删除diff_new.f90中冗余部分，方便后续工作

.gitignore

@@ -1,6 +1,3 @@
 __pycache__
 GW150914
 GW150914-origin
-docs
-*.tmp
-

AMSS_NCKU_ABEtest.py

@@ -0,0 +1,445 @@
+
+##################################################################
+##
+## AMSS-NCKU ABE Test Program (Skip TwoPuncture if data exists)
+## Modified from AMSS_NCKU_Program.py
+## Author: Xiaoqu
+## Modified: 2026/02/01
+##
+##################################################################
+
+
+##################################################################
+
+## Print program introduction
+
+import print_information
+
+print_information.print_program_introduction()
+
+##################################################################
+
+import AMSS_NCKU_Input as input_data
+
+##################################################################
+
+## Create directories to store program run data
+
+import os
+import shutil
+import sys
+import time
+
+## Set the output directory according to the input file
+File_directory = os.path.join(input_data.File_directory)
+
+## Check if output directory exists and if TwoPuncture data is available
+skip_twopuncture = False
+output_directory = os.path.join(File_directory, "AMSS_NCKU_output")
+binary_results_directory = os.path.join(output_directory, input_data.Output_directory)
+
+if os.path.exists(File_directory):
+    print( " Output directory already exists." )
+    print()
+
+    # Check if TwoPuncture initial data files exist
+    if (input_data.Initial_Data_Method == "Ansorg-TwoPuncture"):
+        twopuncture_output = os.path.join(output_directory, "TwoPunctureABE")
+        input_par = os.path.join(output_directory, "input.par")
+
+        if os.path.exists(twopuncture_output) and os.path.exists(input_par):
+            print( " Found existing TwoPuncture initial data." )
+            print( " Do you want to skip TwoPuncture phase and reuse existing data?" )
+            print( " Input 'skip' to skip TwoPuncture and start ABE directly" )
+            print( " Input 'regenerate' to regenerate everything from scratch" )
+            print()
+
+            while True:
+                try:
+                    inputvalue = input()
+                    if ( inputvalue == "skip" ):
+                        print( " Skipping TwoPuncture phase, will reuse existing initial data." )
+                        print()
+                        skip_twopuncture = True
+                        break
+                    elif ( inputvalue == "regenerate" ):
+                        print( " Regenerating everything from scratch." )
+                        print()
+                        skip_twopuncture = False
+                        break
+                    else:
+                        print( " Please input 'skip' or 'regenerate'." )
+                except ValueError:
+                    print( " Please input 'skip' or 'regenerate'." )
+        else:
+            print( " TwoPuncture initial data not found, will regenerate everything." )
+            print()
+
+    # If not skipping, remove and recreate directory
+    if not skip_twopuncture:
+        shutil.rmtree(File_directory, ignore_errors=True)
+        os.mkdir(File_directory)
+        os.mkdir(output_directory)
+        os.mkdir(binary_results_directory)
+        figure_directory = os.path.join(File_directory, "figure")
+        os.mkdir(figure_directory)
+        shutil.copy("AMSS_NCKU_Input.py", File_directory)
+        print( " Output directory has been regenerated." )
+        print()
+else:
+    # Create fresh directory structure
+    os.mkdir(File_directory)
+    shutil.copy("AMSS_NCKU_Input.py", File_directory)
+    os.mkdir(output_directory)
+    os.mkdir(binary_results_directory)
+    figure_directory = os.path.join(File_directory, "figure")
+    os.mkdir(figure_directory)
+    print( " Output directory has been generated." )
+    print()
+
+# Ensure figure directory exists
+figure_directory = os.path.join(File_directory, "figure")
+if not os.path.exists(figure_directory):
+    os.mkdir(figure_directory)
+
+##################################################################
+
+## Output related parameter information
+
+import setup
+
+## Print and save input parameter information
+setup.print_input_data( File_directory )
+
+if not skip_twopuncture:
+    setup.generate_AMSSNCKU_input()
+
+setup.print_puncture_information()
+
+
+##################################################################
+
+## Generate AMSS-NCKU program input files based on the configured parameters
+
+if not skip_twopuncture:
+    print()
+    print( " Generating the AMSS-NCKU input parfile for the ABE executable." )
+    print()
+
+    ## Generate cgh-related input files from the grid information
+
+    import numerical_grid
+
+    numerical_grid.append_AMSSNCKU_cgh_input()
+
+    print()
+    print( " The input parfile for AMSS-NCKU C++ executable file ABE has been generated." )
+    print( " However, the input relevant to TwoPuncture need to be appended later." )
+    print()
+
+
+##################################################################
+
+## Plot the initial grid configuration
+
+if not skip_twopuncture:
+    print()
+    print( " Schematically plot the numerical grid structure." )
+    print()
+
+    import numerical_grid
+    numerical_grid.plot_initial_grid()
+
+
+##################################################################
+
+## Generate AMSS-NCKU macro files according to the numerical scheme and parameters
+
+if not skip_twopuncture:
+    print()
+    print( " Automatically generating the macro file for AMSS-NCKU C++ executable file ABE " )
+    print( " (Based on the finite-difference numerical scheme) " )
+    print()
+
+    import generate_macrodef
+
+    generate_macrodef.generate_macrodef_h()
+    print( " AMSS-NCKU macro file macrodef.h has been generated. " )
+
+    generate_macrodef.generate_macrodef_fh()
+    print( " AMSS-NCKU macro file macrodef.fh has been generated. " )
+
+
+##################################################################
+
+# Compile the AMSS-NCKU program according to user requirements
+# NOTE: ABE compilation is always performed, even when skipping TwoPuncture
+
+print()
+print( " Preparing to compile and run the AMSS-NCKU code as requested " )
+print( " Compiling the AMSS-NCKU code based on the generated macro files " )
+print()
+
+AMSS_NCKU_source_path = "AMSS_NCKU_source"
+AMSS_NCKU_source_copy = os.path.join(File_directory, "AMSS_NCKU_source_copy")
+
+## If AMSS_NCKU source folder is missing, create it and prompt the user
+if not os.path.exists(AMSS_NCKU_source_path):
+    os.makedirs(AMSS_NCKU_source_path)
+    print( " The AMSS-NCKU source files are incomplete; copy all source files into ./AMSS_NCKU_source. " )
+    print( " Press Enter to continue. " )
+    inputvalue = input()
+
+# Copy AMSS-NCKU source files to prepare for compilation
+# If skipping TwoPuncture and source_copy already exists, remove it first
+if skip_twopuncture and os.path.exists(AMSS_NCKU_source_copy):
+    shutil.rmtree(AMSS_NCKU_source_copy)
+
+shutil.copytree(AMSS_NCKU_source_path, AMSS_NCKU_source_copy)
+
+# Copy the generated macro files into the AMSS_NCKU source folder
+if not skip_twopuncture:
+    macrodef_h_path  = os.path.join(File_directory, "macrodef.h")
+    macrodef_fh_path = os.path.join(File_directory, "macrodef.fh")
+else:
+    # When skipping TwoPuncture, use existing macro files from previous run
+    macrodef_h_path  = os.path.join(File_directory, "macrodef.h")
+    macrodef_fh_path = os.path.join(File_directory, "macrodef.fh")
+
+shutil.copy2(macrodef_h_path,  AMSS_NCKU_source_copy)
+shutil.copy2(macrodef_fh_path, AMSS_NCKU_source_copy)
+
+# Compile related programs
+import makefile_and_run
+
+## Change working directory to the target source copy
+os.chdir(AMSS_NCKU_source_copy)
+
+## Build the main AMSS-NCKU executable (ABE or ABEGPU)
+makefile_and_run.makefile_ABE()
+
+## If the initial-data method is Ansorg-TwoPuncture, build the TwoPunctureABE executable
+## Only build TwoPunctureABE if not skipping TwoPuncture phase
+if (input_data.Initial_Data_Method == "Ansorg-TwoPuncture" ) and not skip_twopuncture:
+    makefile_and_run.makefile_TwoPunctureABE()
+
+## Change current working directory back up two levels
+os.chdir('..')
+os.chdir('..')
+
+print()
+
+##################################################################
+
+## Copy the AMSS-NCKU executable (ABE/ABEGPU) to the run directory
+
+if (input_data.GPU_Calculation == "no"):
+    ABE_file = os.path.join(AMSS_NCKU_source_copy, "ABE")
+elif (input_data.GPU_Calculation == "yes"):
+    ABE_file = os.path.join(AMSS_NCKU_source_copy, "ABEGPU")
+
+if not os.path.exists( ABE_file ):
+    print()
+    print( " Lack of AMSS-NCKU executable file ABE/ABEGPU; recompile AMSS_NCKU_source manually. " )
+    print( " When recompilation is finished, press Enter to continue. " )
+    inputvalue = input()
+
+## Copy the executable ABE (or ABEGPU) into the run directory
+shutil.copy2(ABE_file, output_directory)
+
+## If the initial-data method is TwoPuncture, copy the TwoPunctureABE executable to the run directory
+## Only copy TwoPunctureABE if not skipping TwoPuncture phase
+if (input_data.Initial_Data_Method == "Ansorg-TwoPuncture" ) and not skip_twopuncture:
+    TwoPuncture_file = os.path.join(AMSS_NCKU_source_copy, "TwoPunctureABE")
+
+    if not os.path.exists( TwoPuncture_file ):
+        print()
+        print( " Lack of AMSS-NCKU executable file TwoPunctureABE; recompile TwoPunctureABE in AMSS_NCKU_source. " )
+        print( " When recompilation is finished, press Enter to continue. " )
+        inputvalue = input()
+
+    ## Copy the TwoPunctureABE executable into the run directory
+    shutil.copy2(TwoPuncture_file, output_directory)
+
+##################################################################
+
+## If the initial-data method is TwoPuncture, generate the TwoPuncture input files
+
+if (input_data.Initial_Data_Method == "Ansorg-TwoPuncture" ) and not skip_twopuncture:
+
+    print()
+    print( " Initial data is chosen as Ansorg-TwoPuncture" )
+    print()
+    
+    print()
+    print( " Automatically generating the input parfile for the TwoPunctureABE executable " )
+    print()
+    
+    import generate_TwoPuncture_input
+    
+    generate_TwoPuncture_input.generate_AMSSNCKU_TwoPuncture_input()
+    
+    print()
+    print( " The input parfile for the TwoPunctureABE executable has been generated. " )
+    print()
+    
+    ## Generated AMSS-NCKU TwoPuncture input filename
+    AMSS_NCKU_TwoPuncture_inputfile      = 'AMSS-NCKU-TwoPuncture.input'
+    AMSS_NCKU_TwoPuncture_inputfile_path = os.path.join( File_directory, AMSS_NCKU_TwoPuncture_inputfile )
+ 
+    ## Copy and rename the file
+    shutil.copy2( AMSS_NCKU_TwoPuncture_inputfile_path, os.path.join(output_directory, 'TwoPunctureinput.par') )
+    
+    ## Run TwoPuncture to generate initial-data files
+    
+    start_time = time.time()  # Record start time
+
+    print()
+    print()
+    
+    ## Change to the output (run) directory
+    os.chdir(output_directory)
+
+    ## Run the TwoPuncture executable
+    import makefile_and_run
+    makefile_and_run.run_TwoPunctureABE()
+    
+    ## Change current working directory back up two levels
+    os.chdir('..')
+    os.chdir('..')
+
+elif (input_data.Initial_Data_Method == "Ansorg-TwoPuncture" ) and skip_twopuncture:
+    print()
+    print( " Skipping TwoPuncture execution, using existing initial data." )
+    print()
+    start_time = time.time()  # Record start time for ABE only
+else:
+    start_time = time.time()  # Record start time
+    
+##################################################################
+    
+## Update puncture data based on TwoPuncture run results
+
+if not skip_twopuncture:
+    import renew_puncture_parameter
+    renew_puncture_parameter.append_AMSSNCKU_BSSN_input(File_directory, output_directory)
+
+    ## Generated AMSS-NCKU input filename
+    AMSS_NCKU_inputfile      = 'AMSS-NCKU.input'
+    AMSS_NCKU_inputfile_path = os.path.join(File_directory, AMSS_NCKU_inputfile)
+ 
+    ## Copy and rename the file
+    shutil.copy2( AMSS_NCKU_inputfile_path, os.path.join(output_directory, 'input.par') )
+
+    print()
+    print( " Successfully copy all AMSS-NCKU input parfile to target dictionary. " )
+    print()
+else:
+    print()
+    print( " Using existing input.par file from previous run." )
+    print()
+
+##################################################################
+
+## Launch the AMSS-NCKU program
+
+print()
+print()
+
+## Change to the run directory
+os.chdir( output_directory )
+
+import makefile_and_run
+makefile_and_run.run_ABE()
+
+## Change current working directory back up two levels
+os.chdir('..')
+os.chdir('..')
+
+end_time = time.time()
+elapsed_time = end_time - start_time
+
+##################################################################
+
+## Copy some basic input and log files out to facilitate debugging
+
+## Path to the file that stores calculation settings
+AMSS_NCKU_error_file_path = os.path.join(binary_results_directory, "setting.par")
+## Copy and rename the file for easier inspection
+shutil.copy( AMSS_NCKU_error_file_path, os.path.join(output_directory, "AMSSNCKU_setting_parameter") )
+
+## Path to the error log file
+AMSS_NCKU_error_file_path = os.path.join(binary_results_directory, "Error.log")
+## Copy and rename the error log
+shutil.copy( AMSS_NCKU_error_file_path, os.path.join(output_directory, "Error.log") )
+
+## Primary program outputs
+AMSS_NCKU_BH_data         = os.path.join(binary_results_directory, "bssn_BH.dat"        )
+AMSS_NCKU_ADM_data        = os.path.join(binary_results_directory, "bssn_ADMQs.dat"     )
+AMSS_NCKU_psi4_data       = os.path.join(binary_results_directory, "bssn_psi4.dat"      )
+AMSS_NCKU_constraint_data = os.path.join(binary_results_directory, "bssn_constraint.dat")
+## copy and rename the file
+shutil.copy( AMSS_NCKU_BH_data,         os.path.join(output_directory, "bssn_BH.dat"        ) )
+shutil.copy( AMSS_NCKU_ADM_data,        os.path.join(output_directory, "bssn_ADMQs.dat"     ) )
+shutil.copy( AMSS_NCKU_psi4_data,       os.path.join(output_directory, "bssn_psi4.dat"      ) )
+shutil.copy( AMSS_NCKU_constraint_data, os.path.join(output_directory, "bssn_constraint.dat") )
+
+## Additional program outputs
+if (input_data.Equation_Class == "BSSN-EM"):
+    AMSS_NCKU_phi1_data = os.path.join(binary_results_directory, "bssn_phi1.dat" )
+    AMSS_NCKU_phi2_data = os.path.join(binary_results_directory, "bssn_phi2.dat" )
+    shutil.copy( AMSS_NCKU_phi1_data, os.path.join(output_directory, "bssn_phi1.dat" ) )
+    shutil.copy( AMSS_NCKU_phi2_data, os.path.join(output_directory, "bssn_phi2.dat" ) )
+elif (input_data.Equation_Class == "BSSN-EScalar"):
+    AMSS_NCKU_maxs_data = os.path.join(binary_results_directory, "bssn_maxs.dat" )
+    shutil.copy( AMSS_NCKU_maxs_data, os.path.join(output_directory, "bssn_maxs.dat" ) )
+
+##################################################################
+
+## Plot the AMSS-NCKU program results
+
+print()
+print( " Plotting the txt and binary results data from the AMSS-NCKU simulation " )
+print()
+
+
+import plot_xiaoqu
+import plot_GW_strain_amplitude_xiaoqu
+
+## Plot black hole trajectory
+plot_xiaoqu.generate_puncture_orbit_plot(   binary_results_directory, figure_directory )
+plot_xiaoqu.generate_puncture_orbit_plot3D( binary_results_directory, figure_directory )
+
+## Plot black hole separation vs. time
+plot_xiaoqu.generate_puncture_distence_plot( binary_results_directory, figure_directory )
+
+## Plot gravitational waveforms (psi4 and strain amplitude)
+for i in range(input_data.Detector_Number):
+    plot_xiaoqu.generate_gravitational_wave_psi4_plot( binary_results_directory, figure_directory, i )
+    plot_GW_strain_amplitude_xiaoqu.generate_gravitational_wave_amplitude_plot( binary_results_directory, figure_directory, i )
+
+## Plot ADM mass evolution
+for i in range(input_data.Detector_Number):
+    plot_xiaoqu.generate_ADMmass_plot( binary_results_directory, figure_directory, i )
+
+## Plot Hamiltonian constraint violation over time
+for i in range(input_data.grid_level):
+    plot_xiaoqu.generate_constraint_check_plot( binary_results_directory, figure_directory, i )
+
+## Plot stored binary data
+plot_xiaoqu.generate_binary_data_plot( binary_results_directory, figure_directory )
+
+print()
+print( f" This Program Cost = {elapsed_time} Seconds " )
+print()
+
+
+##################################################################
+
+print()
+print( " The AMSS-NCKU-Python simulation is successfully finished, thanks for using !!! " )
+print()
+
+##################################################################
+
+

