~reducedmodelling/fluidity/ROM_Non-intrusive-ann

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
!    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 momentum_equation_reduced_adjoint
    use fields
    use state_module
    use spud
    use fldebug
    !      use momentum_cg
    use divergence_matrix_cv
    use divergence_matrix_cg
    use momentum_dg
    !      use assemble_cmc
    use field_priority_lists
    use momentum_diagnostic_fields, only: calculate_momentum_diagnostics
    use field_options
    !      use compressible_projection
    use boundary_conditions
    use boundary_conditions_from_options
    use sparse_matrices_fields
    use sparse_tools
    use sparse_tools_petsc
    !      use free_surface_module
    use solvers
    !      use full_projection
    !      use petsc_solve_state_module
    use Profiler
    !      use geostrophic_pressure
    !      use hydrostatic_pressure
    !      use vertical_balance_pressure
    !      use foam_drainage, only: calculate_drainage_source_absor
    !      use oceansurfaceforcing
    !      use drag_module
    use parallel_tools
    use linked_lists
    use sparsity_patterns_meshes
    use state_matrices_module
    use vtk_interfaces
    !      use rotated_boundary_conditions
    !      use Weak_BCs
    use reduced_model_runtime
    use state_fields_module
    !      use Tidal_module
    use Coordinates
    use diagnostic_fields, only: calculate_diagnostic_variable
    use dgtools, only: dg_apply_mass
    !      use slope_limiters_dg
    !      use implicit_solids
    use multiphase_module
    !      use pressure_dirichlet_bcs_cv
    use reduced_projection
    use sparse_tools_petsc 
    use global_parameters, only:theta
    
    
    implicit none

    private
    
    public :: solve_momentum_reduced_adjoint
    ! The timestep
    ! The timestep
    real :: dt
    
    ! Are we going to form the Diagonal Schur complement preconditioner?
    logical :: get_diag_schur
    ! Do we need the scaled pressure mass matrix?
    logical :: get_scaled_pressure_mass_matrix
    ! Do we need an auxiliary matrix for full_projection solve?
    logical :: assemble_schur_auxiliary_matrix
    
    ! Do we want to use the compressible projection method?
    logical :: use_compressible_projection
    ! Are we doing a full Schur solve?
    logical :: full_schur
    ! Are we lumping mass or assuming consistent mass?
    logical, dimension(:), allocatable :: lump_mass
    ! are we using a cv pressure 
    logical :: cv_pressure 
    ! for a CG pressure are we testing the continuity with cv
    logical :: cg_pressure_cv_test_continuity
    
    ! Do we need to reassemble the C^T or CMC matrices?
    logical :: reassemble_all_ct_m, reassemble_all_cmc_m
    
    ! Do we want to apply a theta weighting to the pressure gradient term?
    logical :: use_theta_pg
    
    ! Is a theta-weighting term present in the velocity-divergence?
    logical :: use_theta_divergence
    
    ! Are we using a discontinuous Galerkin discretisation?
    logical, dimension(:), allocatable :: dg
    ! True if advection-subcycling is performed
    logical, dimension(:), allocatable :: subcycle
    
    ! Apply KMK stabilisation?
    logical :: apply_kmk
    
    logical :: diagonal_big_m
    logical :: pressure_debugging_vtus
    
    ! Add viscous terms to inverse_masslump for low Re which is only used for pressure correction
    logical :: low_re_p_correction_fix
    
    ! Increased each call to momentum equation, used as index for pressure debugging vtus
    integer, save :: pdv_count = -1
    
    logical, dimension(:), allocatable :: sphere_absorption
    
    ! Are we running a multi-phase simulation?
    logical :: multiphase
    
    !! True if the momentum equation should be solved with the reduced model.
    logical :: reduced_model,DEIM
    integer :: ii,jj,deim_number,udim
    type(vector_field) :: deim_rhs_u
    type(scalar_field) :: deim_rhs_p
    ! real, dimension(:), allocatable :: pod_coef_deim
    type(state_type), dimension(:), pointer :: deim_state => null()   !output state from full model
    type(state_type), dimension(:), pointer :: deim_state_res => null()   !output state from reduced_model
    ! type(state_type), dimension(:), pointer :: deim_state_resl => null() 
    type(state_type), dimension(:), allocatable :: deim_state_resl
    

  contains
    
    subroutine solve_momentum_reduced_adjoint(state, at_first_timestep, timestep, POD_state, snapmean, eps, its)
      !!< Construct and solve the momentum and continuity equations
      !!< using Chorin's projection method (Chorin, 1968)
       
