diff --git a/matlab/load_marconi.m b/matlab/load_marconi.m
new file mode 100644
index 0000000000000000000000000000000000000000..a8e687dbdca443c5d18faba5aac0137be5152fb6
--- /dev/null
+++ b/matlab/load_marconi.m
@@ -0,0 +1,10 @@
+function [ RESDIR ] = load_marconi( outfilename )
+%UNTITLED2 Summary of this function goes here
+% Detailed explanation goes here
+ hostfolder = outfilename(1:end-8);
+ hostfile = [hostfolder,'/outputs*'];
+ RESDIR = ['../',outfilename(46:end-8),'/'];
+ localfolder = [RESDIR,'.'];
+ system(['scp -r ahoffman@login.marconi.cineca.it:',hostfile,' ',localfolder])
+end
+
diff --git a/matlab/load_params.m b/matlab/load_params.m
new file mode 100644
index 0000000000000000000000000000000000000000..d1eee972ddda5a92ec3e73fa1f2c45a59c2ed955
--- /dev/null
+++ b/matlab/load_params.m
@@ -0,0 +1,18 @@
+CONAME = h5readatt(filename,'/data/input','CO');
+if CONAME == -1
+ CONAME = 'FC';
+elseif CONAME == 0
+ CONAME = 'LB';
+elseif CONAME == 1
+ CONAME = 'DG';
+end
+ETAB = h5readatt(filename,'/data/input','eta_B');
+ETAN = h5readatt(filename,'/data/input','eta_n');
+ETAT = h5readatt(filename,'/data/input','eta_T');
+
+NON_LIN = h5readatt(filename,'/data/input','NON_LIN');
+if NON_LIN == 'y'
+ NON_LIN = 1;
+else
+ NON_LIN = 0;
+end
diff --git a/matlab/write_sbash.m b/matlab/write_sbash.m
index 7590654231b6c6421f1d12b600ee86371fd7cff2..ba785d725051314556609535cb7cbec36faa1728 100644
--- a/matlab/write_sbash.m
+++ b/matlab/write_sbash.m
@@ -25,6 +25,7 @@ fid = fopen(INPUT,'wt');
fprintf(fid,[...
'#!/bin/bash\n',...
+'#SBATCH --job-name=',CLUSTER.JNAME,'\n',...
'#SBATCH --time=', CLUSTER.TIME,'\n',...
'#SBATCH --nodes=', CLUSTER.NODES,'\n',...
'#SBATCH --cpus-per-task=', CLUSTER.CPUPT,'\n',...
diff --git a/src/auxval.F90 b/src/auxval.F90
index 0ef6890bd1b1f0ec7cf68fa0a247f6098f56f582..07f7acde91e4074425cbd213350ed23a7b5f5391 100644
--- a/src/auxval.F90
+++ b/src/auxval.F90
@@ -4,11 +4,11 @@ subroutine auxval
USE basic
USE grid
USE array
- USE fourier, ONLY: init_grid_distr_and_plans
+ USE fourier, ONLY: init_grid_distr_and_plans, alloc_local_1, alloc_local_2
use prec_const
IMPLICIT NONE
- INTEGER :: irows,irowe, irow, icol
+ INTEGER :: irows,irowe, irow, icol, i_
IF (my_id .EQ. 0) WRITE(*,*) '=== Set auxiliary values ==='
CALL init_grid_distr_and_plans(Nr,Nz)
@@ -21,4 +21,12 @@ subroutine auxval
CALL memory ! Allocate memory for global arrays
+ DO i_ = 0,num_procs-1
+ CALL mpi_barrier(MPI_COMM_WORLD, ierr)
+ IF (my_id .EQ. i_) WRITE (*,'(I2,A9,I3,A8,I3,A8,I3,A8,I3,A15,I6)') &
+ i_,': ikrs = ', ikrs, ' ikre = ', ikre, 'ikzs = ', ikzs, ' ikze = ', ikze, &
+ 'alloc_local = ', alloc_local_1+alloc_local_2
+ CALL mpi_barrier(MPI_COMM_WORLD, ierr)
+ ENDDO
+
END SUBROUTINE auxval
diff --git a/src/compute_Sapj_old.F90 b/src/compute_Sapj_old.F90
deleted file mode 100644
index b4d02513667ab91c92554c6c62f5bc8d89a95336..0000000000000000000000000000000000000000
--- a/src/compute_Sapj_old.F90
+++ /dev/null
@@ -1,207 +0,0 @@
-SUBROUTINE compute_Sapj
- ! This routine is meant to compute the non linear term for each specie and degree
- !! In real space Sapj ~ b*(grad(phi) x grad(g)) which in moments in fourier becomes
- !! Sapj = Sum_n (ikr Kn phi)#(ikz Sum_s d_njs Naps) - (ikz Kn phi)#(ikr Sum_s d_njs Naps)
- !! where # denotes the convolution.
- USE array, ONLY : dnjs, Sepj, Sipj
- USE basic
- USE fourier
- USE fields!, ONLY : phi, moments_e, moments_i
- USE grid
- USE model
- USE prec_const
- USE time_integration!, ONLY : updatetlevel
- IMPLICIT NONE
-
- COMPLEX(dp), DIMENSION(ikrs:ikre,ikzs:ikze) :: F_, G_, CONV
- INTEGER :: in, is
- REAL(dp):: kr, kz, kerneln, be_2, bi_2, factn
- REAL(dp):: sigmae2_taue_o2, sigmai2_taui_o2
-
- ! Execution time start
- CALL cpu_time(t0_Sapj)
-
- !!!!!!!!!!!!!!!!!!!! ELECTRON non linear term computation (Sepj)!!!!!!!!!!
- sigmae2_taue_o2 = sigma_e**2 * tau_e/2._dp ! factor of the kerneln argument
-
- ploope: DO ip = ips_e,ipe_e ! Loop over Hermite moments
- jloope: DO ij = ijs_e, ije_e ! Loop over Laguerre moments
- Sepj(ip,ij,:,:) = 0._dp
- factn = 1
-
- nloope: DO in = 1,jmaxe+1 ! Loop over laguerre for the sum
-
- krloope1: DO ikr = ikrs,ikre ! Loop over kr
- kzloope1: DO ikz = ikzs,ikze ! Loop over kz
- kr = krarray(ikr)
- kz = kzarray(ikz)
- be_2 = sigmae2_taue_o2 * (kr**2 + kz**2)
- kerneln = be_2**(in-1)/factn * EXP(-be_2)
- ! First convolution term
- IF ( NON_LIN ) THEN
- F_(ikr,ikz) = imagu*kr* phi(ikr,ikz) * kerneln
- ELSE
- F_(ikr,ikz) = 0._dp
- ENDIF
- ! Background sinusoidal electrostatic potential phi_0 for KH inst.