AMSS_NCKU_Verify_ASC26.py

@@ -277,3 +277,4 @@ def main():
 
 if __name__ == "__main__":
     main()
+

AMSS_NCKU_source/FFT.f90

@@ -37,51 +37,57 @@ close(77)
 end program checkFFT
 #endif
 
-!-------------
-! Optimized FFT using Intel oneMKL DFTI
-! Mathematical equivalence: Standard DFT definition
-!   Forward (isign=1):  X[k] = sum_{n=0}^{N-1} x[n] * exp(-2*pi*i*k*n/N)
-!   Backward (isign=-1): X[k] = sum_{n=0}^{N-1} x[n] * exp(+2*pi*i*k*n/N)
-! Input/Output: dataa is interleaved complex array [Re(0),Im(0),Re(1),Im(1),...]
 !-------------
 SUBROUTINE four1(dataa,nn,isign)
-use MKL_DFTI
 implicit none
-INTEGER, intent(in) :: isign, nn
+INTEGER::isign,nn
-DOUBLE PRECISION, dimension(2*nn), intent(inout) :: dataa
+double precision,dimension(2*nn)::dataa
-
+INTEGER::i,istep,j,m,mmax,n
-type(DFTI_DESCRIPTOR), pointer :: desc
+double precision::tempi,tempr
-integer :: status
+DOUBLE PRECISION::theta,wi,wpi,wpr,wr,wtemp
-
+n=2*nn
-! Create DFTI descriptor for 1D complex-to-complex transform
+j=1
-status = DftiCreateDescriptor(desc, DFTI_DOUBLE, DFTI_COMPLEX, 1, nn)
+do i=1,n,2
-if (status /= 0) return
+  if(j.gt.i)then
-
+     tempr=dataa(j)
-! Set input/output storage as interleaved complex (default)
+     tempi=dataa(j+1)
-status = DftiSetValue(desc, DFTI_PLACEMENT, DFTI_INPLACE)
+     dataa(j)=dataa(i)
-if (status /= 0) then
+     dataa(j+1)=dataa(i+1)
-   status = DftiFreeDescriptor(desc)
+     dataa(i)=tempr
-   return
+     dataa(i+1)=tempi
+  endif
+  m=nn
+1 if ((m.ge.2).and.(j.gt.m)) then
+  j=j-m
+  m=m/2
+goto 1
+  endif
+j=j+m
+enddo
+mmax=2
+2  if (n.gt.mmax) then
+     istep=2*mmax
+     theta=6.28318530717959d0/(isign*mmax)
+     wpr=-2.d0*sin(0.5d0*theta)**2
+     wpi=sin(theta)
+     wr=1.d0
+     wi=0.d0
+     do m=1,mmax,2
+       do i=m,n,istep
+         j=i+mmax
+         tempr=sngl(wr)*dataa(j)-sngl(wi)*dataa(j+1)
+         tempi=sngl(wr)*dataa(j+1)+sngl(wi)*dataa(j)
+         dataa(j)=dataa(i)-tempr
+         dataa(j+1)=dataa(i+1)-tempi
+         dataa(i)=dataa(i)+tempr
+         dataa(i+1)=dataa(i+1)+tempi
+       enddo
+          wtemp=wr
+          wr=wr*wpr-wi*wpi+wr
+          wi=wi*wpr+wtemp*wpi+wi
+     enddo
+mmax=istep
+goto 2
 endif
-
-! Commit the descriptor
-status = DftiCommitDescriptor(desc)
-if (status /= 0) then
-   status = DftiFreeDescriptor(desc)
-   return
-endif
-
-! Execute FFT based on direction
-if (isign == 1) then
-   ! Forward FFT: exp(-2*pi*i*k*n/N)
-   status = DftiComputeForward(desc, dataa)
-else
-   ! Backward FFT: exp(+2*pi*i*k*n/N)
-   status = DftiComputeBackward(desc, dataa)
-endif
-
-! Free descriptor
-status = DftiFreeDescriptor(desc)
-
 return
 END SUBROUTINE four1

AMSS_NCKU_source/TwoPunctures.C

@@ -5,7 +5,6 @@
 #include <cstdio>
 #include <cstdlib>
 #include <string>
-#include <cstring>
 #include <iostream>
 #include <iomanip>
 #include <fstream>
@@ -28,7 +27,6 @@ using namespace std;
 #endif
 
