!------------------------------------------------------------------------------- !VERS ! This file is part of Code_Saturne, a general-purpose CFD tool. ! ! Copyright (C) 1998-2021 EDF S.A. ! ! This program is free software; you can redistribute it and/or modify it under ! the terms of the GNU General Public License as published by the Free Software ! Foundation; either version 2 of the License, or (at your option) any later ! version. ! ! This program 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 General Public License for more ! details. ! ! You should have received a copy of the GNU General Public License along with ! this program; if not, write to the Free Software Foundation, Inc., 51 Franklin ! Street, Fifth Floor, Boston, MA 02110-1301, USA. !------------------------------------------------------------------------------- !============================================! !== Pinzo@15102024: Fortran module for ANN variables == module ANNvariables use, intrinsic :: iso_c_binding implicit none ! Scalar fields plus velocity for ANN input ! Dimension - > n_features, n_samples, n_snapshots double precision, dimension(:,:,:), pointer, save :: inputANN double precision, dimension(:,:,:), pointer, save :: inputANN_b ! Setting up output to the function double precision, dimension(:), pointer :: zANN_o ! Setting up constant parameter integer, parameter :: num_samples = 81897, num_features = 11, num_snapshots = 50 integer, parameter :: num_features_out = 1, num_samples_b = 55708, num_features_b = 4 ! Variables to collect ANN processing real (c_float), target :: output(num_samples) real (c_float), target :: input_1(num_features, num_samples) real (c_float), target :: input_0(num_features, num_samples) real (c_float), target :: input_2(num_features, num_samples) real (c_float), target :: input_3(num_features, num_samples) real (c_float), target :: input_4(num_features, num_samples) real (c_float), target :: input_5(num_features, num_samples) real (c_float), target :: input_6(num_features, num_samples) real (c_float), target :: input_7(num_features, num_samples) real (c_float), target :: input_8(num_features, num_samples) real (c_float), target :: input_9(num_features, num_samples) real (c_float), target :: input_10(num_features, num_samples) real (c_float), target :: input_11(num_features, num_samples) real (c_float), target :: input_12(num_features, num_samples) real (c_float), target :: input_13(num_features, num_samples) real (c_float), target :: input_14(num_features, num_samples) real (c_float), target :: input_15(num_features, num_samples) real (c_float), target :: input_16(num_features, num_samples) real (c_float), target :: input_17(num_features, num_samples) real (c_float), target :: input_18(num_features, num_samples) real (c_float), target :: input_19(num_features, num_samples) real (c_float), target :: input_20(num_features, num_samples) real (c_float), target :: input_21(num_features, num_samples) real (c_float), target :: input_22(num_features, num_samples) real (c_float), target :: input_23(num_features, num_samples) real (c_float), target :: input_24(num_features, num_samples) real (c_float), target :: input_25(num_features, num_samples) real (c_float), target :: input_26(num_features, num_samples) real (c_float), target :: input_27(num_features, num_samples) real (c_float), target :: input_28(num_features, num_samples) real (c_float), target :: input_29(num_features, num_samples) real (c_float), target :: input_30(num_features, num_samples) real (c_float), target :: input_31(num_features, num_samples) real (c_float), target :: input_32(num_features, num_samples) real (c_float), target :: input_33(num_features, num_samples) real (c_float), target :: input_40(num_features, num_samples) real (c_float), target :: input_41(num_features, num_samples) real (c_float), target :: input_42(num_features, num_samples) real (c_float), target :: input_43(num_features, num_samples) real (c_float), target :: input_44(num_features, num_samples) real (c_float), target :: input_45(num_features, num_samples) real (c_float), target :: input_46(num_features, num_samples) real (c_float), target :: input_47(num_features, num_samples) real (c_float), target :: input_48(num_features, num_samples) real (c_float), target :: input_49(num_features, num_samples) real (c_float), target :: input_1_b(num_features_b, num_samples) real (c_float), target :: input_0_b(num_features_b, num_samples) real (c_float), target :: input_2_b(num_features_b, num_samples) real (c_float), target :: input_3_b(num_features_b, num_samples) real (c_float), target :: input_4_b(num_features_b, num_samples) real (c_float), target :: input_5_b(num_features_b, num_samples) real (c_float), target :: input_6_b(num_features_b, num_samples) real (c_float), target :: input_7_b(num_features_b, num_samples) real (c_float), target :: input_8_b(num_features_b, num_samples) real (c_float), target :: input_9_b(num_features_b, num_samples) real (c_float), target :: input_10_b(num_features_b, num_samples) real (c_float), target :: input_11_b(num_features_b, num_samples) real (c_float), target :: input_12_b(num_features_b, num_samples) real (c_float), target :: input_13_b(num_features_b, num_samples) real (c_float), target :: input_14_b(num_features_b, num_samples) real (c_float), target :: input_15_b(num_features_b, num_samples) real (c_float), target :: input_16_b(num_features_b, num_samples) real (c_float), target :: input_17_b(num_features_b, num_samples) real (c_float), target :: input_18_b(num_features_b, num_samples) real (c_float), target :: input_19_b(num_features_b, num_samples) real (c_float), target :: input_20_b(num_features_b, num_samples) real (c_float), target :: input_21_b(num_features_b, num_samples) real (c_float), target :: input_22_b(num_features_b, num_samples) real (c_float), target :: input_23_b(num_features_b, num_samples) real (c_float), target :: input_24_b(num_features_b, num_samples) real (c_float), target :: input_25_b(num_features_b, num_samples) real (c_float), target :: input_26_b(num_features_b, num_samples) real (c_float), target :: input_27_b(num_features_b, num_samples) real (c_float), target :: input_28_b(num_features_b, num_samples) real (c_float), target :: input_29_b(num_features_b, num_samples) real (c_float), target :: input_30_b(num_features_b, num_samples) real (c_float), target :: input_31_b(num_features_b, num_samples) real (c_float), target :: input_32_b(num_features_b, num_samples) real (c_float), target :: input_33_b(num_features_b, num_samples) real (c_float), target :: input_40_b(num_features_b, num_samples) real (c_float), target :: input_41_b(num_features_b, num_samples) real (c_float), target :: input_42_b(num_features_b, num_samples) real (c_float), target :: input_43_b(num_features_b, num_samples) real (c_float), target :: input_44_b(num_features_b, num_samples) real (c_float), target :: input_45_b(num_features_b, num_samples) real (c_float), target :: input_46_b(num_features_b, num_samples) real (c_float), target :: input_47_b(num_features_b, num_samples) real (c_float), target :: input_48_b(num_features_b, num_samples) real (c_float), target :: input_49_b(num_features_b, num_samples) contains subroutine deallocate_var() deallocate(inputANN, inputANN_b) end subroutine end module ANNvariables !============================================! !