- IF ( q_e .NE. 0._dp ) THEN ! If electron are not removed
- IF ( kz .EQ. 0._dp ) THEN ! Kronecker kz=0
- IF ( ABS(kr) .EQ. ABS(kr0KH) ) THEN ! kronecker kr=kr0
- F_(ikr,ikz) = -A0KH/2._dp + F_(ikr,ikz)
- ENDIF
- ENDIF
- ENDIF
- ! Second convolution term
- G_(ikr,ikz) = 0._dp ! initialization of the sum
- DO is = 1, MIN( in+ij-1, jmaxe+1 ) ! sum truncation on number of moments
- G_(ikr,ikz) = G_(ikr,ikz) + &
- dnjs(in,ij,is) * moments_e(ip,is,ikr,ikz,updatetlevel)
- ENDDO
- G_(ikr,ikz) = imagu*kz*G_(ikr,ikz)
-
- ENDDO kzloope1
- ENDDO krloope1
-
- CALL convolve_2D_F2F( F_, G_, CONV ) ! Convolve Fourier to Fourier
-
- DO ikr = ikrs,ikre ! Loop over kr
- DO ikz = ikzs,ikze ! Loop over kz
- Sepj(ip,ij,ikr,ikz) = Sepj(ip,ij,ikr,ikz) + CONV(ikr,ikz) ! Add it to Sepj
- ENDDO
- ENDDO
-
- IF ( NON_LIN ) THEN ! Fully non linear term ikz phi * ikr Napj
- krloope2: DO ikr = ikrs,ikre ! Loop over kr
- kzloope2: DO ikz = ikzs,ikze ! Loop over kz
- kr = krarray(ikr)
- kz = kzarray(ikz)
- be_2 = sigmae2_taue_o2 * (kr**2 + kz**2)
- kerneln = be_2**(in-1)/factn * EXP(-be_2)
- ! First convolution term
- F_(ikr,ikz) = imagu*kz* phi(ikr,ikz) * kerneln
- ! Second convolution term
- G_(ikr,ikz) = 0._dp ! initialization of the sum
- DO is = 1, MIN( in+ij-1, jmaxe+1 ) ! sum truncation on number of moments
- G_(ikr,ikz) = G_(ikr,ikz) + &
- dnjs(in,ij,is) * moments_e(ip,is,ikr,ikz,updatetlevel)
- ENDDO
- G_(ikr,ikz) = imagu*kr*G_(ikr,ikz)
-
- ENDDO kzloope2
- ENDDO krloope2
-
- CALL convolve_2D_F2F( F_, G_, CONV )
-
- DO ikr = ikrs,ikre
- DO ikz = ikzs,ikze
- Sepj(ip,ij,ikr,ikz) = Sepj(ip,ij,ikr,ikz) - CONV(ikr,ikz) ! substract it to Sepj
- ENDDO
- ENDDO
-
- ENDIF
-
- IF ( in + 1 .LE. jmaxe+1 ) THEN
- factn = real(in,dp) * factn ! compute (n+1)!
- ENDIF
- ENDDO nloope
-
- ENDDO jloope
- ENDDO ploope
- !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
-
- !!!!!!!!!!!!!!!!!!!! ION non linear term computation (Sipj)!!!!!!!!!!
- sigmai2_taui_o2 = sigma_i**2 * tau_i/2._dp
-
- ploopi: DO ip = ips_i,ipe_i ! Loop over Hermite moments
- jloopi: DO ij = ijs_i, ije_i ! Loop over Laguerre moments
- Sipj(ip,ij,:,:) = 0._dp
- factn = 1
-
- nloopi: DO in = 1,jmaxi+1 ! Loop over laguerre for the sum
-
- krloopi1: DO ikr = ikrs,ikre ! Loop over kr
- kzloopi1: DO ikz = ikzs,ikze ! Loop over kz
- kr = krarray(ikr)
- kz = kzarray(ikz)
- bi_2 = sigmai2_taui_o2 * (kr**2 + kz**2)
- kerneln = bi_2**(in-1)/factn * EXP(-bi_2)
-
- ! First convolution term
- IF ( NON_LIN ) THEN
- F_(ikr,ikz) = imagu*kr* phi(ikr,ikz) * kerneln
- ELSE
- F_(ikr,ikz) = 0._dp
- ENDIF
- ! Background sinusoidal electrostatic potential phi_0 for KH inst.
- IF ( kz .EQ. 0._dp ) THEN ! Kronecker kz=0
- IF ( ABS(kr) .EQ. ABS(kr0KH) ) THEN ! kronecker kr=kr0
- F_(ikr,ikz) = -A0KH/2._dp + F_(ikr,ikz)
- ENDIF
- ENDIF
- ! Second convolution term
- G_(ikr,ikz) = 0._dp ! initialization of the sum
- DO is = 1, MIN( in+ij-1, jmaxi+1 )
- G_(ikr,ikz) = G_(ikr,ikz) + &
- dnjs(in,ij,is) * moments_i(ip,is,ikr,ikz,updatetlevel)
- ENDDO
- G_(ikr,ikz) = imagu*kz*G_(ikr,ikz)
-
- ENDDO kzloopi1
- ENDDO krloopi1
-
- CALL convolve_2D_F2F( F_, G_, CONV ) ! Convolve Fourier to Fourier
- DO ikr = ikrs,ikre
- DO ikz = ikzs,ikze
- Sipj(ip,ij,ikr,ikz) = Sipj(ip,ij,ikr,ikz) + CONV(ikr,ikz) ! add it to Sipj
- ENDDO
- ENDDO
-
- krloopi2: DO ikr = ikrs,ikre ! Loop over kr
- kzloopi2: DO ikz = ikzs,ikze ! Loop over kz
- kr = krarray(ikr)
- kz = kzarray(ikz)
- bi_2 = sigmai2_taui_o2 * (kr**2 + kz**2)
- kerneln = bi_2**(in-1)/factn * EXP(-bi_2)
- ! First convolution term
- F_(ikr,ikz) = imagu*kz*phi(ikr,ikz) * kerneln
- ! Second convolution term
- G_(ikr,ikz) = 0._dp ! initialization of the sum
- IF ( NON_LIN ) THEN
- DO is = 1, MIN( in+ij-1, jmaxi+1 )
- G_(ikr,ikz) = G_(ikr,ikz) + &
- dnjs(in,ij,is) * moments_i(ip,is,ikr,ikz,updatetlevel)
- ENDDO
- G_(ikr,ikz) = imagu*kr*G_(ikr,ikz)
- ENDIF
- ! Background sinusoidal ion denstiy n_i0 for KH inst.