 #include "TwoPunctures.h"
-#include <mkl_cblas.h>
 
 TwoPunctures::TwoPunctures(double mp, double mm, double b,
                            double P_plusx, double P_plusy, double P_plusz,
@@ -61,110 +59,13 @@ TwoPunctures::TwoPunctures(double mp, double mm, double b,
   F = dvector(0, ntotal - 1);
   allocate_derivs(&u, ntotal);
   allocate_derivs(&v, ntotal);
-
-  // Allocate workspace buffers for hot-path allocation elimination
-  int N = maximum3(n1, n2, n3);
-  int maxn = maximum2(n1, n2);
-
-  // LineRelax_be workspace (sized for n2)
-  ws_diag_be = new double[n2];
-  ws_e_be = new double[n2 - 1];
-  ws_f_be = new double[n2 - 1];
-  ws_b_be = new double[n2];
-  ws_x_be = new double[n2];
-
-  // LineRelax_al workspace (sized for n1)
-  ws_diag_al = new double[n1];
-  ws_e_al = new double[n1 - 1];
-  ws_f_al = new double[n1 - 1];
-  ws_b_al = new double[n1];
-  ws_x_al = new double[n1];
-
-  // ThomasAlgorithm workspace (sized for max(n1,n2))
-  ws_thomas_y = new double[maxn];
-
-  // JFD_times_dv workspace (sized for nvar)
-  ws_jfd_values = dvector(0, nvar - 1);
-  allocate_derivs(&ws_jfd_dU, nvar);
-  allocate_derivs(&ws_jfd_U, nvar);
-
-  // chebft_Zeros workspace (sized for N+1)
-  ws_cheb_c = dvector(0, N);
-
-  // fourft workspace (sized for N/2+1 each)
-  ws_four_a = dvector(0, N / 2);
-  ws_four_b = dvector(0, N / 2);
-
-  // Derivatives_AB3 workspace
-  ws_deriv_p = dvector(0, N);
-  ws_deriv_dp = dvector(0, N);
-  ws_deriv_d2p = dvector(0, N);
-  ws_deriv_q = dvector(0, N);
-  ws_deriv_dq = dvector(0, N);
-  ws_deriv_r = dvector(0, N);
-  ws_deriv_dr = dvector(0, N);
-  ws_deriv_indx = ivector(0, N);
-
-  // F_of_v workspace
-  ws_fov_sources = new double[n1 * n2 * n3];
-  ws_fov_values = dvector(0, nvar - 1);
-  allocate_derivs(&ws_fov_U, nvar);
-
-  // J_times_dv workspace
-  ws_jtdv_values = dvector(0, nvar - 1);
-  allocate_derivs(&ws_jtdv_dU, nvar);
-  allocate_derivs(&ws_jtdv_U, nvar);
 }
 
 TwoPunctures::~TwoPunctures()
 {
-  int const nvar = 1, n1 = npoints_A, n2 = npoints_B, n3 = npoints_phi;
-  int N = maximum3(n1, n2, n3);
-
   free_dvector(F, 0, ntotal - 1);
   free_derivs(&u, ntotal);
   free_derivs(&v, ntotal);
-
-  // Free workspace buffers
-  delete[] ws_diag_be;
-  delete[] ws_e_be;
-  delete[] ws_f_be;
-  delete[] ws_b_be;
-  delete[] ws_x_be;
-
-  delete[] ws_diag_al;
-  delete[] ws_e_al;
-  delete[] ws_f_al;
-  delete[] ws_b_al;
-  delete[] ws_x_al;
-
-  delete[] ws_thomas_y;
-
-  free_dvector(ws_jfd_values, 0, nvar - 1);
-  free_derivs(&ws_jfd_dU, nvar);
-  free_derivs(&ws_jfd_U, nvar);
-
-  free_dvector(ws_cheb_c, 0, N);
-
-  free_dvector(ws_four_a, 0, N / 2);
-  free_dvector(ws_four_b, 0, N / 2);
-
-  free_dvector(ws_deriv_p, 0, N);
-  free_dvector(ws_deriv_dp, 0, N);
-  free_dvector(ws_deriv_d2p, 0, N);
-  free_dvector(ws_deriv_q, 0, N);
-  free_dvector(ws_deriv_dq, 0, N);
-  free_dvector(ws_deriv_r, 0, N);
-  free_dvector(ws_deriv_dr, 0, N);
-  free_ivector(ws_deriv_indx, 0, N);
-
-  delete[] ws_fov_sources;
-  free_dvector(ws_fov_values, 0, nvar - 1);
-  free_derivs(&ws_fov_U, nvar);
-
-  free_dvector(ws_jtdv_values, 0, nvar - 1);
-  free_derivs(&ws_jtdv_dU, nvar);
-  free_derivs(&ws_jtdv_U, nvar);
 }
 
 void TwoPunctures::Solve()
@@ -753,7 +654,7 @@ void TwoPunctures::chebft_Zeros(double u[], int n, int inv)
   int k, j, isignum;
   double fac, sum, Pion, *c;
 
-  c = ws_cheb_c;
+  c = dvector(0, n);
   Pion = Pi / n;
   if (inv == 0)
   {
@@ -784,6 +685,7 @@ void TwoPunctures::chebft_Zeros(double u[], int n, int inv)
   }
   for (j = 0; j < n; j++)
     u[j] = c[j];
+  free_dvector(c, 0, n);
 }
 
 /* --------------------------------------------------------------------------*/
@@ -871,8 +773,8 @@ void TwoPunctures::fourft(double *u, int N, int inv)
   double x, x1, fac, Pi_fac, *a, *b;
 
   M = N / 2;
-  a = ws_four_a;
+  a = dvector(0, M);
-  b = ws_four_b - 1; /* offset to match dvector(1,M) indexing */
+  b = dvector(1, M); /* Actually: b=vector(1,M-1) but this is problematic if M=1*/
   fac = 1. / M;
   Pi_fac = Pi * fac;
   if (inv == 0)
@@ -921,6 +823,8 @@ void TwoPunctures::fourft(double *u, int N, int inv)
       iy = -iy;
     }
   }
+  free_dvector(a, 0, M);
+  free_dvector(b, 1, M);
 }
 
 /* -----------------------------------------*/
@@ -987,17 +891,25 @@ double TwoPunctures::norm1(double *v, int n)
 /* -------------------------------------------------------------------------*/
 double TwoPunctures::norm2(double *v, int n)
 {
-  // Optimized with oneMKL BLAS DNRM2
+  int i;
-  // Computes: sqrt(sum(v[i]^2))
+  double result = 0;
-  return cblas_dnrm2(n, v, 1);
+
+  for (i = 0; i < n; i++)
+    result += v[i] * v[i];
+
+  return sqrt(result);
 }
 
 /* -------------------------------------------------------------------------*/
 double TwoPunctures::scalarproduct(double *v, double *w, int n)
 {
-  // Optimized with oneMKL BLAS DDOT
+  int i;
-  // Computes: sum(v[i] * w[i])
+  double result = 0;
-  return cblas_ddot(n, v, 1, w, 1);
+
+  for (i = 0; i < n; i++)
+    result += v[i] * w[i];
+
+  return result;
 }
 
 /* -------------------------------------------------------------------------*/
@@ -1213,14 +1125,14 @@ void TwoPunctures::Derivatives_AB3(int nvar, int n1, int n2, int n3, derivs v)
   double *p, *dp, *d2p, *q, *dq, *r, *dr;
 
   N = maximum3(n1, n2, n3);
-  p = ws_deriv_p;
+  p = dvector(0, N);
-  dp = ws_deriv_dp;
+  dp = dvector(0, N);
-  d2p = ws_deriv_d2p;
+  d2p = dvector(0, N);
-  q = ws_deriv_q;
+  q = dvector(0, N);
-  dq = ws_deriv_dq;
+  dq = dvector(0, N);
-  r = ws_deriv_r;
+  r = dvector(0, N);
-  dr = ws_deriv_dr;
+  dr = dvector(0, N);
-  indx = ws_deriv_indx;
+  indx = ivector(0, N);
 
   for (ivar = 0; ivar < nvar; ivar++)
   {
@@ -1303,6 +1215,14 @@ void TwoPunctures::Derivatives_AB3(int nvar, int n1, int n2, int n3, derivs v)
       }
     }
   }
+  free_dvector(p, 0, N);
+  free_dvector(dp, 0, N);
+  free_dvector(d2p, 0, N);
+  free_dvector(q, 0, N);
+  free_dvector(dq, 0, N);
+  free_dvector(r, 0, N);
+  free_dvector(dr, 0, N);
+  free_ivector(indx, 0, N);
 }
 /* --------------------------------------------------------------------------*/
 void TwoPunctures::Newton(int const nvar, int const n1, int const n2, int const n3,
@@ -1371,11 +1291,10 @@ void TwoPunctures::F_of_v(int nvar, int n1, int n2, int n3, derivs v, double *F,
   derivs U;
   double *sources;
 
-  values = ws_fov_values;
+  values = dvector(0, nvar - 1);
-  U = ws_fov_U;
+  allocate_derivs(&U, nvar);
 
-  sources = ws_fov_sources;
+  sources = (double *)calloc(n1 * n2 * n3, sizeof(double));
-  memset(sources, 0, n1 * n2 * n3 * sizeof(double));
   if (0)
   {
     double *s_x, *s_y, *s_z;
@@ -1530,6 +1449,9 @@ void TwoPunctures::F_of_v(int nvar, int n1, int n2, int n3, derivs v, double *F,
   {
     fclose(debugfile);
   }
+  free(sources);
+  free_dvector(values, 0, nvar - 1);
+  free_derivs(&U, nvar);
 }
 /* --------------------------------------------------------------------------*/
 double TwoPunctures::norm_inf(double const *F, int const ntotal)
@@ -1935,12 +1857,11 @@ void TwoPunctures::J_times_dv(int nvar, int n1, int n2, int n3, derivs dv, doubl
 
   Derivatives_AB3(nvar, n1, n2, n3, dv);
 
-  values = ws_jtdv_values;
-  dU = ws_jtdv_dU;
-  U = ws_jtdv_U;
-
   for (i = 0; i < n1; i++)
   {
+    values = dvector(0, nvar - 1);
+    allocate_derivs(&dU, nvar);
+    allocate_derivs(&U, nvar);
     for (j = 0; j < n2; j++)
     {
       for (k = 0; k < n3; k++)
@@ -1994,6 +1915,9 @@ void TwoPunctures::J_times_dv(int nvar, int n1, int n2, int n3, derivs dv, doubl
         }
       }
     }
+    free_dvector(values, 0, nvar - 1);
+    free_derivs(&dU, nvar);
+    free_derivs(&U, nvar);
   }
 }
 /* --------------------------------------------------------------------------*/
@@ -2040,11 +1964,17 @@ void TwoPunctures::LineRelax_be(double *dv,
 {
   int j, m, Ic, Ip, Im, col, ivar;
 
-  double *diag = ws_diag_be;
+  double *diag = new double[n2];
-  double *e = ws_e_be;     /* above diagonal */
+  double *e = new double[n2 - 1]; /* above diagonal */
-  double *f = ws_f_be;     /* below diagonal */
+  double *f = new double[n2 - 1]; /* below diagonal */
-  double *b = ws_b_be;     /* rhs */
+  double *b = new double[n2];     /* rhs */
-  double *x = ws_x_be;     /* solution vector */
+  double *x = new double[n2];     /* solution vector */
+
+  //  gsl_vector *diag = gsl_vector_alloc(n2);
+  //  gsl_vector *e = gsl_vector_alloc(n2-1); /* above diagonal */
+  //  gsl_vector *f = gsl_vector_alloc(n2-1); /* below diagonal */
+  //  gsl_vector *b = gsl_vector_alloc(n2);   /* rhs */
+  //  gsl_vector *x = gsl_vector_alloc(n2);   /* solution vector */
 
   for (ivar = 0; ivar < nvar; ivar++)
   {
@@ -2054,35 +1984,62 @@ void TwoPunctures::LineRelax_be(double *dv,
     }
     diag[n2 - 1] = 0;
 