      ! An array of buckets full of fields
      ! The whole array is needed for the sake of multimaterial assembly
      type(state_type), dimension(:), intent(inout) :: state
     
      logical, intent(in) :: at_first_timestep
      integer, intent(in) :: timestep
      type(state_type), dimension(:,:,:), intent(inout) :: POD_state 
      
      ! Counter iterating over each state
      integer :: istate 
      
      ! The pressure projection matrix (extracted from state)
      type(csr_matrix), pointer :: cmc_m
      
      ! logical to indicate whether ct_m and cmc_m need reassembling
      ! (used for each state within the assembly loop)
      logical :: reassemble_ct_m, reassemble_cmc_m
      ! is there a pressure in state?
      logical :: have_pressure
      ! Are we solving a Poisson pressure equation?
      logical :: poisson_p
      
      ! Matrix sparsity patterns for the matrices we allocate locally
      type(csr_sparsity), pointer :: u_sparsity
      
      !! Locally allocated matrices:
      ! Matrix for split explicit advection
      type(petsc_csr_matrix), dimension(:), allocatable, target  :: big_m_tmp
      type(block_csr_matrix), dimension(:), allocatable :: subcycle_m
      ! Pointer to matrix for full projection solve:
      type(petsc_csr_matrix_pointer), dimension(:), allocatable :: inner_m
      ! Pointer to preconditioner matrix for full projection solve:
      type(csr_matrix), pointer :: full_projection_preconditioner
      ! Auxiliary matrix for full_projection solve
      type(csr_sparsity), pointer :: schur_auxiliary_matrix_sparsity
      type(csr_matrix) :: schur_auxiliary_matrix
      ! Scaled pressure mass matrix - used for preconditioning full projection solve:
      type(csr_matrix), target :: scaled_pressure_mass_matrix
      type(csr_sparsity), pointer :: scaled_pressure_mass_matrix_sparsity
      ! Left hand matrix of CMC. For incompressibe flow this points to ct_m as they are identical, 
      ! unless for CG pressure with CV tested continuity case when this matrix will be the 
      ! CV divergence tested matrix and ct_m the CG divergence tested matrix (right hand matrix of CMC).
      ! For compressible flow this differs to ct_m in that it will contain the variable density.
      type(block_csr_matrix_pointer), dimension(:), allocatable :: ctp_m
      ! The lumped mass matrix (may vary per component as absorption could be included)
      type(vector_field), dimension(1:size(state)) :: inverse_masslump, visc_inverse_masslump
      ! Mass matrix
      type(petsc_csr_matrix), dimension(1:size(state)), target :: mass
      ! For DG:
      type(block_csr_matrix), dimension(1:size(state)):: inverse_mass
       
      ! Momentum RHS
      type(vector_field), dimension(1:size(state)):: rhs_deim, rhs_advec,rhs_deim_res
      
      ! Projection RHS
      type(scalar_field) :: projec_rhs
      
      ! Do we want to assemble the KMK stabilisation matrix?
      logical :: assemble_kmk
      
      ! Change in pressure
      type(scalar_field) :: delta_p
      ! Change in velocity
      type(vector_field) :: delta_u
      
      ! Dummy fields
      type(scalar_field), pointer :: dummyscalar, dummydensity, dummypressure
      
      ! Pressure and density
      type(scalar_field), pointer :: p, density
      type(mesh_type), pointer :: p_mesh
      ! Velocity and space
      type(vector_field), pointer :: u, x
      
      ! with free-surface or compressible pressure projection pressures 
      ! are at integer time levels and we apply a theta weighting to the
      ! pressure gradient term
      real :: theta_pg
      ! in this case p_theta=theta_pg*p+(1-theta_pg)*old_p
      type(scalar_field), pointer :: old_p, p_theta
      ! With free-surface or compressible-projection the velocity divergence is
      ! calculated at time n+theta_divergence instead of at the end of the timestep
      real :: theta_divergence
      type(vector_field), pointer :: old_u
      ! all of this only applies if use_theta_pg .eqv. .true.
      ! without a free surface, or with a free surface and theta==1
      ! use_theta_pg .eqv. .false. and p_theta => p
      