- IF ( kz .EQ. 0._dp ) THEN ! Kronecker kz=0
- IF ( ABS(kr) .EQ. ABS(kr0KH) ) THEN ! kronecker kr=+-kr0
- G_(ikr,ikz) = -A0KH*kr0KH**2/2._dp + G_(ikr,ikz)
- ENDIF
- ENDIF
-
- ENDDO kzloopi2
- ENDDO krloopi2
-
- CALL convolve_2D_F2F( F_, G_, CONV ) ! Convolve and back to Fourier
- DO ikr = ikrs,ikre
- DO ikz = ikzs,ikze
- Sipj(ip,ij,ikr,ikz) = Sipj(ip,ij,ikr,ikz) - CONV(ikr,ikz) ! substract it to Sepj
- ENDDO
- ENDDO
-
- IF ( in + 1 .LE. jmaxi+1 ) THEN
- factn = real(in,dp) * factn ! factorial(n+1)
- ENDIF
-
- ENDDO nloopi
-
- ENDDO jloopi
- ENDDO ploopi
- !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
-
- ! Execution time end
- CALL cpu_time(t1_Sapj)
- tc_Sapj = tc_Sapj + (t1_Sapj - t0_Sapj)
-
-END SUBROUTINE compute_Sapj
diff --git a/src/compute_Sapj_opt.F90 b/src/compute_Sapj_opt.F90
deleted file mode 100644
index 3fed22876f9a3fd3e060899d62049b95eea37be5..0000000000000000000000000000000000000000
--- a/src/compute_Sapj_opt.F90
+++ /dev/null
@@ -1,189 +0,0 @@
-SUBROUTINE compute_Sapj
- ! This routine is meant to compute the non linear term for each specie and degree
- !! In real space Sapj ~ b*(grad(phi) x grad(g)) which in moments in fourier becomes
- !! Sapj = Sum_n (ikr Kn phi)#(ikz Sum_s d_njs Naps) - (ikz Kn phi)#(ikr Sum_s d_njs Naps)
- !! where # denotes the convolution.
- USE array, ONLY : dnjs, Sepj, Sipj
- USE basic
- USE fourier
- USE fields!, ONLY : phi, moments_e, moments_i
- USE grid
- USE model
- USE prec_const
- USE time_integration!, ONLY : updatetlevel
- IMPLICIT NONE
- INCLUDE 'fftw3-mpi.f03'
-
- COMPLEX(dp), DIMENSION(ikrs:ikre,ikzs:ikze) :: Fr_cmpx, Gz_cmpx
- COMPLEX(dp), DIMENSION(ikrs:ikre,ikzs:ikze) :: Fz_cmpx, Gr_cmpx, F_conv_G
- REAL(dp), DIMENSION(irs:ire,izs:ize) :: fr_real, gz_real
- REAL(dp), DIMENSION(irs:ire,izs:ize) :: fz_real, gr_real, f_times_g
-
- INTEGER :: in, is
- REAL(dp):: kr, kz, kerneln, be_2, bi_2, factn
- REAL(dp):: sigmae2_taue_o2, sigmai2_taui_o2
-
- ! Execution time start
- CALL cpu_time(t0_Sapj)
-
- !!!!!!!!!!!!!!!!!!!! ELECTRON non linear term computation (Sepj)!!!!!!!!!!
- sigmae2_taue_o2 = sigma_e**2 * tau_e/2._dp ! factor of the kerneln argument
-
- ploope: DO ip = ips_e,ipe_e ! Loop over Hermite moments
- jloope: DO ij = ijs_e, ije_e ! Loop over Laguerre moments
- real_data_c = 0._dp ! initialize sum over real nonlinear term
- factn = 1
-
- nloope: DO in = 1,jmaxe+1 ! Loop over laguerre for the sum
-
- krloope: DO ikr = ikrs,ikre ! Loop over kr
- kzloope: DO ikz = ikzs,ikze ! Loop over kz
- kr = krarray(ikr)
- kz = kzarray(ikz)
- be_2 = sigmae2_taue_o2 * (kr**2 + kz**2)
- kerneln = be_2**(in-1)/factn * EXP(-be_2)
- ! First convolution terms
- Fr_cmpx(ikr,ikz) = imagu*kr* phi(ikr,ikz) * kerneln
- Fz_cmpx(ikr,ikz) = imagu*kz* phi(ikr,ikz) * kerneln
- ! Second convolution terms
- Gz_cmpx(ikr,ikz) = 0._dp ! initialization of the sum
- Gr_cmpx(ikr,ikz) = 0._dp ! initialization of the sum
- DO is = 1, MIN( in+ij-1, jmaxe+1 ) ! sum truncation on number of moments
- Gz_cmpx(ikr,ikz) = Gz_cmpx(ikr,ikz) + &
- dnjs(in,ij,is) * moments_e(ip,is,ikr,ikz,updatetlevel)
- Gr_cmpx(ikr,ikz) = Gr_cmpx(ikr,ikz) + &
- dnjs(in,ij,is) * moments_e(ip,is,ikr,ikz,updatetlevel)
- ENDDO
- Gz_cmpx(ikr,ikz) = imagu*kz*Gz_cmpx(ikr,ikz)
- Gr_cmpx(ikr,ikz) = imagu*kr*Gr_cmpx(ikr,ikz)
- ENDDO kzloope
- ENDDO krloope
-
- ! First term drphi x dzf
- DO ikr = ikrs, ikre
- DO ikz = ikzs, ikze
- cmpx_data_f(ikz,ikr-local_nkr_offset) = Fr_cmpx(ikr,ikz)
- cmpx_data_g(ikz,ikr-local_nkr_offset) = Gz_cmpx(ikr,ikz)
- ENDDO
- ENDDO
-
- call fftw_mpi_execute_dft_c2r(planb, cmpx_data_f, real_data_f)
- call fftw_mpi_execute_dft_c2r(planb, cmpx_data_g, real_data_g)
-
- real_data_c = real_data_c + real_data_f/Nz/Nr * real_data_g/Nz/Nr
-
- ! Second term -dzphi x drf
- DO ikr = ikrs, ikre
- DO ikz = ikzs, ikze
- cmpx_data_f(ikz,ikr-local_nkr_offset) = Fz_cmpx(ikr,ikz)
- cmpx_data_g(ikz,ikr-local_nkr_offset) = Gr_cmpx(ikr,ikz)
- ENDDO
- ENDDO
-
- call fftw_mpi_execute_dft_c2r(planb, cmpx_data_f, real_data_f)
- call fftw_mpi_execute_dft_c2r(planb, cmpx_data_g, real_data_g)
-
- real_data_c = real_data_c - real_data_f/Nz/Nr * real_data_g/Nz/Nr
-
- IF ( in + 1 .LE. jmaxe+1 ) THEN
- factn = real(in,dp) * factn ! compute (n+1)!
- ENDIF
- ENDDO nloope
-
- ! Put the real nonlinear product into k-space
- call fftw_mpi_execute_dft_r2c(planf, real_data_c, cmpx_data_c)
-
- ! Retrieve convolution in input format
- DO ikr = ikrs, ikre
- DO ikz = ikzs, ikze
- Sepj(ip,ij,ikr,ikz) = cmpx_data_c(ikz,ikr-local_nkr_offset)*AA_r(ikr)*AA_z(ikz)
- ENDDO
- ENDDO
-
- ENDDO jloope
- ENDDO ploope
- !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
-
- !!!!!!!!!!!!!!!!!!!! ION non linear term computation (Sipj)!!!!!!!!!!