+    //    gsl_vector_set_zero(diag);
+    //    gsl_vector_set_zero(e);
+    //    gsl_vector_set_zero(f);
     for (j = 0; j < n2; j++)
     {
       Ip = Index(ivar, i, j + 1, k, nvar, n1, n2, n3);
       Ic = Index(ivar, i, j, k, nvar, n1, n2, n3);
       Im = Index(ivar, i, j - 1, k, nvar, n1, n2, n3);
       b[j] = rhs[Ic];
+      //      gsl_vector_set(b,j,rhs[Ic]);
       for (m = 0; m < ncols[Ic]; m++)
       {
         col = cols[Ic][m];
         if (col != Ip && col != Ic && col != Im)
           b[j] -= JFD[Ic][m] * dv[col];
+        //          *gsl_vector_ptr(b, j) -= JFD[Ic][m] * dv[col];
         else
         {
           if (col == Im && j > 0)
             f[j - 1] = JFD[Ic][m];
+          //            gsl_vector_set(f,j-1,JFD[Ic][m]);
           if (col == Ic)
             diag[j] = JFD[Ic][m];
+          //            gsl_vector_set(diag,j,JFD[Ic][m]);
           if (col == Ip && j < n2 - 1)
             e[j] = JFD[Ic][m];
+          //            gsl_vector_set(e,j,JFD[Ic][m]);
         }
       }
     }
+    //          A x = b
+    //          A = ( d_0 e_0  0   0  )
+    //              ( f_0 d_1 e_1  0  )
+    //              (  0  f_1 d_2 e_2 )
+    //              (  0   0  f_2 d_3 )
+    //
     ThomasAlgorithm(n2, f, diag, e, x, b);
+    //    gsl_linalg_solve_tridiag(diag, e, f, b, x);
     for (j = 0; j < n2; j++)
     {
       Ic = Index(ivar, i, j, k, nvar, n1, n2, n3);
       dv[Ic] = x[j];
+      //      dv[Ic] = gsl_vector_get(x, j);
     }
   }
+
+  delete[] diag;
+  delete[] e;
+  delete[] f;
+  delete[] b;
+  delete[] x;
+  //  gsl_vector_free(diag);
+  //  gsl_vector_free(e);
+  //  gsl_vector_free(f);
+  //  gsl_vector_free(b);
+  //  gsl_vector_free(x);
 }
 /* --------------------------------------------------------------------------*/
 void TwoPunctures::JFD_times_dv(int i, int j, int k, int nvar, int n1, int n2,
@@ -2099,8 +2056,8 @@ void TwoPunctures::JFD_times_dv(int i, int j, int k, int nvar, int n1, int n2,
       ha, ga, ga2, hb, gb, gb2, hp, gp, gp2, gagb, gagp, gbgp;
   derivs dU, U;
 
-  dU = ws_jfd_dU;
+  allocate_derivs(&dU, nvar);
-  U = ws_jfd_U;
+  allocate_derivs(&U, nvar);
 
   if (k < 0)
     k = k + n3;
@@ -2218,6 +2175,9 @@ void TwoPunctures::JFD_times_dv(int i, int j, int k, int nvar, int n1, int n2,
   LinEquations(A, B, X, R, x, r, phi, y, z, dU, U, values);
   for (ivar = 0; ivar < nvar; ivar++)
     values[ivar] *= FAC;
+
+  free_derivs(&dU, nvar);
+  free_derivs(&U, nvar);
 }
 #undef FAC
 /*-----------------------------------------------------------*/
@@ -2249,11 +2209,17 @@ void TwoPunctures::LineRelax_al(double *dv,
 {
   int i, m, Ic, Ip, Im, col, ivar;
 
-  double *diag = ws_diag_al;
+  double *diag = new double[n1];
-  double *e = ws_e_al;     /* above diagonal */
+  double *e = new double[n1 - 1]; /* above diagonal */
-  double *f = ws_f_al;     /* below diagonal */
+  double *f = new double[n1 - 1]; /* below diagonal */
-  double *b = ws_b_al;     /* rhs */
+  double *b = new double[n1];     /* rhs */
-  double *x = ws_x_al;     /* solution vector */
+  double *x = new double[n1];     /* solution vector */
+
+  //  gsl_vector *diag = gsl_vector_alloc(n1);
+  //  gsl_vector *e = gsl_vector_alloc(n1-1); /* above diagonal */
+  //  gsl_vector *f = gsl_vector_alloc(n1-1); /* below diagonal */
+  //  gsl_vector *b = gsl_vector_alloc(n1);   /* rhs */
+  //  gsl_vector *x = gsl_vector_alloc(n1);   /* solution vector */
 
   for (ivar = 0; ivar < nvar; ivar++)
   {
@@ -2263,35 +2229,57 @@ void TwoPunctures::LineRelax_al(double *dv,
     }
     diag[n1 - 1] = 0;
 
+    //    gsl_vector_set_zero(diag);
+    //    gsl_vector_set_zero(e);
+    //    gsl_vector_set_zero(f);
     for (i = 0; i < n1; i++)
     {
       Ip = Index(ivar, i + 1, j, k, nvar, n1, n2, n3);
       Ic = Index(ivar, i, j, k, nvar, n1, n2, n3);
       Im = Index(ivar, i - 1, j, k, nvar, n1, n2, n3);
       b[i] = rhs[Ic];
+      //      gsl_vector_set(b,i,rhs[Ic]);
       for (m = 0; m < ncols[Ic]; m++)
       {
         col = cols[Ic][m];
         if (col != Ip && col != Ic && col != Im)
           b[i] -= JFD[Ic][m] * dv[col];
+        //          *gsl_vector_ptr(b, i) -= JFD[Ic][m] * dv[col];
         else
         {
           if (col == Im && i > 0)
             f[i - 1] = JFD[Ic][m];
+          //            gsl_vector_set(f,i-1,JFD[Ic][m]);
           if (col == Ic)
             diag[i] = JFD[Ic][m];
+          //            gsl_vector_set(diag,i,JFD[Ic][m]);
           if (col == Ip && i < n1 - 1)
             e[i] = JFD[Ic][m];
+          //            gsl_vector_set(e,i,JFD[Ic][m]);
         }
       }
     }
     ThomasAlgorithm(n1, f, diag, e, x, b);
+    //    gsl_linalg_solve_tridiag(diag, e, f, b, x);
     for (i = 0; i < n1; i++)
     {
       Ic = Index(ivar, i, j, k, nvar, n1, n2, n3);
       dv[Ic] = x[i];
+      //      dv[Ic] = gsl_vector_get(x, i);
     }
   }
+
+  delete[] diag;
+  delete[] e;
+  delete[] f;
+  delete[] b;
+  delete[] x;
+
+  //  gsl_vector_free(diag);
+  //  gsl_vector_free(e);
+  //  gsl_vector_free(f);
+  //  gsl_vector_free(b);
+  //  gsl_vector_free(x);
 }
 /* -------------------------------------------------------------------------*/
 // a[N], b[N-1], c[N-1], x[N], q[N]
@@ -2303,29 +2291,44 @@ void TwoPunctures::LineRelax_al(double *dv,
 //"Parallel Scientific Computing in C++ and MPI" P361
 void TwoPunctures::ThomasAlgorithm(int N, double *b, double *a, double *c, double *x, double *q)
 {
-  // In-place Thomas algorithm: uses a[] as d workspace, b[] as l workspace.
-  // c[] is already u (above-diagonal). ws_thomas_y is pre-allocated workspace.
   int i;
-  double *y = ws_thomas_y;
+  double *l, *u, *d, *y;
+  l = new double[N - 1];
+  u = new double[N - 1];
+  d = new double[N];
+  y = new double[N];
+
+  /* LU Decomposition */
+  d[0] = a[0];
+  u[0] = c[0];
 
-  /* LU Decomposition (in-place: a becomes d, b becomes l) */
   for (i = 0; i < N - 2; i++)
   {
-    b[i] = b[i] / a[i];
+    l[i] = b[i] / d[i];
-    a[i + 1] = a[i + 1] - b[i] * c[i];
+    d[i + 1] = a[i + 1] - l[i] * u[i];
+    u[i + 1] = c[i + 1];
   }
-  b[N - 2] = b[N - 2] / a[N - 2];
+
-  a[N - 1] = a[N - 1] - b[N - 2] * c[N - 2];
+  l[N - 2] = b[N - 2] / d[N - 2];
+  d[N - 1] = a[N - 1] - l[N - 2] * u[N - 2];
 
   /* Forward Substitution [L][y] = [q] */
   y[0] = q[0];
   for (i = 1; i < N; i++)
-    y[i] = q[i] - b[i - 1] * y[i - 1];
+    y[i] = q[i] - l[i - 1] * y[i - 1];
 
   /* Backward Substitution [U][x] = [y] */
-  x[N - 1] = y[N - 1] / a[N - 1];
+  x[N - 1] = y[N - 1] / d[N - 1];
+
   for (i = N - 2; i >= 0; i--)
-    x[i] = (y[i] - c[i] * x[i + 1]) / a[i];
+    x[i] = (y[i] - u[i] * x[i + 1]) / d[i];
+
+  delete[] l;
+  delete[] u;
+  delete[] d;
+  delete[] y;
+
+  return;
 }
 // --------------------------------------------------------------------------*/
 // Calculates the value of v at an arbitrary position (x,y,z) if the spectral coefficients are know*/*/

AMSS_NCKU_source/TwoPunctures.h

@@ -42,33 +42,6 @@ private:
 
        int ntotal;
 
-       // Pre-allocated workspace buffers for hot-path allocation elimination
-       // LineRelax_be workspace (sized for n2)
-       double *ws_diag_be, *ws_e_be, *ws_f_be, *ws_b_be, *ws_x_be;
-       // LineRelax_al workspace (sized for n1)
-       double *ws_diag_al, *ws_e_al, *ws_f_al, *ws_b_al, *ws_x_al;
-       // ThomasAlgorithm workspace (sized for max(n1,n2))
-       double *ws_thomas_y;
-       // JFD_times_dv workspace (sized for nvar)
-       double *ws_jfd_values;
-       derivs ws_jfd_dU, ws_jfd_U;
-       // chebft_Zeros workspace (sized for max(n1,n2,n3)+1)
-       double *ws_cheb_c;
-       // fourft workspace (sized for max(n1,n2,n3)/2+1 each)
-       double *ws_four_a, *ws_four_b;
-       // Derivatives_AB3 workspace
-       double *ws_deriv_p, *ws_deriv_dp, *ws_deriv_d2p;
-       double *ws_deriv_q, *ws_deriv_dq;
-       double *ws_deriv_r, *ws_deriv_dr;
-       int *ws_deriv_indx;
-       // F_of_v workspace
-       double *ws_fov_sources;
-       double *ws_fov_values;
-       derivs ws_fov_U;
-       // J_times_dv workspace
-       double *ws_jtdv_values;
-       derivs ws_jtdv_dU, ws_jtdv_U;
-
        struct parameters
        {
               int nvar, n1, n2, n3;