== Pinzo@250902024: Fortran module for ANN prediction == module ANNtorch use, intrinsic :: iso_c_binding implicit none private interface ! Pinzo@25092024: Subroutine for ANN predict subroutine ANNpredict(output, input_0, input_1, input_2, input_3, input_4, & input_5, input_6, input_7, input_8, input_9, input_10, & input_11, input_12, input_13, input_14, input_15, input_16, & input_17, input_18, input_19, input_20, input_21, input_22, & input_23, input_24, input_25, input_26, input_27, input_28, & input_29, input_30, input_31, input_32, input_33, input_34, & input_35, input_36, input_37, input_38, input_39, input_40, & input_41, input_42, input_43, input_44, input_45, input_46, & input_47, input_48, input_49, & input_0_b, input_1_b, input_2_b, input_3_b, input_4_b, & input_5_b, input_6_b, input_7_b, input_8_b, input_9_b, input_10_b, & input_11_b, input_12_b, input_13_b, input_14_b, input_15_b, input_16_b, & input_17_b, input_18_b, input_19_b, input_20_b, input_21_b, input_22_b, & input_23_b, input_24_b, input_25_b, input_26_b, input_27_b, input_28_b, & input_29_b, input_30_b, input_31_b, input_32_b, input_33_b, input_34_b, & input_35_b, input_36_b, input_37_b, input_38_b, input_39_b, input_40_b, & input_41_b, input_42_b, input_43_b, input_44_b, input_45_b, input_46_b, & input_47_b, input_48_b, input_49_b) bind(C, name="ANNpredict") use, intrinsic :: iso_c_binding implicit none type(c_ptr), value :: output type(c_ptr), value :: input_0 type(c_ptr), value :: input_1 type(c_ptr), value :: input_2 type(c_ptr), value :: input_3 type(c_ptr), value :: input_4 type(c_ptr), value :: input_5 type(c_ptr), value :: input_6 type(c_ptr), value :: input_7 type(c_ptr), value :: input_8 type(c_ptr), value :: input_9 type(c_ptr), value :: input_10 type(c_ptr), value :: input_11 type(c_ptr), value :: input_12 type(c_ptr), value :: input_13 type(c_ptr), value :: input_14 type(c_ptr), value :: input_15 type(c_ptr), value :: input_16 type(c_ptr), value :: input_17 type(c_ptr), value :: input_18 type(c_ptr), value :: input_19 type(c_ptr), value :: input_20 type(c_ptr), value :: input_21 type(c_ptr), value :: input_22 type(c_ptr), value :: input_23 type(c_ptr), value :: input_24 type(c_ptr), value :: input_25 type(c_ptr), value :: input_26 type(c_ptr), value :: input_27 type(c_ptr), value :: input_28 type(c_ptr), value :: input_29 type(c_ptr), value :: input_30 type(c_ptr), value :: input_31 type(c_ptr), value :: input_32 type(c_ptr), value :: input_33 type(c_ptr), value :: input_34 type(c_ptr), value :: input_35 type(c_ptr), value :: input_36 type(c_ptr), value :: input_37 type(c_ptr), value :: input_38 type(c_ptr), value :: input_39 type(c_ptr), value :: input_40 type(c_ptr), value :: input_41 type(c_ptr), value :: input_42 type(c_ptr), value :: input_43 type(c_ptr), value :: input_44 type(c_ptr), value :: input_45 type(c_ptr), value :: input_46 type(c_ptr), value :: input_47 type(c_ptr), value :: input_48 type(c_ptr), value :: input_49 type(c_ptr), value :: input_0_b type(c_ptr), value :: input_1_b type(c_ptr), value :: input_2_b type(c_ptr), value :: input_3_b type(c_ptr), value :: input_4_b type(c_ptr), value :: input_5_b type(c_ptr), value :: input_6_b type(c_ptr), value :: input_7_b type(c_ptr), value :: input_8_b type(c_ptr), value :: input_9_b type(c_ptr), value :: input_10_b type(c_ptr), value :: input_11_b type(c_ptr), value :: input_12_b type(c_ptr), value :: input_13_b type(c_ptr), value :: input_14_b type(c_ptr), value :: input_15_b type(c_ptr), value :: input_16_b type(c_ptr), value :: input_17_b type(c_ptr), value :: input_18_b type(c_ptr), value :: input_19_b type(c_ptr), value :: input_20_b type(c_ptr), value :: input_21_b type(c_ptr), value :: input_22_b type(c_ptr), value :: input_23_b type(c_ptr), value :: input_24_b type(c_ptr), value :: input_25_b