- sigmai2_taui_o2 = sigma_i**2 * tau_i/2._dp ! factor of the kerneln argument
-
- ploopi: DO ip = ips_e,ipe_e ! Loop over Hermite moments
- jloopi: DO ij = ijs_e, ije_e ! Loop over Laguerre moments
- real_data_c = 0._dp ! initialize sum over real nonlinear term
- factn = 1
-
- nloopi: DO in = 1,jmaxi+1 ! Loop over laguerre for the sum
-
- krloopi: DO ikr = ikrs,ikre ! Loop over kr
- kzloopi: DO ikz = ikzs,ikze ! Loop over kz
- kr = krarray(ikr)
- kz = kzarray(ikz)
- bi_2 = sigmai2_taui_o2 * (kr**2 + kz**2)
- kerneln = bi_2**(in-1)/factn * EXP(-bi_2)
- ! First convolution terms
- Fr_cmpx(ikr,ikz) = imagu*kr* phi(ikr,ikz) * kerneln
- Fz_cmpx(ikr,ikz) = imagu*kz* phi(ikr,ikz) * kerneln
- ! Second convolution terms
- Gz_cmpx(ikr,ikz) = 0._dp ! initialization of the sum
- Gr_cmpx(ikr,ikz) = 0._dp ! initialization of the sum
- DO is = 1, MIN( in+ij-1, jmaxi+1 ) ! sum truncation on number of moments
- Gz_cmpx(ikr,ikz) = Gz_cmpx(ikr,ikz) + &
- dnjs(in,ij,is) * moments_i(ip,is,ikr,ikz,updatetlevel)
- Gr_cmpx(ikr,ikz) = Gr_cmpx(ikr,ikz) + &
- dnjs(in,ij,is) * moments_i(ip,is,ikr,ikz,updatetlevel)
- ENDDO
- Gz_cmpx(ikr,ikz) = imagu*kz*Gz_cmpx(ikr,ikz)
- Gr_cmpx(ikr,ikz) = imagu*kr*Gr_cmpx(ikr,ikz)
- ENDDO kzloopi
- ENDDO krloopi
-
- ! First term drphi x dzf
- DO ikr = ikrs, ikre
- DO ikz = ikzs, ikze
- cmpx_data_f(ikz,ikr-local_nkr_offset) = Fr_cmpx(ikr,ikz)
- cmpx_data_g(ikz,ikr-local_nkr_offset) = Gz_cmpx(ikr,ikz)
- ENDDO
- ENDDO
-
- call fftw_mpi_execute_dft_c2r(planb, cmpx_data_f, real_data_f)
- call fftw_mpi_execute_dft_c2r(planb, cmpx_data_g, real_data_g)
-
- real_data_c = real_data_c + real_data_f/Nz/Nr * real_data_g/Nz/Nr
-
- ! Second term -dzphi x drf
- DO ikr = ikrs, ikre
- DO ikz = ikzs, ikze
- cmpx_data_f(ikz,ikr-local_nkr_offset) = Fz_cmpx(ikr,ikz)
- cmpx_data_g(ikz,ikr-local_nkr_offset) = Gr_cmpx(ikr,ikz)
- ENDDO
- ENDDO
-
- call fftw_mpi_execute_dft_c2r(planb, cmpx_data_f, real_data_f)
- call fftw_mpi_execute_dft_c2r(planb, cmpx_data_g, real_data_g)
-
- real_data_c = real_data_c - real_data_f/Nz/Nr * real_data_g/Nz/Nr
-
- IF ( in + 1 .LE. jmaxe+1 ) THEN
- factn = real(in,dp) * factn ! compute (n+1)!
- ENDIF
- ENDDO nloopi
-
- ! Put the real nonlinear product into k-space
- call fftw_mpi_execute_dft_r2c(planf, real_data_c, cmpx_data_c)
-
- ! Retrieve convolution in input format
- DO ikr = ikrs, ikre
- DO ikz = ikzs, ikze
- Sipj(ip,ij,ikr,ikz) = cmpx_data_c(ikz,ikr-local_nkr_offset)*AA_r(ikr)*AA_z(ikz)
- ENDDO
- ENDDO
-
- ENDDO jloopi
- ENDDO ploopi
- !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
-
- ! Execution time END
- CALL cpu_time(t1_Sapj)
- tc_Sapj = tc_Sapj + (t1_Sapj - t0_Sapj)
-
-END SUBROUTINE compute_Sapj
diff --git a/src/fourier_mod.F90 b/src/fourier_mod.F90
index f1d4eb1bd7b806416d57e317869f76c4cf2bdbe2..528e21ddede3301b8f2b3a8b4baa928f19c2b2c9 100644
--- a/src/fourier_mod.F90
+++ b/src/fourier_mod.F90
@@ -18,7 +18,8 @@ MODULE fourier
type(C_PTR) :: cdatar_f, cdatar_g, cdatar_c
type(C_PTR) :: cdatac_f, cdatac_g, cdatac_c
type(C_PTR) , PUBLIC :: planf, planb
- integer(C_INTPTR_T) :: i, ix, iy, alloc_local_1, alloc_local_2
+ integer(C_INTPTR_T) :: i, ix, iy
+ integer(C_INTPTR_T), PUBLIC :: alloc_local_1, alloc_local_2
integer(C_INTPTR_T) :: NR_, NZ_
CONTAINS
diff --git a/src/grid_mod.F90 b/src/grid_mod.F90
index 77b6119bc79e41cffe5104f5e311abc2cc4045c2..8f5be3dfa155bf88a363ed3100a3c52a28e18984 100644
--- a/src/grid_mod.F90
+++ b/src/grid_mod.F90
@@ -94,12 +94,6 @@ CONTAINS
ikrs = local_nkr_offset + 1
ikre = ikrs + local_nkr - 1
- DO i_ = 0,num_procs-1
- CALL mpi_barrier(MPI_COMM_WORLD, ierr)
- IF (my_id .EQ. i_) print *, i_,': ikrs = ', ikrs, ' ikre = ', ikre
- CALL mpi_barrier(MPI_COMM_WORLD, ierr)
- ENDDO
-
IF (my_id .EQ. num_procs-1) ikre = Nkr
!WRITE(*,*) 'ID = ',my_id,' ikrs = ', ikrs, ' ikre = ', ikre
! Grid spacings
diff --git a/wk/analysis_2D.m b/wk/analysis_2D.m
index dc5198904d5581f68e0b5ac458cb358db1a61010..4e01df312d990de8bb22a7dddf1d8e7eec44e60c 100644
--- a/wk/analysis_2D.m
+++ b/wk/analysis_2D.m
@@ -1,12 +1,14 @@
%% Load results
-if LOAD_MARCONI==1
- hostfolder = ['/marconi_scratch/userexternal/ahoffman/HeLaZ/results/',BASIC.SIMID,'/',BASIC.PARAMS];
- localfolder= [BASIC.RESDIR,'..'];
- system(['scp -r ahoffman@login.marconi.cineca.it:',hostfolder,' ',localfolder])
+if 0
+ %%
+ outfile = '/marconi_scratch/userexternal/ahoffman/HeLaZ/results/Marconi/512x256_L_100_Pe_6_Je_3_Pi_6_Ji_3_nB_0.8_nN_1_nu_1e-01_FC_mu_5e-04/out.txt';
+ BASIC.RESDIR = load_marconi(outfile);
end
+%%
% JOBNUM = 0; load_results;
% JOBNUM = 1; load_results;
compile_results
+load_params
%% Retrieving max polynomial degree and sampling info
Npe = numel(Pe); Nje = numel(Je); [JE,PE] = meshgrid(Je,Pe);
@@ -16,10 +18,8 @@ Ns2D = numel(Ts2D);
% renaming and reshaping quantity of interest
Ts5D = Ts5D';
Ts2D = Ts2D';
-if strcmp(OUTPUTS.write_non_lin,'.true.')