AMSS_NCKU_source/bssn_rhs.f90

@@ -106,8 +106,7 @@
   call getpbh(BHN,Porg,Mass)
 #endif
 
-!!! sanity check (disabled in production builds for performance)
+!!! sanity check
-#ifdef DEBUG
   dX = sum(chi)+sum(trK)+sum(dxx)+sum(gxy)+sum(gxz)+sum(dyy)+sum(gyz)+sum(dzz) &
       +sum(Axx)+sum(Axy)+sum(Axz)+sum(Ayy)+sum(Ayz)+sum(Azz)                   &
       +sum(Gamx)+sum(Gamy)+sum(Gamz)                                           &
@@ -137,7 +136,6 @@
      gont = 1
      return
   endif
-#endif
 
   PI = dacos(-ONE)
 
@@ -168,8 +166,6 @@
   call fderivs(ex,gyz,gyzx,gyzy,gyzz,X,Y,Z,SYM ,ANTI,ANTI,Symmetry,Lev)
   call fderivs(ex,dzz,gzzx,gzzy,gzzz,X,Y,Z,SYM ,SYM ,SYM ,Symmetry,Lev)
 
-  !$OMP PARALLEL
-  !$OMP WORKSHARE
   gxx_rhs = - TWO * alpn1 * Axx    -  F2o3 * gxx * div_beta          + &
               TWO *(  gxx * betaxx +   gxy * betayx +   gxz * betazx)
 
@@ -203,8 +199,6 @@
   gupyy =   ( gxx * gzz - gxz * gxz ) / gupzz
   gupyz = - ( gxx * gyz - gxy * gxz ) / gupzz
   gupzz =   ( gxx * gyy - gxy * gxy ) / gupzz
-  !$OMP END WORKSHARE
-  !$OMP END PARALLEL
 
   if(co == 0)then
 ! Gam^i_Res = Gam^i + gup^ij_,j
@@ -238,8 +232,6 @@
   endif
 
 ! second kind of connection
-  !$OMP PARALLEL
-  !$OMP WORKSHARE
   Gamxxx =HALF*( gupxx*gxxx + gupxy*(TWO*gxyx - gxxy ) + gupxz*(TWO*gxzx - gxxz ))
   Gamyxx =HALF*( gupxy*gxxx + gupyy*(TWO*gxyx - gxxy ) + gupyz*(TWO*gxzx - gxxz ))
   Gamzxx =HALF*( gupxz*gxxx + gupyz*(TWO*gxyx - gxxy ) + gupzz*(TWO*gxzx - gxxz ))
@@ -288,8 +280,6 @@
           (gupxy * gupyz       + gupyy * gupxz)* Axy                       + &
           (gupxy * gupzz       + gupyz * gupxz)* Axz                       + &
           (gupyy * gupzz       + gupyz * gupyz)* Ayz
-  !$OMP END WORKSHARE
-  !$OMP END PARALLEL
 
 ! Right hand side for Gam^i without shift terms...
   call fderivs(ex,Lap,Lapx,Lapy,Lapz,X,Y,Z,SYM,SYM,SYM,Symmetry,Lev)
@@ -344,8 +334,6 @@
   call fderivs(ex,Gamy,Gamyx,Gamyy,Gamyz,X,Y,Z,SYM ,ANTI,SYM ,Symmetry,Lev)
   call fderivs(ex,Gamz,Gamzx,Gamzy,Gamzz,X,Y,Z,SYM ,SYM ,ANTI,Symmetry,Lev)
 
-  !$OMP PARALLEL
-  !$OMP WORKSHARE
   Gamx_rhs =               Gamx_rhs +  F2o3 *  Gamxa * div_beta        - &
                      Gamxa * betaxx - Gamya * betaxy - Gamza * betaxz  + &
              F1o3 * (gupxx * fxx    + gupxy * fxy    + gupxz * fxz    ) + &
@@ -385,8 +373,6 @@
   gyyz = gxz * Gamxyy + gyz * Gamyyy + gzz * Gamzyy
   gyzz = gxz * Gamxyz + gyz * Gamyyz + gzz * Gamzyz
   gzzz = gxz * Gamxzz + gyz * Gamyzz + gzz * Gamzzz
-  !$OMP END WORKSHARE
-  !$OMP END PARALLEL
 
 !compute Ricci tensor for tilted metric
    call fdderivs(ex,dxx,fxx,fxy,fxz,fyy,fyz,fzz,X,Y,Z,SYM ,SYM ,SYM ,symmetry,Lev)
@@ -413,8 +399,6 @@
    Ryz =   gupxx * fxx + gupyy * fyy + gupzz * fzz + &
          ( gupxy * fxy + gupxz * fxz + gupyz * fyz ) * TWO
 
-  !$OMP PARALLEL
-  !$OMP WORKSHARE
   Rxx =     - HALF * Rxx                                   + &
                gxx * Gamxx+ gxy * Gamyx   +    gxz * Gamzx + &
              Gamxa * gxxx +  Gamya * gxyx +  Gamza * gxzx  + &
@@ -615,13 +599,9 @@
             Gamxyz * gxzz + Gamyyz * gyzz + Gamzyz * gzzz  + &
             Gamxzz * gxzy + Gamyzz * gyzy + Gamzzz * gzzy  + &
             Gamxyz * gzzx + Gamyyz * gzzy + Gamzyz * gzzz )
-  !$OMP END WORKSHARE
-  !$OMP END PARALLEL
 !covariant second derivative of chi respect to tilted metric
   call fdderivs(ex,chi,fxx,fxy,fxz,fyy,fyz,fzz,X,Y,Z,SYM,SYM,SYM,Symmetry,Lev)
 
-  !$OMP PARALLEL
-  !$OMP WORKSHARE
   fxx = fxx - Gamxxx * chix - Gamyxx * chiy - Gamzxx * chiz
   fxy = fxy - Gamxxy * chix - Gamyxy * chiy - Gamzxy * chiz
   fxz = fxz - Gamxxz * chix - Gamyxz * chiy - Gamzxz * chiz
@@ -644,15 +624,11 @@
   Rxy = Rxy + (fxy - chix*chiy/chin1/TWO + gxy * f)/chin1/TWO
   Rxz = Rxz + (fxz - chix*chiz/chin1/TWO + gxz * f)/chin1/TWO
   Ryz = Ryz + (fyz - chiy*chiz/chin1/TWO + gyz * f)/chin1/TWO
-  !$OMP END WORKSHARE
-  !$OMP END PARALLEL
 
 ! covariant second derivatives of the lapse respect to physical metric
   call fdderivs(ex,Lap,fxx,fxy,fxz,fyy,fyz,fzz,X,Y,Z, &
                 SYM,SYM,SYM,symmetry,Lev)
 
-  !$OMP PARALLEL
-  !$OMP WORKSHARE
   gxxx = (gupxx * chix + gupxy * chiy + gupxz * chiz)/chin1
   gxxy = (gupxy * chix + gupyy * chiy + gupyz * chiz)/chin1
   gxxz = (gupxz * chix + gupyz * chiy + gupzz * chiz)/chin1
@@ -813,8 +789,6 @@
 !!!! gauge variable part
 
   Lap_rhs = -TWO*alpn1*trK
-  !$OMP END WORKSHARE
-  !$OMP END PARALLEL
 #if (GAUGE == 0)
   betax_rhs = FF*dtSfx
   betay_rhs = FF*dtSfy

AMSS_NCKU_source/enforce_algebra.f90

@@ -19,60 +19,48 @@
 
 !~~~~~~~> Local variable:
   
-  integer :: i,j,k
+  real*8, dimension(ex(1),ex(2),ex(3)) :: trA,detg
-  real*8 :: lgxx,lgyy,lgzz,ldetg
+  real*8, dimension(ex(1),ex(2),ex(3)) :: gxx,gyy,gzz 
-  real*8 :: lgupxx,lgupxy,lgupxz,lgupyy,lgupyz,lgupzz
+  real*8, dimension(ex(1),ex(2),ex(3)) :: gupxx,gupxy,gupxz,gupyy,gupyz,gupzz
-  real*8 :: ltrA,lscale
   real*8, parameter :: F1o3 = 1.D0 / 3.D0, ONE = 1.D0, TWO = 2.D0
 
 !~~~~~~>
 
-  do k=1,ex(3)
+  gxx = dxx + ONE
-  do j=1,ex(2)
+  gyy = dyy + ONE
-  do i=1,ex(1)
+  gzz = dzz + ONE
 
-    lgxx = dxx(i,j,k) + ONE
+  detg =  gxx * gyy * gzz + gxy * gyz * gxz + gxz * gxy * gyz - &
-    lgyy = dyy(i,j,k) + ONE
+          gxz * gyy * gxz - gxy * gxy * gzz - gxx * gyz * gyz
-    lgzz = dzz(i,j,k) + ONE
+  gupxx =   ( gyy * gzz - gyz * gyz ) / detg
+  gupxy = - ( gxy * gzz - gyz * gxz ) / detg
+  gupxz =   ( gxy * gyz - gyy * gxz ) / detg
+  gupyy =   ( gxx * gzz - gxz * gxz ) / detg
+  gupyz = - ( gxx * gyz - gxy * gxz ) / detg
+  gupzz =   ( gxx * gyy - gxy * gxy ) / detg
 
-    ldetg =  lgxx * lgyy * lgzz &
+  trA =         gupxx * Axx + gupyy * Ayy + gupzz * Azz &
-           + gxy(i,j,k) * gyz(i,j,k) * gxz(i,j,k) &
+       + TWO * (gupxy * Axy + gupxz * Axz + gupyz * Ayz)
-           + gxz(i,j,k) * gxy(i,j,k) * gyz(i,j,k) &
-           - gxz(i,j,k) * lgyy * gxz(i,j,k) &
-           - gxy(i,j,k) * gxy(i,j,k) * lgzz &
-           - lgxx * gyz(i,j,k) * gyz(i,j,k)
 
-    lgupxx =   ( lgyy * lgzz - gyz(i,j,k) * gyz(i,j,k) ) / ldetg
+  Axx = Axx - F1o3 * gxx * trA
-    lgupxy = - ( gxy(i,j,k) * lgzz - gyz(i,j,k) * gxz(i,j,k) ) / ldetg
+  Axy = Axy - F1o3 * gxy * trA
-    lgupxz =   ( gxy(i,j,k) * gyz(i,j,k) - lgyy * gxz(i,j,k) ) / ldetg
+  Axz = Axz - F1o3 * gxz * trA
-    lgupyy =   ( lgxx * lgzz - gxz(i,j,k) * gxz(i,j,k) ) / ldetg
+  Ayy = Ayy - F1o3 * gyy * trA
-    lgupyz = - ( lgxx * gyz(i,j,k) - gxy(i,j,k) * gxz(i,j,k) ) / ldetg
+  Ayz = Ayz - F1o3 * gyz * trA
-    lgupzz =   ( lgxx * lgyy - gxy(i,j,k) * gxy(i,j,k) ) / ldetg
+  Azz = Azz - F1o3 * gzz * trA
 