type(c_ptr), value :: input_26_b type(c_ptr), value :: input_27_b type(c_ptr), value :: input_28_b type(c_ptr), value :: input_29_b type(c_ptr), value :: input_30_b type(c_ptr), value :: input_31_b type(c_ptr), value :: input_32_b type(c_ptr), value :: input_33_b type(c_ptr), value :: input_34_b type(c_ptr), value :: input_35_b type(c_ptr), value :: input_36_b type(c_ptr), value :: input_37_b type(c_ptr), value :: input_38_b type(c_ptr), value :: input_39_b type(c_ptr), value :: input_40_b type(c_ptr), value :: input_41_b type(c_ptr), value :: input_42_b type(c_ptr), value :: input_43_b type(c_ptr), value :: input_44_b type(c_ptr), value :: input_45_b type(c_ptr), value :: input_46_b type(c_ptr), value :: input_47_b type(c_ptr), value :: input_48_b type(c_ptr), value :: input_49_b end subroutine ANNpredict end interface public :: ANNpredict end module ANNtorch !== (END)Pinzo@250902024: Fortran module for ANN prediction == !============================================! module collect use pointe use numvar use ppincl use optcal use field use entsor use ANNvariables use, intrinsic :: iso_c_binding implicit none double precision, dimension(:,:), pointer :: cvar_vel, bpro_forbr double precision, dimension(:), pointer :: cpro_visct, cvar_pr, cpro_temp double precision, dimension(:), pointer :: bpro_normstress, cvar_z contains subroutine collect_data(inputANN, inputANN_b, alpha) double precision, dimension(:,:,:), pointer :: inputANN, inputANN_b integer :: iel, icg, alpha integer, parameter :: ncel = 81897, ncompvec = 3, ncel_boundary = 55708 call field_get_val_s(ivarfl(ipr), cvar_pr) call field_get_val_s(ivarfl(isca(iscalt)), cpro_temp) call field_get_val_v(ivarfl(iu), cvar_vel) call field_get_val_s(ivisct, cpro_visct) call field_get_val_s_by_name("stress_normal", bpro_normstress) call field_get_val_s_by_name("Z", cvar_z) call field_get_val_v(iforbr, bpro_forbr) ! boundary stress do iel = 1, ncel inputANN(1, iel, alpha) = cvar_pr(iel) inputANN(2, iel, alpha) = cpro_temp(iel) inputANN(3, iel, alpha) = cpro_visct(iel) inputANN(4, iel, alpha) = cvar_z(iel) do icg = 1, ncompvec inputANN(4+icg, iel, alpha) = cvar_vel(icg,iel) enddo enddo do iel = 1, ncel_boundary inputANN_b(1, iel, alpha) = bpro_normstress(iel) do icg = 1, ncompvec inputANN_b(1+icg, iel, alpha) = bpro_forbr(icg, iel) enddo enddo end subroutine collect_data end module collect !============================================! !== Pinzo@26092024: Camille's code == subroutine cs_f_user_extra_operations & ( nvar , nscal , & dt ) use paramx use cstphy use cstnum use parall use period use ppppar use ppthch use mesh use lagran use cs_c_bindings use coincl use radiat use dimens, only: ndimfb use field_operator use ANNtorch use collect implicit none integer :: nvar, nscal, icg, iel, i double precision :: dt(ncelet) ! reading csv character(len=1000) :: row_csv integer :: j, ios real(c_float) :: data_cs(num_samples, 10) character(len=100) :: filename character(len=100), dimension(1) :: filename_list integer counterANN data counterANN /0/ save counterANN integer time_step data time_step /0/ save time_step integer alpha data alpha /0/ save alpha ! Pinzo@26092024: Initialize variable for ANN activation double precision, dimension(:), pointer :: activeANN logical ANNactivated ! clock variables real :: start, end, elapsed_time call cpu_time(start) call field_get_val_s_by_name("activeANN", activeANN) ANNactivated = .FALSE. ! update for print purposes time_step = time_step + 1 if (counterANN.eq.0) allocate(inputANN(num_samples, num_features, num_snapshots), & inputANN_b(num_samples_b, num_features_b, num_snapshots)) if (time_step.ge.0) then ! Pinzo@15102024: Adding data collection (every 1E-5 s) if (mod(counterANN, 10) == 