- Si00 = squeeze(Sipj(1,1,:,:,:));
- Se00 = squeeze(Sepj(1,1,:,:,:));
-end
+Si00 = squeeze(Sipj(1,1,:,:,:));
+Se00 = squeeze(Sepj(1,1,:,:,:));
%% Build grids
Nkr = numel(kr); Nkz = numel(kz);
[KZ,KR] = meshgrid(kz,kr);
@@ -53,11 +53,10 @@ for it = 1:numel(Ts2D)
dzphi(:,:,it) = real(fftshift(ifft2(1i*KZ.*(PH_),Nr,Nz)));
end
-if strcmp(OUTPUTS.write_non_lin,'.true.')
for it = 1:numel(Ts5D)
si00(:,:,it) = real(fftshift(ifft2(squeeze(Si00(:,:,it)),Nr,Nz)));
end
-end
+
% Post processing
disp('- post processing')
E_pot = zeros(1,Ns2D); % Potential energy n^2
@@ -69,12 +68,11 @@ Flux_re = zeros(1,Ns2D); % Particle flux Gamma = <ne drphi>
Flux_ze = zeros(1,Ns2D); % Particle flux Gamma = <ne dzphi>
Ne_norm = zeros(Npe,Nje,Ns5D);% Time evol. of the norm of Napj
Ni_norm = zeros(Npi,Nji,Ns5D);% .
-if strcmp(OUTPUTS.write_non_lin,'.true.')
Se_norm = zeros(Npe,Nje,Ns5D);% Time evol. of the norm of Sapj
Si_norm = zeros(Npi,Nji,Ns5D);% .
Sne00_norm = zeros(1,Ns2D); % Time evol. of the amp of e nonlin term
Sni00_norm = zeros(1,Ns2D); %
-end
+
Ddr = 1i*KR; Ddz = 1i*KZ; lapl = Ddr.^2 + Ddz.^2;
@@ -96,14 +94,11 @@ dEdt = diff(E_pot+E_kin)./dt2D;
for it = 1:numel(Ts5D) % Loop over 5D arrays
Ne_norm(:,:,it)= sum(sum(abs(Nepj(:,:,:,:,it)),3),4)/Nkr/Nkz;
Ni_norm(:,:,it)= sum(sum(abs(Nipj(:,:,:,:,it)),3),4)/Nkr/Nkz;
-if strcmp(OUTPUTS.write_non_lin,'.true.')
Se_norm(:,:,it)= sum(sum(abs(Sepj(:,:,:,:,it)),3),4)/Nkr/Nkz;
Si_norm(:,:,it)= sum(sum(abs(Sipj(:,:,:,:,it)),3),4)/Nkr/Nkz;
Sne00_norm(it) = sum(sum(abs(Se00(:,:,it))))/Nkr/Nkz;
Sni00_norm(it) = sum(sum(abs(Si00(:,:,it))))/Nkr/Nkz;
end
-end
-
%% Compute growth rate
if NON_LIN == 0
@@ -241,7 +236,7 @@ save_figure
end
%%
-t0 = 0;
+t0 = 1;
skip_ = 1;
DELAY = 0.01*skip_;
FRAMES = floor(t0/(Ts2D(2)-Ts2D(1)))+1:skip_:numel(Ts2D);
@@ -310,56 +305,29 @@ fig = figure; FIGNAME = 'space_time_drphi';set(gcf, 'Position', [100, 100, 1200
save_figure
end
-if 0
-%% Flow
-skip = 10;
-plt = @(x) x(1:skip:end,1:skip:end);
-tf = 120; [~,it] = min(abs(Ts2D-tf));
-
-fig = figure; FIGNAME = 'space_time_drphi';
- quiver(plt(RR),plt(ZZ),plt(dzphi(:,:,it)),plt(-drphi(:,:,it)),0);
- xlabel('$r$'), ylabel('$z$')
- title(sprintf('$v_E$, $t=%.0f c_s/R$',Ts2D(it)));
-end
-
-if 0
-%% Growth rate
-fig = figure; FIGNAME = ['growth_rate',sprintf('_%.2d',JOBNUM)];
- [~,ikr0KH] = min(abs(kr-KR0KH));
- plot(kz/kr(ikr0KH),gkr0kz_Ni00/(kr(ikr0KH)*A0KH),'DisplayName',['J = ',num2str(JMAXI)])
- xlabel('$k_z/k_{r0}$'); ylabel('$\gamma_{Ni}/(A_0k_{r0})$');
- xlim([0. 1.0]); ylim([0. 0.6]);
-save_figure
-
-end
%%
if 0
%% Mode time evolution
+[~,ikr ] = min(abs(kr-dkr));
[~,ik00] = min(abs(kz));
[~,idk] = min(abs(kz-dkz));
-[~,ik50] = min(abs(kz-0.5*KR0KH));
-[~,ik75] = min(abs(kz-0.7*KR0KH));
-[~,ik10] = min(abs(kz-1.00*KR0KH));
+[~,ik50] = min(abs(kz-0.1*max(kz)));
+[~,ik75] = min(abs(kz-0.2*max(kz)));
+[~,ik10] = min(abs(kz-0.3*max(kz)));
plt = @(x) abs(squeeze(x));
fig = figure; FIGNAME = ['mode_time_evolution',sprintf('_%.2d',JOBNUM)];
- semilogy(Ts2D,plt(Ni00(ikr0KH,ik00,:)),'DisplayName', ...
- ['$k_z/k_{r0} = $',num2str(kz(ik00)/KR0KH)]); hold on
- semilogy(Ts2D,plt(Ni00(ikr0KH,idk,:)),'DisplayName', ...
- ['$k_z/k_{r0} = $',num2str(kz(idk)/KR0KH)]); hold on
- semilogy(Ts2D,plt(Ni00(ikr0KH,ik50,:)),'DisplayName', ...
- ['$k_z/k_{r0} = $',num2str(kz(ik50)/KR0KH)]); hold on
- semilogy(Ts2D,plt(Ni00(ikr0KH,ik75,:)),'DisplayName', ...
- ['$k_z/k_{r0} = $',num2str(kz(ik75)/KR0KH)]); hold on
- semilogy(Ts2D,plt(Ni00(ikr0KH,ik10,:)),'DisplayName', ...
- ['$k_z/k_{r0} = $',num2str(kz(ik10)/KR0KH)]); hold on
+ semilogy(Ts2D,plt(Ni00(ikr,ik00,:)),'DisplayName', ...
+ ['$k_z = $',num2str(kz(ik00))]); hold on
+ semilogy(Ts2D,plt(Ni00(ikr,idk,:)),'DisplayName', ...
+ ['$k_z = $',num2str(kz(idk))]); hold on
+ semilogy(Ts2D,plt(Ni00(ikr,ik50,:)),'DisplayName', ...
+ ['$k_z = $',num2str(kz(ik50))]); hold on
+ semilogy(Ts2D,plt(Ni00(ikr,ik75,:)),'DisplayName', ...