-    ltrA =         lgupxx * Axx(i,j,k) + lgupyy * Ayy(i,j,k) &
+  detg = ONE / ( detg ** F1o3 ) 
-                 + lgupzz * Azz(i,j,k) &
-         + TWO * (lgupxy * Axy(i,j,k) + lgupxz * Axz(i,j,k) &
-                 + lgupyz * Ayz(i,j,k))
   
-    Axx(i,j,k) = Axx(i,j,k) - F1o3 * lgxx * ltrA
+  gxx = gxx * detg
-    Axy(i,j,k) = Axy(i,j,k) - F1o3 * gxy(i,j,k) * ltrA
+  gxy = gxy * detg
-    Axz(i,j,k) = Axz(i,j,k) - F1o3 * gxz(i,j,k) * ltrA
+  gxz = gxz * detg
-    Ayy(i,j,k) = Ayy(i,j,k) - F1o3 * lgyy * ltrA
+  gyy = gyy * detg
-    Ayz(i,j,k) = Ayz(i,j,k) - F1o3 * gyz(i,j,k) * ltrA
+  gyz = gyz * detg
-    Azz(i,j,k) = Azz(i,j,k) - F1o3 * lgzz * ltrA
+  gzz = gzz * detg
 
-    lscale = ONE / ( ldetg ** F1o3 )
+  dxx = gxx - ONE
-
+  dyy = gyy - ONE
-    dxx(i,j,k) = lgxx * lscale - ONE
+  dzz = gzz - ONE
-    gxy(i,j,k) = gxy(i,j,k) * lscale
-    gxz(i,j,k) = gxz(i,j,k) * lscale
-    dyy(i,j,k) = lgyy * lscale - ONE
-    gyz(i,j,k) = gyz(i,j,k) * lscale
-    dzz(i,j,k) = lgzz * lscale - ONE
-
-  enddo
-  enddo
-  enddo
 
   return
 
@@ -95,70 +83,50 @@
 
 !~~~~~~~> Local variable:
   
-  integer :: i,j,k
+  real*8, dimension(ex(1),ex(2),ex(3)) :: trA
-  real*8 :: lgxx,lgyy,lgzz,lscale
+  real*8, dimension(ex(1),ex(2),ex(3)) :: gxx,gyy,gzz 
-  real*8 :: lgxy,lgxz,lgyz
+  real*8, dimension(ex(1),ex(2),ex(3)) :: gupxx,gupxy,gupxz,gupyy,gupyz,gupzz
-  real*8 :: lgupxx,lgupxy,lgupxz,lgupyy,lgupyz,lgupzz
-  real*8 :: ltrA
   real*8, parameter :: F1o3 = 1.D0 / 3.D0, ONE = 1.D0, TWO = 2.D0
 
 !~~~~~~>
 
-  do k=1,ex(3)
+  gxx = dxx + ONE
-  do j=1,ex(2)
+  gyy = dyy + ONE
-  do i=1,ex(1)
+  gzz = dzz + ONE
+! for g
+  gupzz =  gxx * gyy * gzz + gxy * gyz * gxz + gxz * gxy * gyz - &
+           gxz * gyy * gxz - gxy * gxy * gzz - gxx * gyz * gyz
 
-! for g: normalize determinant first
+  gupzz = ONE / ( gupzz ** F1o3 ) 
-    lgxx = dxx(i,j,k) + ONE
-    lgyy = dyy(i,j,k) + ONE
-    lgzz = dzz(i,j,k) + ONE
-    lgxy = gxy(i,j,k)
-    lgxz = gxz(i,j,k)
-    lgyz = gyz(i,j,k)
   
-    lscale =  lgxx * lgyy * lgzz + lgxy * lgyz * lgxz &
+  gxx = gxx * gupzz
-            + lgxz * lgxy * lgyz - lgxz * lgyy * lgxz &
+  gxy = gxy * gupzz
-            - lgxy * lgxy * lgzz - lgxx * lgyz * lgyz
+  gxz = gxz * gupzz
+  gyy = gyy * gupzz
+  gyz = gyz * gupzz
+  gzz = gzz * gupzz
 
-    lscale = ONE / ( lscale ** F1o3 )
+  dxx = gxx - ONE
+  dyy = gyy - ONE
+  dzz = gzz - ONE
+! for A  
 
-    lgxx = lgxx * lscale
+  gupxx =   ( gyy * gzz - gyz * gyz )
-    lgxy = lgxy * lscale
+  gupxy = - ( gxy * gzz - gyz * gxz )
-    lgxz = lgxz * lscale
+  gupxz =   ( gxy * gyz - gyy * gxz )
-    lgyy = lgyy * lscale
+  gupyy =   ( gxx * gzz - gxz * gxz )
-    lgyz = lgyz * lscale
+  gupyz = - ( gxx * gyz - gxy * gxz )
-    lgzz = lgzz * lscale
+  gupzz =   ( gxx * gyy - gxy * gxy )
 
-    dxx(i,j,k) = lgxx - ONE
+  trA =         gupxx * Axx + gupyy * Ayy + gupzz * Azz &
-    gxy(i,j,k) = lgxy
+       + TWO * (gupxy * Axy + gupxz * Axz + gupyz * Ayz)
-    gxz(i,j,k) = lgxz
-    dyy(i,j,k) = lgyy - ONE
-    gyz(i,j,k) = lgyz
-    dzz(i,j,k) = lgzz - ONE
 
-! for A: trace-free using normalized metric (det=1, no division needed)
+  Axx = Axx - F1o3 * gxx * trA
-    lgupxx =   ( lgyy * lgzz - lgyz * lgyz )
+  Axy = Axy - F1o3 * gxy * trA
-    lgupxy = - ( lgxy * lgzz - lgyz * lgxz )
+  Axz = Axz - F1o3 * gxz * trA
-    lgupxz =   ( lgxy * lgyz - lgyy * lgxz )
+  Ayy = Ayy - F1o3 * gyy * trA
-    lgupyy =   ( lgxx * lgzz - lgxz * lgxz )
+  Ayz = Ayz - F1o3 * gyz * trA
-    lgupyz = - ( lgxx * lgyz - lgxy * lgxz )
+  Azz = Azz - F1o3 * gzz * trA
-    lgupzz =   ( lgxx * lgyy - lgxy * lgxy )
-
-    ltrA =         lgupxx * Axx(i,j,k) + lgupyy * Ayy(i,j,k) &
-                 + lgupzz * Azz(i,j,k) &
-         + TWO * (lgupxy * Axy(i,j,k) + lgupxz * Axz(i,j,k) &
-                 + lgupyz * Ayz(i,j,k))
-
-    Axx(i,j,k) = Axx(i,j,k) - F1o3 * lgxx * ltrA
-    Axy(i,j,k) = Axy(i,j,k) - F1o3 * lgxy * ltrA
-    Axz(i,j,k) = Axz(i,j,k) - F1o3 * lgxz * ltrA
-    Ayy(i,j,k) = Ayy(i,j,k) - F1o3 * lgyy * ltrA
-    Ayz(i,j,k) = Ayz(i,j,k) - F1o3 * lgyz * ltrA
-    Azz(i,j,k) = Azz(i,j,k) - F1o3 * lgzz * ltrA
-
-  enddo
-  enddo
-  enddo
 
   return
 

AMSS_NCKU_source/fmisc.f90

@@ -324,6 +324,7 @@ subroutine symmetry_bd(ord,extc,func,funcc,SoA)
 
   integer::i
 
+  funcc = 0.d0
   funcc(1:extc(1),1:extc(2),1:extc(3)) = func
    do i=0,ord-1
       funcc(-i,1:extc(2),1:extc(3)) = funcc(i+2,1:extc(2),1:extc(3))*SoA(1)
@@ -349,6 +350,7 @@ subroutine symmetry_tbd(ord,extc,func,funcc,SoA)
 
   integer::i
 
+  funcc = 0.d0
   funcc(1:extc(1),1:extc(2),1:extc(3)) = func
    do i=0,ord-1
       funcc(-i,1:extc(2),1:extc(3)) = funcc(i+2,1:extc(2),1:extc(3))*SoA(1)
@@ -377,6 +379,7 @@ subroutine symmetry_stbd(ord,extc,func,funcc,SoA)
 
   integer::i
 
+  funcc = 0.d0
   funcc(1:extc(1),1:extc(2),1:extc(3)) = func
    do i=0,ord-1
       funcc(-i,1:extc(2),1:extc(3)) = funcc(i+2,1:extc(2),1:extc(3))*SoA(1)
@@ -881,18 +884,10 @@ subroutine symmetry_bd(ord,extc,func,funcc,SoA)
   real*8, dimension(-ord+1:extc(1),-ord+1:extc(2),-ord+1:extc(3)),intent(out):: funcc
   real*8, dimension(1:3), intent(in) :: SoA
 
-  integer::i,j,k
+  integer::i
-
-  !$OMP PARALLEL DO COLLAPSE(2) SCHEDULE(static) PRIVATE(i,j,k)
-  do k=1,extc(3)
-  do j=1,extc(2)
-  do i=1,extc(1)
-     funcc(i,j,k) = func(i,j,k)
-  enddo
-  enddo
-  enddo
-  !$OMP END PARALLEL DO
 
+  funcc = 0.d0
+  funcc(1:extc(1),1:extc(2),1:extc(3)) = func
    do i=0,ord-1
       funcc(-i,1:extc(2),1:extc(3)) = funcc(i+1,1:extc(2),1:extc(3))*SoA(1)
    enddo
@@ -917,6 +912,7 @@ subroutine symmetry_tbd(ord,extc,func,funcc,SoA)
 
   integer::i
 
+  funcc = 0.d0
   funcc(1:extc(1),1:extc(2),1:extc(3)) = func
    do i=0,ord-1
       funcc(-i,1:extc(2),1:extc(3)) = funcc(i+1,1:extc(2),1:extc(3))*SoA(1)
@@ -945,6 +941,7 @@ subroutine symmetry_stbd(ord,extc,func,funcc,SoA)
 
   integer::i
 
+  funcc = 0.d0
   funcc(1:extc(1),1:extc(2),1:extc(3)) = func
    do i=0,ord-1
       funcc(-i,1:extc(2),1:extc(3)) = funcc(i+1,1:extc(2),1:extc(3))*SoA(1)
@@ -1120,8 +1117,7 @@ end subroutine d2dump
 !------------------------------------------------------------------------------
 ! Lagrangian polynomial interpolation
 !------------------------------------------------------------------------------
-
+ subroutine polint(xa, ya, x, y, dy, ordn)
-  subroutine polint(xa, ya, x, y, dy, ordn)
   implicit none
 
   integer, intent(in) :: ordn
@@ -1133,6 +1129,7 @@ end subroutine d2dump
   real*8, dimension(ordn) :: c, d, ho
   real*8 :: dif, dift, hp, h, den_val
 
+  ! Initialization
   c = ya
   d = ya
   ho = xa - x
@@ -1140,6 +1137,7 @@ end subroutine d2dump
   ns = 1
   dif = abs(x - xa(1))
   