0) then write(nfecra, *) ' Pinzo cs_user_physical_properties.f90: Collecting data at ', time_step+3000, & 'with counter', counterANN ! t0 if (counterANN.lt.10) then write(nfecra, *) ' *t', 0 alpha = alpha + 1 call collect_data(inputANN, inputANN_b, alpha) ! t1 -> t49 else if (counterANN.eq.alpha*10) then write(nfecra, *) ' *t', alpha alpha = alpha + 1 call collect_data(inputANN, inputANN_b, alpha) if (alpha.eq.50) activeANN(:) = 1 endif endif ! update counter counterANN = counterANN + 1 endif ! activating ANN in all cells if (activeANN(1).ge.1) ANNactivated = .TRUE. ! Pinzo@21102024: Overwriting cs variables at time step following ANN update if (counterANN.ge.502) then do iel = 1, ncel ! reset ANN activeANN(iel) = 0 enddo ! reset iteration ANNactivated = .false. counterANN = -1 ! for debug endif if (ANNactivated) then write(nfecra,*) ' Pinzo cs_user_physical_properties.f90: ANN upload' !call to ANNpredict call ANNpredict(c_loc(output), c_loc(inputANN(:,:,1)), c_loc(inputANN(:,:,2)), & c_loc(inputANN(:,:,3)), c_loc(inputANN(:,:,4)), c_loc(inputANN(:,:,5)), c_loc(inputANN(:,:,6)), & c_loc(inputANN(:,:,7)), c_loc(inputANN(:,:,8)), c_loc(inputANN(:,:,9)), c_loc(inputANN(:,:,10)), & c_loc(inputANN(:,:,11)), c_loc(inputANN(:,:,12)), c_loc(inputANN(:,:,13)), c_loc(inputANN(:,:,14)), & c_loc(inputANN(:,:,15)), c_loc(inputANN(:,:,16)), c_loc(inputANN(:,:,17)), c_loc(inputANN(:,:,18)), & c_loc(inputANN(:,:,19)), c_loc(inputANN(:,:,20)), c_loc(inputANN(:,:,21)), c_loc(inputANN(:,:,22)), & c_loc(inputANN(:,:,23)), c_loc(inputANN(:,:,24)), c_loc(inputANN(:,:,25)), c_loc(inputANN(:,:,26)), & c_loc(inputANN(:,:,27)), c_loc(inputANN(:,:,28)), c_loc(inputANN(:,:,29)), c_loc(inputANN(:,:,30)), & c_loc(inputANN(:,:,31)), c_loc(inputANN(:,:,32)), c_loc(inputANN(:,:,33)), c_loc(inputANN(:,:,34)), & c_loc(inputANN(:,:,35)), c_loc(inputANN(:,:,36)), c_loc(inputANN(:,:,37)), c_loc(inputANN(:,:,38)), & c_loc(inputANN(:,:,39)), c_loc(inputANN(:,:,40)), c_loc(inputANN(:,:,41)), c_loc(inputANN(:,:,42)), & c_loc(inputANN(:,:,43)), c_loc(inputANN(:,:,44)), c_loc(inputANN(:,:,45)), c_loc(inputANN(:,:,46)), & c_loc(inputANN(:,:,47)), c_loc(inputANN(:,:,48)), c_loc(inputANN(:,:,49)), c_loc(inputANN(:,:,50)), & c_loc(inputANN_b(:,:,1)), c_loc(inputANN_b(:,:,2)), & c_loc(inputANN_b(:,:,3)), c_loc(inputANN_b(:,:,4)), c_loc(inputANN_b(:,:,5)), & c_loc(inputANN_b(:,:,6)), c_loc(inputANN_b(:,:,7)), c_loc(inputANN_b(:,:,8)), & c_loc(inputANN_b(:,:,9)), c_loc(inputANN_b(:,:,10)), c_loc(inputANN_b(:,:,11)), & c_loc(inputANN_b(:,:,12)), c_loc(inputANN_b(:,:,13)), c_loc(inputANN_b(:,:,14)), & c_loc(inputANN_b(:,:,15)), c_loc(inputANN_b(:,:,16)), c_loc(inputANN_b(:,:,17)), & c_loc(inputANN_b(:,:,18)), c_loc(inputANN_b(:,:,19)), c_loc(inputANN_b(:,:,20)), & c_loc(inputANN_b(:,:,21)), c_loc(inputANN_b(:,:,22)), c_loc(inputANN_b(:,:,23)), & c_loc(inputANN_b(:,:,24)), c_loc(inputANN_b(:,:,25)), c_loc(inputANN_b(:,:,26)), & c_loc(inputANN_b(:,:,27)), c_loc(inputANN_b(:,:,28)), c_loc(inputANN_b(:,:,29)), & c_loc(inputANN_b(:,:,30)), c_loc(inputANN_b(:,:,31)), c_loc(inputANN_b(:,:,32)), & c_loc(inputANN_b(:,:,33)), c_loc(inputANN_b(:,:,34)), c_loc(inputANN_b(:,:,35)), & c_loc(inputANN_b(:,:,36)), c_loc(inputANN_b(:,:,37)), c_loc(inputANN_b(:,:,38)), & c_loc(inputANN_b(:,:,39)), c_loc(inputANN_b(:,:,40)), c_loc(inputANN_b(:,:,41)), & c_loc(inputANN_b(:,:,42)), c_loc(inputANN_b(:,:,43)), c_loc(inputANN_b(:,:,44)), & c_loc(inputANN_b(:,:,45)), c_loc(inputANN_b(:,:,46)), c_loc(inputANN_b(:,:,47)), & c_loc(inputANN_b(:,:,48)), c_loc(inputANN_b(:,:,49)), c_loc(inputANN_b(:,:,50))) ! ! Debug: calling fields from standard simulation ! if (time_step.ge.141) then ! if (time_step.lt.150) then ! filename_list(1) = "/home/cenvinzf@coria.fr/Desktop/test_cs_coupling/t15.csv" ! else if (time_step.ge.150.and.time_step.lt.300) then ! filename_list(1) = "/home/cenvinzf@coria.fr/Desktop/test_cs_coupling/t30.csv" ! else if (time_step.ge.300.and.time_step.lt.450) then ! filename_list(1) = "/home/cenvinzf@coria.fr/Desktop/test_cs_coupling/t45.csv" ! ! debug ! activeANN(:) = 0 ! endif ! ! open file (reading only) ! open(unit=10, file=filename_list(1), status="old", action="read", iostat=ios) ! if (ios /= 0) then ! print *, "Errore nell'apertura del file." ! stop ! end if ! do i = 1, num_samples ! read(10, '(A)', iostat=ios) row_csv ! if (ios /= 0) then ! print *, "Errore nella lettura della riga ", i ! exit ! end if ! read(row_csv, *) (data_cs(i, j), j=1,10) ! end do ! close(10) ! ! (END) open file ! endif ! Pinzo@21102024: Commented because overwrite of variables moved in d3pint.f90 ! Overwriting cs arrays ! ! debug ! output(:,:) = time_step ! activeANN(:) = 0 call field_get_val_s_by_name("zANN_o", zANN_o) zANN_o(:) = output(:) endif call cpu_time(end) elapsed_time = end - start write(nfecra,*) ' Pinzo cs_user_physical_properties.f90: Time elapsed in routine:', elapsed_time return end subroutine cs_f_user_extra_operations !=============================================================================== !> \brief Modify turbulent viscosity !> !> This subroutine is called at beginning of each time step !> after the computation of the turbulent viscosity !> (physical quantities have already been computed in \ref usphyv). !> !> Turbulent viscosity \f$ \mu_T \f$ (kg/(m s)) can be modified. !> !> A modification of the turbulent viscosity can lead to very !> significant differences between solutions and even give wrong !> results. !> !> This subroutine is therefore reserved to expert users. ! !------------------------------------------------------------------------------- !------------------------------------------------------------------------------- ! Arguments !______________________________________________________________________________. ! mode name role ! !______________________________________________________________________________! !> \param[in] nvar total number of variables !> \param[in] nscal total number of scalars !> \param[in] ncepdp number of cells with head loss !> \param[in] ncesmp number of cells with mass source term !> \param[in] icepdc head loss cell numbering !> \param[in] icetsm numbering of cells with mass source term !> \param[in] itypsm kind of mass source for each variable !> \param[in] dt time step (per cell) !> \param[in] ckupdc work array for head loss terms !> \param[in] smacel values of variables related to mass source !> term. If ivar=ipr, smacel=mass flux !_______________________________________________________________________________ subroutine usvist & ( nvar , nscal , ncepdp , ncesmp , & icepdc , icetsm , itypsm , & dt , & ckupdc , smacel ) !=============================================================================== !=============================================================================== ! Module files !=============================================================================== use paramx use numvar use optcal use cstphy use entsor use parall use period use mesh use field use field_operator !=============================================================================== implicit none ! Arguments integer nvar , nscal integer ncepdp , ncesmp integer icepdc(ncepdp) integer icetsm(ncesmp), itypsm(ncesmp,nvar) double precision dt(ncelet) double precision ckupdc(6,ncepdp), smacel(ncesmp,nvar) !=============================================================================== !-------- ! Formats !-------- !---- ! End !---- return end subroutine usvist