+ ['$k_z = $',num2str(kz(ik75))]); hold on
+ semilogy(Ts2D,plt(Ni00(ikr,ik10,:)),'DisplayName', ...
+ ['$k_z = $',num2str(kz(ik10))]); hold on
xlabel('$t$'); ylabel('$\hat n_i^{00}$'); legend('show');
- semilogy(Ts2D,plt(Ni00(ikr0KH,ik00,end))*exp(gkr0kz_Ni00(ik00)*(Ts2D-Ts2D(end))),'k--')
- semilogy(Ts2D,plt(Ni00(ikr0KH,idk,end))*exp(gkr0kz_Ni00(idk)*(Ts2D-Ts2D(end))),'k--')
- semilogy(Ts2D,plt(Ni00(ikr0KH,ik50,end))*exp(gkr0kz_Ni00(ik50)*(Ts2D-Ts2D(end))),'k--')
- semilogy(Ts2D,plt(Ni00(ikr0KH,ik75,end))*exp(gkr0kz_Ni00(ik75)*(Ts2D-Ts2D(end))),'k--')
- semilogy(Ts2D,plt(Ni00(ikr0KH,ik10,end))*exp(gkr0kz_Ni00(ik10)*(Ts2D-Ts2D(end))),'k--')
- plot(tstart*[1 1],ylim,'k-','LineWidth',0.5);
- plot(tend*[1 1],ylim,'k-','LineWidth',0.5);
+title(sprintf('$k_r=$ %1.1f',kr(ikr)))
save_figure
end
@@ -384,3 +352,37 @@ if strcmp(OUTPUTS.write_non_lin,'.true.')
end
save_figure
end
+
+%% Phase space distribution function
+% M_ = 25;
+% spar = linspace(0,4,M_); xperp = spar;
+%
+% PSDF_e = zeros(numel(kr),numel(kz),M_,M_,numel(Ts5D));
+% for ikr = 1:numel(kr)
+% for ikz = 1:numel(kz)
+% for it = 1:numel(Ts5D)
+% PSDF_e(ikr,ikz,:,:,it) = compute_fa(Nepj(:,:,ikr,ikz,it), spar, xperp);
+% end
+% end
+% end
+%
+% ktarget = 0.3*max(kz);
+% ttarget = 310;
+%
+% [~,ik10] = min(abs(kz-ktarget));
+% [~,it] = min(abs(Ts5D-ttarget));
+% if 0
+% %%
+% plt = @(x) real(x(ik10,ik10,:,:,it));
+% pclr = pcolor(spar,xperp,plt(PSDF_e)); set(pclr, 'edgecolor','none'); colorbar;
+% xlabel('$s_\parallel$'); ylabel('$x_\perp$'); title(sprintf('$t c_s/R=%.0f$',Ts5D(it)));
+% legend(['$Re(f_e),k_\perp \approx$',sprintf('%01.0f',norm([kr(ik10),kz(ik10)]))]);
+% end
+% if 0
+% %% Phase space distribution function time evolution
+% GIFNAME = ['f_e',sprintf('_%.2d',JOBNUM)]; INTERP = 1;
+% FIELD = real(ni00+ne00); X = RR; Y = ZZ; T = Ts5D;
+% FIELDNAME = '$n_i-n_e$'; XNAME = '$r\rho_s$'; YNAME = '$z\rho_s$';
+% create_gif
+% end
+
diff --git a/wk/load_results.m b/wk/load_results.m
index 43bdbdeb1d508f8aef6373ca1528731c7b1e45f2..db7c69dc3571dc64f0bd6e82836f209510d61573 100644
--- a/wk/load_results.m
+++ b/wk/load_results.m
@@ -4,10 +4,8 @@ disp(['Loading ',filename])
% Loading from output file
CPUTIME = h5readatt(filename,'/data/input','cpu_time');
DT_SIM = h5readatt(filename,'/data/input','dt');
-if strcmp(OUTPUTS.write_moments,'.true.')
[Nipj, Pi, Ji, kr, kz, Ts5D, dt5D] = load_5D_data(filename, 'moments_i');
[Nepj, Pe, Je ] = load_5D_data(filename, 'moments_e');
-end
[Ni00, kr, kz, Ts2D, dt2D] = load_2D_data(filename, 'Ni00');
Ne00 = load_2D_data(filename, 'Ne00');
%Pi = [0]; Ji = Pi; Pe = Pi; Je = Pi;
@@ -16,7 +14,5 @@ end
% Nipj(1,1,:,:,:) = Ni00; Nepj(1,1,:,:,:) = Ne00;
PHI = load_2D_data(filename, 'phi');
-if strcmp(OUTPUTS.write_non_lin,'.true.')
- Sipj = load_5D_data(filename, 'Sipj');
- Sepj = load_5D_data(filename, 'Sepj');
-end
\ No newline at end of file
+Sipj = load_5D_data(filename, 'Sipj');
+Sepj = load_5D_data(filename, 'Sepj');
diff --git a/wk/marconi_run.m b/wk/marconi_run.m
index 4a2ae9df68cd32a7b06d353f2738715857497c4b..3fc57d14fbc0c9bf0a4804f72a20653b2bcf78bb 100644
--- a/wk/marconi_run.m
+++ b/wk/marconi_run.m
@@ -8,26 +8,27 @@ CLUSTER.TIME = '24:00:00'; % allocation time hh:mm:ss
CLUSTER.NODES = '1'; % MPI process
CLUSTER.CPUPT = '1'; % CPU per task
CLUSTER.NTPN = '20'; % N tasks per node
-CLUSTER.PART = 'prod'; % dbg or prod
-CLUSTER.MEM = '32GB'; % Memory
+CLUSTER.PART = 'prod'; % dbg or prod
+CLUSTER.MEM = '16GB'; % Memory
+CLUSTER.JNAME = 'gamma_inf'; % Job name
%% PHYSICAL PARAMETERS
NU = 1e-1; % Collision frequency
TAU = 1.0; % e/i temperature ratio
-ETAB = 0.4; % Magnetic gradient
+ETAB = 0.66; % Magnetic gradient
ETAN = 1.0; % Density gradient
ETAT = 0.0; % Temperature gradient
MU = 5e-4; % Hyper diffusivity coefficient
NOISE0 = 1.0e-5;
%% GRID PARAMETERS
N = 512; % Frequency gridpoints (Nkr = N/2)
-L = 150; % Size of the squared frequency domain
-PMAXE = 2; % Highest electron Hermite polynomial degree
-JMAXE = 1; % Highest '' Laguerre ''
-PMAXI = 2; % Highest ion Hermite polynomial degree
-JMAXI = 1; % Highest '' Laguerre ''
+L = 100; % Size of the squared frequency domain
+PMAXE = 8; % Highest electron Hermite polynomial degree
+JMAXE = 4; % Highest '' Laguerre ''
+PMAXI = 8; % Highest ion Hermite polynomial degree
+JMAXI = 4; % Highest '' Laguerre ''
%% TIME PARAMETERS
-TMAX = 200; % Maximal time unit
-DT = 5e-3; % Time step
+TMAX = 400; % Maximal time unit
+DT = 1e-2; % Time step
SPS0D = 10; % Sampling per time unit for profiler
SPS2D = 1; % Sampling per time unit for 2D arrays
SPS5D = 1/10; % Sampling per time unit for 5D arrays
diff --git a/wk/marconi_scaling.m b/wk/marconi_scaling.m
index a65e63789f7fd01c817dfa2808d1ea78488e6246..fb087636e48d9402eba222268caac3aead745bc4 100644
--- a/wk/marconi_scaling.m