+  ! Find the index of the closest table entry
   do i = 2, ordn
     dift = abs(x - xa(i))
     if (dift < dif) then
@@ -1151,6 +1149,7 @@ end subroutine d2dump
   y = ya(ns)
   ns = ns - 1
   
+  ! Main Neville's algorithm loop
   do m = 1, ordn - 1
     n_m = ordn - m
     do i = 1, n_m
@@ -1158,18 +1157,22 @@ end subroutine d2dump
       h  = ho(i+m)
       den_val = hp - h
       
+      ! Check for division by zero locally
       if (den_val == 0.0d0) then
         write(*,*) 'failure in polint for point',x
         write(*,*) 'with input points: ',xa
         stop
       end if
       
+      ! Reuse den_val to avoid redundant divisions
       den_val = (c(i+1) - d(i)) / den_val
       
+      ! Update c and d in place
       d(i) = h * den_val
       c(i) = hp * den_val
     end do
 
+    ! Decide which path (up or down the tableau) to take
     if (2 * ns < n_m) then
       dy = c(ns + 1)
     else
@@ -1186,89 +1189,65 @@ end subroutine d2dump
 ! interpolation in 2 dimensions, follow yx order
 !
 !------------------------------------------------------------------------------
-  subroutine polin2(x1a,x2a,ya,x1,x2,y,dy,ordn)
+subroutine polin2(x1a,x2a,ya,x1,x2,y,dy,ordn)
-  implicit none
+    implicit none
+    integer,intent(in) :: ordn
+    real*8, dimension(ordn), intent(in) :: x1a,x2a
+    real*8, dimension(ordn,ordn), intent(in) :: ya
+    real*8, intent(in) :: x1,x2
+    real*8, intent(out) :: y,dy
 
-  integer,intent(in) :: ordn
+    integer  :: j
-  real*8, dimension(1:ordn), intent(in) :: x1a,x2a
+    real*8, dimension(ordn) :: ymtmp
-  real*8, dimension(1:ordn,1:ordn), intent(in) :: ya
+    real*8 :: dy_temp ! Local variable to prevent overwriting result
-  real*8, intent(in) :: x1,x2
-  real*8, intent(out) :: y,dy
 
-#ifdef POLINT_LEGACY_ORDER
+    ! Optimized sequence: Loop over columns (j) 
-  integer  :: i,m
+    ! ya(:,j) is a contiguous memory block in Fortran
-  real*8, dimension(ordn) :: ymtmp
+    do j=1,ordn
-  real*8, dimension(ordn) :: yntmp
+      call polint(x1a, ya(:,j), x1, ymtmp(j), dy_temp, ordn)
+    end do
 
-  m=size(x1a)
+    ! Final interpolation on the results
-  do i=1,m
+    call polint(x2a, ymtmp, x2, y, dy, ordn)
-    yntmp=ya(i,:)
-    call polint(x2a,yntmp,x2,ymtmp(i),dy,ordn)
-  end do
-  call polint(x1a,ymtmp,x1,y,dy,ordn)
-#else
-  integer  :: j
-  real*8, dimension(ordn) :: ymtmp
-  real*8 :: dy_temp
 
-  do j=1,ordn
+    return
-    call polint(x1a, ya(:,j), x1, ymtmp(j), dy_temp, ordn)
-  end do
-  call polint(x2a, ymtmp, x2, y, dy, ordn)
-#endif
-
-  return
   end subroutine polin2
 !------------------------------------------------------------------------------
 !
 ! interpolation in 3 dimensions, follow zyx order
 !
 !------------------------------------------------------------------------------
-  subroutine polin3(x1a,x2a,x3a,ya,x1,x2,x3,y,dy,ordn)
+subroutine polin3(x1a,x2a,x3a,ya,x1,x2,x3,y,dy,ordn)
-  implicit none
+    implicit none
+    integer,intent(in) :: ordn
+    real*8, dimension(ordn), intent(in) :: x1a,x2a,x3a
+    real*8, dimension(ordn,ordn,ordn), intent(in) :: ya
+    real*8, intent(in) :: x1,x2,x3
+    real*8, intent(out) :: y,dy
 
-  integer,intent(in) :: ordn
+    integer  :: j, k
-  real*8, dimension(1:ordn), intent(in) :: x1a,x2a,x3a
+    real*8, dimension(ordn,ordn) :: yatmp
-  real*8, dimension(1:ordn,1:ordn,1:ordn), intent(in) :: ya
+    real*8, dimension(ordn) :: ymtmp
-  real*8, intent(in) :: x1,x2,x3
+    real*8 :: dy_temp
-  real*8, intent(out) :: y,dy
 
-#ifdef POLINT_LEGACY_ORDER
+    ! Sequence change: Process the contiguous first dimension (x1) first.
-  integer  :: i,j,m,n
+    ! We loop through the 'slow' planes (j, k) to extract 'fast' columns.
-  real*8, dimension(ordn,ordn) :: yatmp
+    do k=1,ordn
-  real*8, dimension(ordn) :: ymtmp
+      do j=1,ordn
-  real*8, dimension(ordn) :: yntmp
+        ! ya(:,j,k) is contiguous; much faster than ya(i,j,:)
-  real*8, dimension(ordn) :: yqtmp
+        call polint(x1a, ya(:,j,k), x1, yatmp(j,k), dy_temp, ordn)
-
+      end do
-  m=size(x1a)
-  n=size(x2a)
-  do i=1,m
-   do j=1,n
-    yqtmp=ya(i,j,:)
-    call polint(x3a,yqtmp,x3,yatmp(i,j),dy,ordn)
-   end do
-    yntmp=yatmp(i,:)
-    call polint(x2a,yntmp,x2,ymtmp(i),dy,ordn)
-  end do
-  call polint(x1a,ymtmp,x1,y,dy,ordn)
-#else
-  integer  :: j, k
-  real*8, dimension(ordn,ordn) :: yatmp
-  real*8, dimension(ordn) :: ymtmp
-  real*8 :: dy_temp
-
-  do k=1,ordn
-    do j=1,ordn
-      call polint(x1a, ya(:,j,k), x1, yatmp(j,k), dy_temp, ordn)
     end do
-  end do
-  do k=1,ordn
-    call polint(x2a, yatmp(:,k), x2, ymtmp(k), dy_temp, ordn)
-  end do
-  call polint(x3a, ymtmp, x3, y, dy, ordn)
-#endif
 
-  return
+    ! Now process the second dimension
+    do k=1,ordn
+      call polint(x2a, yatmp(:,k), x2, ymtmp(k), dy_temp, ordn)
+    end do
+
+    ! Final dimension
+    call polint(x3a, ymtmp, x3, y, dy, ordn)
+
+    return
   end subroutine polin3
 !--------------------------------------------------------------------------------------
 ! calculate L2norm  
@@ -1288,9 +1267,7 @@ end subroutine d2dump
   real*8            :: dX, dY, dZ
   integer::imin,jmin,kmin
   integer::imax,jmax,kmax
-  integer::i,j,k,n_elements
+  integer::i,j,k
-  real*8, dimension(:), allocatable :: f_flat
-  real*8, external :: DDOT
 
   dX = X(2) - X(1)
   dY = Y(2) - Y(1)
@@ -1314,12 +1291,7 @@ if(dabs(X(1)-xmin) < dX) imin = 1
 if(dabs(Y(1)-ymin) < dY) jmin = 1
 if(dabs(Z(1)-zmin) < dZ) kmin = 1
 
-! Optimized with oneMKL BLAS DDOT for dot product
+f_out = sum(f(imin:imax,jmin:jmax,kmin:kmax)*f(imin:imax,jmin:jmax,kmin:kmax))
-n_elements = (imax-imin+1)*(jmax-jmin+1)*(kmax-kmin+1)
-allocate(f_flat(n_elements))
-f_flat = reshape(f(imin:imax,jmin:jmax,kmin:kmax), [n_elements])
-f_out = DDOT(n_elements, f_flat, 1, f_flat, 1)
-deallocate(f_flat)
 
 f_out = f_out*dX*dY*dZ
 
@@ -1344,9 +1316,7 @@ f_out = f_out*dX*dY*dZ
   real*8            :: dX, dY, dZ
   integer::imin,jmin,kmin
   integer::imax,jmax,kmax
-  integer::i,j,k,n_elements
+  integer::i,j,k
-  real*8, dimension(:), allocatable :: f_flat
-  real*8, external :: DDOT
 
   real*8 :: PIo4
 
@@ -1409,12 +1379,7 @@ if(Symmetry==2)then
   if(dabs(ymin+gw*dY)<dY.and.Y(1)<0.d0) jmin = gw+1
 endif
 
-! Optimized with oneMKL BLAS DDOT for dot product
+f_out = sum(f(imin:imax,jmin:jmax,kmin:kmax)*f(imin:imax,jmin:jmax,kmin:kmax))
-n_elements = (imax-imin+1)*(jmax-jmin+1)*(kmax-kmin+1)
-allocate(f_flat(n_elements))
-f_flat = reshape(f(imin:imax,jmin:jmax,kmin:kmax), [n_elements])
-f_out = DDOT(n_elements, f_flat, 1, f_flat, 1)
-deallocate(f_flat)
 
 f_out = f_out*dX*dY*dZ
 
@@ -1442,8 +1407,6 @@ f_out = f_out*dX*dY*dZ
   integer::imin,jmin,kmin
   integer::imax,jmax,kmax
   integer::i,j,k
-  real*8, dimension(:), allocatable :: f_flat
-  real*8, external :: DDOT
 
   real*8 :: PIo4
 
@@ -1506,12 +1469,11 @@ if(Symmetry==2)then
   if(dabs(ymin+gw*dY)<dY.and.Y(1)<0.d0) jmin = gw+1
 endif
 
-! Optimized with oneMKL BLAS DDOT for dot product
+f_out = sum(f(imin:imax,jmin:jmax,kmin:kmax)*f(imin:imax,jmin:jmax,kmin:kmax))
+
+f_out = f_out
+
 Nout = (imax-imin+1)*(jmax-jmin+1)*(kmax-kmin+1)
-allocate(f_flat(Nout))
-f_flat = reshape(f(imin:imax,jmin:jmax,kmin:kmax), [Nout])
-f_out = DDOT(Nout, f_flat, 1, f_flat, 1)
-deallocate(f_flat)
 
   return
 
@@ -1709,7 +1671,6 @@ deallocate(f_flat)
   real*8, dimension(ORDN,ORDN) :: tmp2
   real*8, dimension(ORDN) :: tmp1
   real*8, dimension(3) :: SoAh
-  real*8, external :: DDOT
 
 ! +1 because c++ gives 0 for first point
   cxB = inds+1  
@@ -1745,21 +1706,20 @@ deallocate(f_flat)
      ya=fh(cxB(1):cxT(1),cxB(2):cxT(2),cxB(3):cxT(3))
   endif 
 
-  ! Optimized with BLAS operations for better performance
-  ! First dimension: z-direction weighted sum
   tmp2=0
   do m=1,ORDN
     tmp2 = tmp2 + coef(2*ORDN+m)*ya(:,:,m)
   enddo
 