      ! What is the equation type?
      character(len=FIELD_NAME_LEN) :: equation_type, poisson_scheme, schur_scheme, pressure_pmat
      
      integer :: stat
      real ::  theta_pp
      
      ! The list of stiff nodes
      ! This is saved because the list is only formed when cmc is assembled, which
      ! isn't necessarily every time this subroutine is called but the list is
      ! still needed to fix the rhs (applying the fix to cmc itself wipes out the
      ! information that would be required to recompile the list)
      type(ilist), save :: stiff_nodes_list
      
      
      !! Variables for multi-phase flow model
      integer :: prognostic_count
      ! Do we have a prognostic pressure field to solve for?
      logical :: prognostic_p = .false.
      ! Prognostic pressure field's state index (if present)
      integer :: prognostic_p_istate 
      ! The 'global' CMC matrix (the sum of all individual phase CMC matrices)
      type(csr_matrix), pointer :: cmc_global
      ! An array of submaterials of the current phase in state(istate).
      type(state_type), dimension(:), pointer :: submaterials
      ! The index of the current phase (i.e. state(istate)) in the submaterials array
      integer :: submaterials_istate
      ! Do we have fluid-particle drag between phases?
      logical :: have_fp_drag
      real :: scale
      
      
      !!for reduced model
      type(vector_field), pointer :: snapmean_velocity
      type(scalar_field), pointer :: snapmean_pressure
      type(vector_field), pointer :: POD_velocity, POD_velocity_deim, velocity_deim,velocity_deim_snapmean
      type(scalar_field), pointer :: POD_pressure
      
      type(pod_matrix_type) :: pod_matrix, pod_matrix_mass, pod_matrix_adv,pod_matrix_B
      type(pod_rhs_type) :: pod_rhs
      real, dimension(:), allocatable :: pod_coef,pod_coef_dt
      real, dimension(:,:), allocatable :: pod_sol_velocity, pod_ct_m
      real, dimension(:), allocatable :: pod_sol_pressure
      integer :: d, i, j, k,dim,jd
      real, intent(in) :: eps
      logical, intent(in) :: snapmean
      logical :: timestep_check
      type(scalar_field) :: u_cpt
      
      !  real, dimension(:,:), allocatable :: A_deim ,mom_rhs_deim  !() name change
      !  real, dimension(:), allocatable :: b_deim,Ny
      real, dimension(:,:), allocatable :: P_mn,Temp_pu,Temp_vu,Temp_vupu,mom_rhs_deim !A_deim,
      real, dimension(:,:,:), allocatable :: V_kn,U_nm, U_mn
      real, dimension(:), allocatable :: Ny ,b_deim
      real, dimension(:,:), allocatable :: A_deim    !() name change
      integer :: unodes,d1, AI,AJ,AM,AN,d2   
      
      !petro
      type(scalar_field), pointer :: POD_u_scalar  !petro
      type(pod_rhs_type)::pod_rhs_old
      type(vector_field), pointer :: POD_u
      real, dimension(:), allocatable :: theta_pet,smean_gi,KB_pod,b_pod,PSI_GI,PSI_OLD_GI,psi_old,GRADX,GRADT,DIFFGI_pod,PSI
      real, dimension(:,:), allocatable ::leftsvd,leftsvd_gi,leftsvd_x_gi, AMAT_pod, KMAT_pod
      
      !real, dimension(:,:), allocatable :: snapmean_velocity
      type(element_type), pointer :: x_shape,xf_shape
      real, dimension(:,:),allocatable ::X_val !for getting the result of detwei
      real, dimension(:), allocatable :: detwei(:),res(:)
      real ,dimension(:,:),allocatable :: JJ   !(X%dim, mesh_dim(X)),
      real :: det ! for getting the result of detwei
      real, dimension(:,:,:), allocatable ::  dshape
      
      ! real, dimension(POD_u%dim,ele_loc(POD_u,ele)) :: X_val !for getting the result of detwei
      ! new added for petro-galerkin
      integer :: TOTELE,NLOC,NGI,nsvd !,stat
      integer ::  POD_num
      type(mesh_type), pointer :: pod_umesh
      REAL PSI_THETA
      INTEGER ELE,globi,globj,isvd,jsvd,jloc,GI 
      REAL A_STAR_COEF,AX_STAR,P_STAR_POD
      REAL DT1
      real noloc,nogas
      logical :: petrov
      real, dimension(:,:), allocatable :: pod_matrix_snapmean,  pod_matrix_adv_mean
      real, dimension(:), allocatable :: pod_rhs_snapmean
      real, dimension(:,:), allocatable :: pod_matrix_perturbed,pod_matrix_adv_perturbed
      real, dimension(:), allocatable :: pod_rhs_perturbed,pod_ct_rhs,pod_rhs_adv_perturbed
      