+++ b/wk/marconi_scaling.m
@@ -5,11 +5,12 @@ addpath(genpath('../matlab')) % ... add
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% CLUSTER PARAMETERS
CLUSTER.TIME = '01:00:00'; % allocation time hh:mm:ss
-CLUSTER.NODES = '01'; % MPI process
+CLUSTER.NODES = '02'; % MPI process
CLUSTER.CPUPT = '01'; % CPU per task
-CLUSTER.NTPN = '24'; % N tasks per node (openMP)
-CLUSTER.PART = 'prod'; % dbg or prod
-CLUSTER.MEM = '16GB'; % Memory
+CLUSTER.NTPN = '14'; % N tasks per node (openMP)
+CLUSTER.PART = 'dbg'; % dbg or prod
+CLUSTER.MEM = '8GB'; % Memory
+CLUSTER.JNAME = 'scaling'; % Job name
%% PHYSICAL PARAMETERS
NU = 1e-1; % Collision frequency
TAU = 1.0; % e/i temperature ratio
@@ -19,14 +20,14 @@ ETAT = 0.0; % Temperature gradient
MU = 0e-4; % Hyper diffusivity coefficient
NOISE0 = 1.0e-5;
%% GRID PARAMETERS
-N = 128; % Frequency gridpoints (Nkr = N/2)
+N = 1024; % Frequency gridpoints (Nkr = N/2)
L = 10; % Size of the squared frequency domain
-PMAXE = 6; % Highest electron Hermite polynomial degree
-JMAXE = 4; % Highest '' Laguerre ''
-PMAXI = 6; % Highest ion Hermite polynomial degree
-JMAXI = 4; % Highest '' Laguerre ''
+PMAXE = 2; % Highest electron Hermite polynomial degree
+JMAXE = 1; % Highest '' Laguerre ''
+PMAXI = 2; % Highest ion Hermite polynomial degree
+JMAXI = 1; % Highest '' Laguerre ''
%% TIME PARAMETERS
-TMAX = 2; % Maximal time unit
+TMAX = 5; % Maximal time unit
DT = 5e-2; % Time step
SPS0D = 1/DT; % Sampling per time unit for profiler
SPS2D = -1; % Sampling per time unit for 2D arrays
@@ -35,7 +36,7 @@ SPSCP = -1; % Sampling per time unit for checkpoints
RESTART = 0; % To restart from last checkpoint
JOB2LOAD= 0;
%% OPTIONS
-SIMID = ['Scaling_np',num2str(CLUSTER.NTPN)]; % Name of the simulation
+SIMID = ['nn',num2str(CLUSTER.NODES),'_np',num2str(CLUSTER.NTPN)]; % Name of the simulation
CO = -2; % Collision operator (0 : L.Bernstein, -1 : Full Coulomb, -2 : Dougherty)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
@@ -53,4 +54,8 @@ setup
write_sbash
+system(['mkdir -p ../results/Scaling']);
+system(['cp -r ../results/',SIMID,' ../results/Scaling/']);
+system(['rm -r ../results/',SIMID]);
+
MARCONI = 1;
diff --git a/wk/parallel_scaling_new.m b/wk/parallel_scaling_new.m
index c9608265e7ed5bdc343f27a279464439d2bce4cb..1ad96fa61aedff5d11c5219dd668749cad167246 100644
--- a/wk/parallel_scaling_new.m
+++ b/wk/parallel_scaling_new.m
@@ -1,68 +1,103 @@
default_plots_options
-NPS = [01 02 04 08 12 16 20 24];
+% NPS = [01 02 04 08 12 16 20 24];
+NPS = [02 04 10];
TIMES = NPS;
if 0
%% Load times
i_ = 1;
+nn = 2;
for np = NPS
- SIM_NAME = sprintf('Scaling_np%02d',np);
+ SIM_NAME = sprintf('nn%02d_np%02d',nn,np)
hostfile = ['/marconi_scratch/userexternal/ahoffman/HeLaZ/results/',SIM_NAME,'/',BASIC.PARAMS];
- localfile= ['../results/',SIM_NAME,'/'];
- system(['scp -r ahoffman@login.marconi.cineca.it:',hostfile,' ',localfile]);
- filename = ['../results/',SIM_NAME,'/',BASIC.PARAMS,'/outputs_00.h5'];
+ localtarget= ['../results/',sprintf('Scaling/nn%02d_np%02d',nn,np),'/'];
+ system(['scp -r ahoffman@login.marconi.cineca.it:',hostfile,' ',localtarget]);
+ filename = ['../results/',sprintf('Scaling/nn%02d_np%02d',nn,np),'/',BASIC.PARAMS,'/outputs_00.h5'];
TIMES(i_) = h5readatt(filename,'/data/input','cpu_time');
i_ = i_ + 1;
end
-TIMES
+disp(PARAMS)
+disp(TIMES')
end
%% Strong scaling measurement (no diagnostics)
% Handwritten results for 512x256, P,J=2,1, Tmax = 10, mu=0, dt = 5e-2
Results_512_21.np = NPS;
-Results_512_21.time = [799 422 206 106 81 72 66 82]; %tmax 10
-% Results_512_21.time = [0162, 0108, 0055, 0032, 0030, 0045, 0061, 0084]; %tmax 2
+Results_512_21.tproc = [857 447 217 107 74 59 46 41]; %Nthreads
+Results_512_21.tnode = [854 442 199 96 75 64 57 55]; %Nnodes
% Handwritten results for 512x256, P,J=3,2, Tmax = 10, mu=0, dt 5e-2
Results_512_32.np = [ 1, 2, 4, 8, 12, 16, 20, 24];
-Results_512_32.time = [2421 1241 0598 0309 0221 0188 0159 0148];
+Results_512_32.tproc = [2546 1290 0630 0316 0220 0177 0143 0125];
+
+% Handwritten results for 512x256, P,J=6,4, Tmax = 5, mu=0, dt 5e-2
+Results_512_64.np = [ 1, 2, 4, 8, 12, 16, 20, 24];
+Results_512_64.tproc = [5801 2974 1467 0746 0507 0408 0320 0279];
% Handwritten results for 1024x512, P,J=2,1, Tmax = 5 dt = 0.05, mu = 0
Results_1024_21.np = [ 1, 2, 4, 8, 12, 16, 20, 24];
-% Results_1024_21.time = [1920, 0000, 0563, 0306, 0247, 0240, 0000, 0000];
-Results_1024_21.time = [1920, 1260, 0556, 0289, 0219, 0189, 0172, 0167];
+Results_1024_21.tproc = [2051, 1296, 0583, 0296, 0205, 0163, 0132, 0115];
+Results_1024_21.tnode = [2052 1327 0511 0231 0157 0109 0099 0083]; %Nnodes
% Handwritten results for 1024x512, P,J=3,2, Tmax = 2 dt = 0.05, mu = 0
Results_1024_32.np = [ 1, 2, 4, 8, 12, 16, 20, 24];
