENDDO
open(1,file='maxmin')
write(1,*) Minin(1),Maxin(1),Minin(2),Maxin(2),Maxout(1),Minout(1)
close(1)
DO i=1, in_num
do j=1,sample_num
d_in(j,i)=(d_in(j,i)-Minin(i)+1)/(Maxin(i)-Minin(i)+1)
enddo
ENDDO
DO i=1,out_num
do j=1,sample_num
d_out(j,i)=(d_out(j,i)-Minout(i)+1)/(Maxout(i)-Minout(i)+1);
enddo
ENDDO
do i=1,820
! print *, 'out' ,d_in(i,1), d_in(i,2), d_out(i,1)
enddo
! stop 208
call random_seed ()
! print *, 'Neuroninnit', Neuron
DO i=1,Neuron
do j=1,in_num
call random_number (x)
w(i,j)=x*2-1 !rand()*2.0/RAND_MAX-1;
dw(i,j)=0;
enddo
ENDDO
call random_seed ()
DO i=1,Neuron
do j=1,out_num
call random_number (x)
v(j,i)=x*2-1!rand()*2.0/RAND_MAX-1;
dv(j,i)=0;
enddo
ENDDO
! print *, 'w,v', w,v
end subroutine initBPNetwork
subroutine computO(var)
integer :: var,i,j
real :: s,y !s :sum
DO i=1,Neuron
s=0
do j=1,in_num
s=s+w(i,j)*d_in(var,j)
enddo
o(i)=1./(1.+exp(-s))
ENDDO
DO i=1,out_num
s=0
do j=1,Neuron
s=s+v(i,j)*o(j)
enddo
OutputData(i)=s
ENDDO
end subroutine computO
subroutine backUpdate(var)
integer :: var,i,j
real :: t
DO i=1,Neuron
t=0
do j=1,out_num
t=t+(OutputData(j)-d_out(var,j))*v(j,i)
dv(j,i)=A*dv(j,i)+B*(OutputData(j)-d_out(var,j))*o(i)
v(j,i)=v(j,i)-dv(j,i)
enddo
do j=1,in_num
dw(i,j)=aa*dw(i,j)+bb*t*o(i)*(1-o(i))*d_in(var,j)
w(i,j)=w(i,j)-dw(i,j)
enddo
ENDDO
end subroutine backUpdate
subroutine trainNetwork()
integer :: var,i,j,training_num
real :: e
training_num=0
120 e=0
DO i=1, sample_num
call computO(i)
do j=1,out_num
e=e+abs((OutputData(j)-d_out(i,j))/d_out(i,j))
enddo
call backUpdate(i)
ENDDO
! print *, 'sample_num , whole_error and traning number' ,sample_num, e/sample_num ,training_num
training_num=training_num+1
if (e/sample_num>0.001 .and.training_num<Train_num) goto 120
print *,'training is over the training number is ', training_num
end subroutine
function bpresult(var1,var2) result (s)
real :: var1,var2,s,y
integer :: i,j
var1=(var1-Minin(1)+1)/(Maxin(1)-Minin(1)+1);
var2=(var2-Minin(2)+1)/(Maxin(2)-Minin(2)+1);
do i=1,Neuron
s=0
s=w(i,1)*var1+w(i,2)*var2
o(i)=1/(1+exp(-s))
enddo
s=0
do j=1,Neuron
s=s+v(1,j)*o(j)
enddo
s=s*(Maxout(1)-Minout(1)+1)+Minout(1)-1
end function bpresult
subroutine bpresultok(var1,var2,s)
real::var1,var2
real, intent(inout) :: s
integer :: i,j
! call initBPNetwork()
! Neuron=45
! print *, 'var1,var2',var1,var2
var1=(var1-Minin(1)+1)/(Maxin(1)-Minin(1)+1);
var2=(var2-Minin(2)+1)/(Maxin(2)-Minin(2)+1);
print *, 'var1,var2,bpresult',var1,var2,Neuron
print *, 'Minin(1),Maxin(1),Minin(2),Maxin(2)',Minin(1),Maxin(1),Minin(2),Maxin(2)
!stop 286
! open(1,file='neuron.txt')
! read(1,*)((w(i,j),i=1,45),j=1,2)
! read(1,*)(v(1,i),i=1,45)
! close(1)
do i=1,Neuron
s=0
s=w(i,1)*var1+w(i,2)*var2
! print *, 'we',w(i,1),w(i,2)
o(i)=1/(1+exp(-s))
! print *, 'out',o(i)
!stop 326
enddo
s=0
do j=1,Neuron
s=s+v(1,j)*o(j)
enddo
s=s*(Maxout(1)-Minout(1)+1)+Minout(1)-1
print *, 's' , s
end subroutine bpresultok
subroutine writeweight(l)
integer, intent(in) :: l
integer :: i,j
if(l.eq. 1) then
open(1,file='weight_w')
write(1,*)((w(i,j),j=1,2),i=1,45)
! write(1,*)(o(i),i=1,45)
close(1)
else
open(1,file='weight_w', position='append',ACTION='WRITE')
write(1,*)((w(i,j),j=1,2),i=1,45)
! write(1,*)(o(i),i=1,45)
close(1)
endif
if(l.eq. 1) then
open(2,file='weight_o')
!write(1,*)((w(i,j),j=1,2),i=1,45)
write(2,*)(o(i),i=1,45)
close(2)
else
open(2,file='weight_o', position='append',ACTION='WRITE')
!write(1,*)((w(i,j),j=1,2),i=1,45)
write(2,*)(o(i),i=1,45)
close(2)
endif
if(l.eq. 1) then
open(1,file='weight_v')
write(1,*)((v(i,j),j=1,45),i=1,1)
close(1)
else
open(1,file='weight_v', position='append',ACTION='WRITE')
write(1,*)((v(i,j),j=1,45),i=1,1)
close(1)
endif
end subroutine writeweight
subroutine readallweight()
integer :: i,j,k
!real :: w_all(360,45,2),o_all(360,45),v_all(360,1,45)
open(1,file='maxmin')
read(1,*) Minin(1),Maxin(1),Minin(2),Maxin(2),Maxout(1),Minout(1)
close(1)
do k=1, coef_num
open(1,file='weight_w')
read(1,*)((w_all(k,i,j),j=1,2),i=1,45)
close(1)
enddo
do k=1, coef_num
open(2,file='weight_o')
read(2,*)(o_all(k,i),i=1,45)
close(2)
enddo
do k=1, coef_num
open(1,file='weight_v')
read(1,*)((v_all(k,i,j),j=1,45),i=1,1)
close(1)
enddo
end subroutine readallweight
subroutine readweight(l)
integer, intent(in) :: l
integer :: i,j,k
! real :: w_all(360,45,2),o_all(360,45),v_all(360,1,45)
w(:,:)=w_all(l,:,:)
o(:)=o_all(l,:)
v(:,:)=v_all(l,:,:)
end subroutine readweight
end module ann_bp
|