      !nonlinear_iteration_loop
      integer :: nonlinear_iterations
      integer, intent(in) :: its
      
      !free surface matrix
      type(csr_matrix) :: fs_m
      logical :: on_sphere, have_absorption, have_vertical_stabilization
      type(vector_field), pointer :: dummy_absorption
      logical :: lump_mass_form
      real :: finish_time,current_time
      integer :: total_timestep
      ! Adjoint model
      logical adjoint_reduced
      type(pod_rhs_type) :: pod_rhs_adjoint
      type(pod_matrix_type) :: adjoint_pod_A,adjoint_pod_B,adjoint_A_extra,adjoint_A
      real, dimension(:,:), allocatable :: pod_coef_all,pod_coef_adjoint ! solution of adjoint

      ! gradient:
      real, dimension(:), allocatable :: g,ds,ds_tmp
      
      !  else !do adjoint_reduced
      call get_option("/timestepping/current_time", current_time)
      call get_option("/timestepping/finish_time", finish_time)       
      call get_option("/timestepping/timestep", dt)
      call get_option(trim(u%option_path)//"/prognostic/temporal_discretisation/theta", &
           theta)
      total_timestep=(finish_time-current_time)/dt
      call allocate(adjoint_pod_A, POD_velocity, POD_pressure)
      call allocate(adjoint_A_extra, POD_velocity, POD_pressure) 
      call allocate(adjoint_A, POD_velocity, POD_pressure)
      call allocate(adjoint_pod_B, POD_velocity, POD_pressure)
      allocate(pod_coef_all(total_timestep,((u%dim+1)*size(POD_state,1))))
      allocate(pod_coef_adjoint(total_timestep,((u%dim+1)*size(POD_state,1))))
      read(100,*)((pod_coef_all(i,j),j=1,(u%dim+1)*size(POD_state,1)),i=0,total_timestep)
      open(unit=20,file='pod_matrix_snapmean')
      read(20,*)((pod_matrix_snapmean(i,j),j=1,(u%dim+1)*size(POD_state,1)),i=1,(u%dim+1)*size(POD_state,1))
      close(20)
      
      if(timestep.eq.1) then  !!! do we run the adjoint from timestep = 1 or total_timestep??
         allocate(g((u%dim+1)*size(POD_state,1))) !!  where we deallocate it?
      endif
      allocate(ds((u%dim+1)*size(POD_state,1)))
      allocate(ds_tmp((u%dim+1)*size(POD_state,1)))

      g = 0.0
      ds = 0.0
      ds_tmp = 0.0

      pod_matrix%val=pod_matrix_snapmean(:,:)
      pod_matrix_B%val=0.0
      
      open(1,file='coef_pod_all_obv')
      read(1,*)((pod_coef_adjoint(i,j),j=1,(u%dim+1)*size(POD_state,1)),i=0,total_timestep)
      ! pod_coef_adjoint() save the coef_pod_all_obv temporarily
      close(1)
      pod_coef_adjoint(:,:)=pod_coef_adjoint(:,:)-pod_coef_all(:,:)
      open(30,file='advection_matrix_perturbed')
      adjoint_pod_A%val =0.0
      adjoint_pod_B%val =0.0
      do k=1,size(POD_state,1)
         do d=1, u%dim
            !adjoint_pod_A%val size same as pod_matrix. read pod_matrix_perturbed here including theta*dt
            read(30,*)((pod_matrix_perturbed(i,j),j=1,(u%dim+1)*size(POD_state,1)),i=1,(u%dim+1)*size(POD_state,1))
            ! delete theta*dt from pod_matrix_perturbed
            pod_matrix_perturbed = pod_matrix_perturbed/(theta*dt)
            pod_matrix%val=pod_matrix%val+  &
                 theta*pod_coef_all(timestep,k+(d-1)*size(POD_state,1))*pod_matrix_perturbed(:,:)
            pod_matrix_B%val = pod_matrix_B%val + &
                 (1.-theta)*pod_coef_all(timestep,k+(d-1)*size(POD_state,1))*pod_matrix_perturbed(:,:)
            adjoint_pod_A%val((d-1)*size(POD_state,1)+k,:)= &
                 theta*matmul(pod_matrix_perturbed(:,:),pod_coef_all(timestep+1, :))
            ! theta*matmul(pod_matrix_perturbed(:,:),pod_coef_all(timestep+1, k+(d-1)*size(POD_state,1)))
            adjoint_pod_B%val((d-1)*size(POD_state,1)+k,:)= &
                 (1.-theta)*matmul(pod_matrix_perturbed(:,:),pod_coef_all(timestep, :))
            !(1.-theta)*matmul(pod_matrix_perturbed(:,:),pod_coef_all(timestep, k+(d-1)*size(POD_state,1)))
            !   pod_matrix%val=pod_matrix%val+pod_coef_all(timestep, k+(d-1)*size(POD_state,1))*pod_matrix_perturbed(:,:)
            