-Results_1024_32.time = [2199, 1424, 0623, 0324, 0243, 0208, 0188, 0180];
+Results_1024_32.tproc = [2248, 1526, 0644, 0326, 0233, 0182, 0148, 0130];
% Handwritten results for 1024x512, P,J=6,4, Tmax = 2 dt = 0.05, mu = 0
Results_1024_64.np = [ 1, 2, 4, 8, 12, 16, 20, 24];
-Results_1024_64.time = [0000, 0000, 0000, 0000, 0000, 0000, 0000, 0000];
+Results_1024_64.tproc = [10330,6548, 2901, 1497, 1001, 0769, 0610, 0520];
%
fig = figure;
-plot(1:24,1:24,'-k','DisplayName','Ideal')
-hold on
-% res = Results_256_21;
-% plot(res.np,res.time(1)./(res.time),'o--','DisplayName','$256\times128$, $P,J=2,1$');
-res = Results_512_21;
-plot(res.np,res.time(1)./(res.time),'v-','DisplayName','$512\times256$, $P,J=2,1$');
-res = Results_512_32;
-plot(res.np,res.time(1)./(res.time),'>-','DisplayName','$512\times256$, $P,J=3,2$');
-res = Results_1024_21;
-plot(res.np,res.time(1)./(res.time),'o-','DisplayName','$1024\times512$, $P,J=2,1$');
-res = Results_1024_32;
-plot(res.np,res.time(1)./(res.time),'s-','DisplayName','$1024\times512$, $P,J=3,2$');xlim([1,max(res.np)]);
-res = Results_1024_64;
-plot(res.np,res.time(1)./(res.time),'d-','DisplayName','$1024\times512$, $P,J=6,4$');xlim([1,max(res.np)]);
-xlabel('$N_p$'); ylabel('speedup')
-xlim([1,24]); ylim([1,24])
-legend('show')
-title('Strong scaling')
-grid on
+ subplot(121)
+ plot(1:24,1:24,'-k','DisplayName','Ideal')
+ hold on
+ res = Results_512_21;
+ plot(res.np,res.tproc(1)./(res.tproc),'v-','Color',line_colors(1,:),...
+ 'DisplayName','$512\times256$, $P,J=2,1$');
+
+ res = Results_512_32;
+ plot(res.np,res.tproc(1)./(res.tproc),'v-','Color',line_colors(2,:),...
+ 'DisplayName','$512\times256$, $P,J=3,2$');
+
+ res = Results_512_64;
+ plot(res.np,res.tproc(1)./(res.tproc),'v-','Color',line_colors(3,:),...
+ 'DisplayName','$512\times256$, $P,J=6,4$');
+
+ res = Results_1024_21;
+ plot(res.np,res.tproc(1)./(res.tproc),'^--','Color',line_colors(1,:),...
+ 'DisplayName','$1024\times512$, $P,J=2,1$');
+
+ res = Results_1024_32;
+ plot(res.np,res.tproc(1)./(res.tproc),'^--','Color',line_colors(2,:),...
+ 'DisplayName','$1024\times512$, $P,J=3,2$');
+
+ res = Results_1024_64;
+ plot(res.np,res.tproc(1)./(res.tproc),'^--','Color',line_colors(3,:),...
+ 'DisplayName','$1024\times512$, $P,J=6,4$');
+
+ xlabel('$N_p$'); ylabel('speedup')
+ xlim([1,24]);
+ legend('show')
+ title('$N_n=01$')
+subplot(122)
+ plot(1:24,1:24,'-k','DisplayName','Ideal')
+ hold on
+ res = Results_512_21;
+ plot(res.np,res.tnode(1)./(res.tnode),'o-','Color',line_colors(1,:),...
+ 'DisplayName','$512\times256$, $P,J=2,1$');
+
+ res = Results_1024_21;
+ plot(res.np,res.tnode(1)./(res.tnode),'o--','Color',line_colors(1,:),...
+ 'DisplayName','$1024\times512$, $P,J=2,1$');
+
+ xlabel('$N_n$'); ylabel('speedup')
+ xlim([1,24]);
+ legend('show')
+ title('$N_p=01$')
+
FIGNAME = '/home/ahoffman/HeLaZ/results/strong_scaling_new.pdf';
saveas(fig,FIGNAME);
diff --git a/wk/profiler.m b/wk/profiler.m
index 4ca0045721d43681bbfcf726c2b66a7a68338484..e578665159ac924839c9c4da56c09a0eb703807c 100644
--- a/wk/profiler.m
+++ b/wk/profiler.m
@@ -1,5 +1,5 @@
%% load profiling
-filename = sprintf([BASIC.RESDIR,'outputs_%.2d.h5'],00);
+% filename = sprintf([BASIC.RESDIR,'outputs_%.2d.h5'],00);
CPUTIME = double(h5readatt(filename,'/data/input','cpu_time'));
DT_SIM = h5readatt(filename,'/data/input','dt');
diff --git a/wk/test_parallel.m b/wk/test_parallel.m
index 4a6afc8c86dcac22e63dba3a124f16a5b1d3e66a..8ae627881de323155b03b22d8b7ee45d1d20b593 100644
--- a/wk/test_parallel.m
+++ b/wk/test_parallel.m
@@ -2,7 +2,7 @@ clear all;
addpath(genpath('../matlab')) % ... add
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% Set Up parameters
-CLUSTER.TIME = '00:00:10'; % allocation time hh:mm:ss
+CLUSTER.TIME = '00:10:00'; % allocation time hh:mm:ss
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% PHYSICAL PARAMETERS
NU = 1e-1; % Collision frequency
@@ -29,18 +29,17 @@ SPSCP = 1/10; % Sampling per time unit for checkpoints
RESTART = 0; % To restart from last checkpoint
JOB2LOAD= 0;
%% OPTIONS
-SIMID = 'test_runtime'; % Name of the simulation
-CO = -1; % Collision operator (0 : L.Bernstein, -1 : Full Coulomb, -2 : Dougherty)
+SIMID = 'debug'; % Name of the simulation
+CO = 0; % Collision operator (0 : L.Bernstein, -1 : Full Coulomb, -2 : Dougherty)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% unused
KR0KH = 0; A0KH = 0; % Background phi mode to drive Ray-Tay inst.
NO_E = 0; % Remove electrons dynamic
-% DK = 0; % Drift kinetic model (put every kernel_n to 0 except n=0 to 1)
KREQ0 = 0; % put kr = 0
KPAR = 0.0; % Parellel wave vector component
LAMBDAD = 0.0;
-NON_LIN = 0 *(1-KREQ0); % activate non-linearity (is cancelled if KREQ0 = 1)
+NON_LIN = 1 *(1-KREQ0); % activate non-linearity (is cancelled if KREQ0 = 1)
LOAD_MARCONI = 0;
setup