         enddo
         
         read(30,*)((pod_matrix_perturbed(i,j),j=1,(u%dim+1)*size(POD_state,1)),i=1,(u%dim+1)*size(POD_state,1))

         pod_matrix%val=pod_matrix%val+theta*pod_coef_all(timestep,k+u%dim*size(POD_state,1))*pod_matrix_perturbed(:,:)
         pod_matrix_B%val = pod_matrix_B%val + (1.-theta)*pod_coef_all(timestep,k+u%dim*size(POD_state,1))*pod_matrix_perturbed(:,:)

         adjoint_pod_A%val((d-1)*size(POD_state,1)+k,:)=&
              theta*matmul(pod_matrix_perturbed(:,:),pod_coef_all(timestep+1, :))
         !theta*matmul(pod_matrix_perturbed(:,:),pod_coef_all(timestep+1, k+(d-1)*size(POD_state,1)))
         adjoint_pod_B%val((d-1)*size(POD_state,1)+k,:)=&
              (1.-theta)*matmul(pod_matrix_perturbed(:,:),pod_coef_all(timestep, :))
         !(1.-theta)*matmul(pod_matrix_perturbed(:,:),pod_coef_all(timestep, k+(d-1)*size(POD_state,1)))
         !   pod_matrix%val=pod_matrix%val+pod_coef_all(timestep, k+(d-1)*size(POD_state,1))*pod_matrix_perturbed(:,:)
      enddo
      
      close(30)
      
      if(timestep.eq.total_timestep) then
         adjoint_A_extra%val=adjoint_pod_A%val      
         adjoint_A_extra%val=adjoint_pod_A%val + pod_matrix_B%val
         adjoint_A_extra%val=transpose(adjoint_A_extra%val)
         adjoint_A%val=transpose(pod_matrix%val)     
         pod_rhs_adjoint%val=matmul(adjoint_A_extra%val,pod_rhs_adjoint%val)
      else
         adjoint_A_extra%val=adjoint_pod_A%val+ adjoint_pod_B%val        
         adjoint_A_extra%val=adjoint_pod_A%val+ adjoint_pod_B%val + pod_matrix_B%val
         adjoint_A_extra%val=transpose(adjoint_A_extra%val)
         adjoint_A%val=transpose(pod_matrix%val)     
         pod_rhs_adjoint%val=matmul(adjoint_A_extra%val,pod_rhs_adjoint%val)    
      endif
      call solve(adjoint_A%val,pod_rhs_adjoint%val)
      pod_coef_adjoint(timestep,:)=pod_rhs_adjoint%val(:)
      if(timestep.eq.total_timestep) then
         ds(:) = matmul(adjoint_pod_B%val, pod_coef_all(1, :) )  !!!Is pod_coef_all(1, :) the initial one??
         g = ds
         do i = 1,size(g)
            ds = 0.0
            ds(i) = 1.0
            ds_tmp = matmul(adjoint_pod_B%val,ds)
            g(i)= g(i) + dot_product(pod_rhs_adjoint%val,ds_tmp)
         enddo
      endif
      
      call deallocate(adjoint_pod_A)
      call deallocate(adjoint_A_extra)
      call deallocate(adjoint_A)
      call deallocate(adjoint_pod_B)
      deallocate(pod_coef_all)
      deallocate(pod_coef_adjoint)
      deallocate(ds)
      deallocate(ds_tmp)
      ! endif !end if(.not.adjoint_reduced)
      
    end subroutine solve_momentum_reduced_adjoint
    
  end module momentum_equation_reduced_adjoint