-  ! Second dimension: y-direction weighted sum
   tmp1=0
   do m=1,ORDN
     tmp1 = tmp1 + coef(ORDN+m)*tmp2(:,m)
   enddo
 
-  ! Third dimension: x-direction weighted sum using BLAS DDOT
+  f_int=0
-  f_int = DDOT(ORDN, coef(1:ORDN), 1, tmp1, 1)
+  do m=1,ORDN
+    f_int = f_int + coef(m)*tmp1(m)
+  enddo
 
   return
 
@@ -1789,7 +1749,6 @@ deallocate(f_flat)
   real*8, dimension(ORDN,ORDN) :: ya
   real*8, dimension(ORDN) :: tmp1
   real*8, dimension(2) :: SoAh
-  real*8, external :: DDOT
 
 ! +1 because c++ gives 0 for first point
   cxB = inds(1:2)+1  
@@ -1819,14 +1778,15 @@ deallocate(f_flat)
      ya=fh(cxB(1):cxT(1),cxB(2):cxT(2),inds(3))
   endif 
 
-  ! Optimized with BLAS operations
   tmp1=0
   do m=1,ORDN
     tmp1 = tmp1 + coef(ORDN+m)*ya(:,m)
   enddo
 
-  ! Use BLAS DDOT for final weighted sum
+  f_int=0
-  f_int = DDOT(ORDN, coef(1:ORDN), 1, tmp1, 1)
+  do m=1,ORDN
+    f_int = f_int + coef(m)*tmp1(m)
+  enddo
 
   return
 
@@ -1857,7 +1817,6 @@ deallocate(f_flat)
   real*8, dimension(ORDN) :: ya
   real*8 :: SoAh
   integer,dimension(3) :: inds
-  real*8, external :: DDOT
 
 ! +1 because c++ gives 0 for first point
   inds = indsi + 1
@@ -1918,8 +1877,10 @@ deallocate(f_flat)
           write(*,*)"error in global_interpind1d, not recognized dumyd = ",dumyd
   endif
 
-  ! Optimized with BLAS DDOT for weighted sum
+  f_int=0
-  f_int = DDOT(ORDN, coef, 1, ya, 1)
+  do m=1,ORDN
+    f_int = f_int + coef(m)*ya(m)
+  enddo
 
   return
 
@@ -2151,38 +2112,24 @@ deallocate(f_flat)
 
   end function fWigner_d_function
 !----------------------------------
-! Optimized factorial function using lookup table for small N
-! and log-gamma for large N to avoid overflow
   function ffact(N) result(gont)
   implicit none
   integer,intent(in) :: N
 
   real*8 :: gont
-  integer :: i
 
-  ! Lookup table for factorials 0! to 20! (precomputed)
+  integer :: i
-  real*8, parameter, dimension(0:20) :: fact_table = [ &
-    1.d0, 1.d0, 2.d0, 6.d0, 24.d0, 120.d0, 720.d0, 5040.d0, 40320.d0, &
-    362880.d0, 3628800.d0, 39916800.d0, 479001600.d0, 6227020800.d0, &
-    87178291200.d0, 1307674368000.d0, 20922789888000.d0, &
-    355687428096000.d0, 6402373705728000.d0, 121645100408832000.d0, &
-    2432902008176640000.d0 ]
 
 ! sanity check
   if(N < 0)then
      write(*,*) "ffact: error input for factorial"
-     gont = 1.d0
      return
   endif
 
-  ! Use lookup table for small N (fast path)
+  gont = 1.d0
-  if(N <= 20)then
+  do i=1,N
-     gont = fact_table(N)
+     gont = gont*i
-  else
+  enddo
-     ! Use log-gamma function for large N: N! = exp(log_gamma(N+1))
-     ! This avoids overflow and is computed efficiently
-     gont = exp(log_gamma(dble(N+1)))
-  endif
 
   return
 
@@ -2316,3 +2263,4 @@ subroutine find_maximum(ext,X,Y,Z,fun,val,pos,llb,uub)
   return
 
 end subroutine
+

AMSS_NCKU_source/gaussj.C

@@ -16,66 +16,115 @@ using namespace std;
 #include <string.h>
 #include <math.h>
 #endif
-
+/* Linear equation solution by Gauss-Jordan elimination.
-// Intel oneMKL LAPACK interface
-#include <mkl_lapacke.h>
-/* Linear equation solution using Intel oneMKL LAPACK.
 a[0..n-1][0..n-1] is the input matrix. b[0..n-1] is input
 containing the right-hand side vectors. On output a is
 replaced by its matrix inverse, and b is replaced by the
-corresponding set of solution vectors.
+corresponding set of solution vectors */
-
-Mathematical equivalence:
-  Solves: A * x = b  =>  x = A^(-1) * b
-  Original Gauss-Jordan and LAPACK dgesv/dgetri produce identical results
-  within numerical precision. */
 
 int gaussj(double *a, double *b, int n)
 {
-  // Allocate pivot array and workspace
+  double swap;
-  lapack_int *ipiv = new lapack_int[n];
-  lapack_int info;
 
-  // Make a copy of matrix a for solving (dgesv modifies it to LU form)
+  int *indxc, *indxr, *ipiv;
-  double *a_copy = new double[n * n];
+  indxc = new int[n];
-  for (int i = 0; i < n * n; i++) {
+  indxr = new int[n];
-    a_copy[i] = a[i];
+  ipiv = new int[n];
+
+  int i, icol, irow, j, k, l, ll;
+  double big, dum, pivinv, temp;
+
+  for (j = 0; j < n; j++)
+    ipiv[j] = 0;
+  for (i = 0; i < n; i++)
+  {
+    big = 0.0;
+    for (j = 0; j < n; j++)
+      if (ipiv[j] != 1)
+        for (k = 0; k < n; k++)
+        {
+          if (ipiv[k] == 0)
+          {
+            if (fabs(a[j * n + k]) >= big)
+            {
+              big = fabs(a[j * n + k]);
+              irow = j;
+              icol = k;
+            }
+          }
+          else if (ipiv[k] > 1)
+          {
+            cout << "gaussj: Singular Matrix-1" << endl;
+            for (int ii = 0; ii < n; ii++)
+            {
+              for (int jj = 0; jj < n; jj++)
+                cout << a[ii * n + jj] << " ";
+              cout << endl;
+            }
+            return 1; // error return
+          }
+        }
+
+    ipiv[icol] = ipiv[icol] + 1;
+    if (irow != icol)
+    {
+      for (l = 0; l < n; l++)
+      {
+        swap = a[irow * n + l];
+        a[irow * n + l] = a[icol * n + l];
+        a[icol * n + l] = swap;
+      }
+
+      swap = b[irow];
+      b[irow] = b[icol];
+      b[icol] = swap;
+    }
+
+    indxr[i] = irow;
+    indxc[i] = icol;
+
+    if (a[icol * n + icol] == 0.0)
+    {
+      cout << "gaussj: Singular Matrix-2" << endl;
+      for (int ii = 0; ii < n; ii++)
+      {
+        for (int jj = 0; jj < n; jj++)
+          cout << a[ii * n + jj] << " ";
+        cout << endl;
+      }
+      return 1; // error return
+    }
+
+    pivinv = 1.0 / a[icol * n + icol];
+    a[icol * n + icol] = 1.0;
+    for (l = 0; l < n; l++)
+      a[icol * n + l] *= pivinv;
+    b[icol] *= pivinv;
+    for (ll = 0; ll < n; ll++)
+      if (ll != icol)
+      {
+        dum = a[ll * n + icol];
+        a[ll * n + icol] = 0.0;
+        for (l = 0; l < n; l++)
+          a[ll * n + l] -= a[icol * n + l] * dum;
+        b[ll] -= b[icol] * dum;
+      }
   }
 
-  // Step 1: Solve linear system A*x = b using LU decomposition
+  for (l = n - 1; l >= 0; l--)
-  // LAPACKE_dgesv uses column-major by default, but we use row-major
+  {
-  info = LAPACKE_dgesv(LAPACK_ROW_MAJOR, n, 1, a_copy, n, ipiv, b, 1);
+    if (indxr[l] != indxc[l])
-
+      for (k = 0; k < n; k++)
-  if (info != 0) {
+      {
-    cout << "gaussj: Singular Matrix (dgesv info=" << info << ")" << endl;
+        swap = a[k * n + indxr[l]];
-    delete[] ipiv;
+        a[k * n + indxr[l]] = a[k * n + indxc[l]];
-    delete[] a_copy;
+        a[k * n + indxc[l]] = swap;
-    return 1;
+      }
-  }
-
-  // Step 2: Compute matrix inverse A^(-1) using LU factorization
-  // First do LU factorization of original matrix a
-  info = LAPACKE_dgetrf(LAPACK_ROW_MAJOR, n, n, a, n, ipiv);
-
-  if (info != 0) {
-    cout << "gaussj: Singular Matrix (dgetrf info=" << info << ")" << endl;
-    delete[] ipiv;
-    delete[] a_copy;
-    return 1;
-  }
-
-  // Then compute inverse from LU factorization
-  info = LAPACKE_dgetri(LAPACK_ROW_MAJOR, n, a, n, ipiv);
-
-  if (info != 0) {
-    cout << "gaussj: Singular Matrix (dgetri info=" << info << ")" << endl;
-    delete[] ipiv;
-    delete[] a_copy;
-    return 1;
   }
 
+  delete[] indxc;
+  delete[] indxr;
   delete[] ipiv;
-  delete[] a_copy;
 
   return 0;
 }

AMSS_NCKU_source/ilucg.f90

@@ -512,10 +512,11 @@
       IMPLICIT DOUBLE PRECISION (A-H,O-Z)
       DIMENSION V(N),W(N)
 !     SUBROUTINE TO COMPUTE DOUBLE PRECISION VECTOR DOT PRODUCT.
-!     Optimized using Intel oneMKL BLAS ddot
-!     Mathematical equivalence: DGVV = sum_{i=1}^{N} V(i)*W(i)
 
-      DOUBLE PRECISION, EXTERNAL :: DDOT
+      SUM = 0.0D0
-      DGVV = DDOT(N, V, 1, W, 1)
+            DO 10 I = 1,N
+            SUM = SUM + V(I)*W(I)
+10          CONTINUE
+      DGVV = SUM
       RETURN
       END

AMSS_NCKU_source/kodiss.f90

@@ -159,42 +159,36 @@ integer, parameter :: NO_SYMM=0, OCTANT=2
 
   call symmetry_bd(3,ex,f,fh,SoA)
 
-! Interior: all stencil points guaranteed in-bounds
-  !$OMP PARALLEL DO COLLAPSE(2) SCHEDULE(static) PRIVATE(i,j,k)
-  do k=4,ex(3)-3
-  do j=4,ex(2)-3
-  !DIR$ IVDEP
-  do i=4,ex(1)-3
-   f_rhs(i,j,k)       = f_rhs(i,j,k) + eps/cof *( (     &
-                              (fh(i-3,j,k)+fh(i+3,j,k)) - &
-                          SIX*(fh(i-2,j,k)+fh(i+2,j,k)) + &
-                          FIT*(fh(i-1,j,k)+fh(i+1,j,k)) - &
-                          TWT* fh(i,j,k)            )/dX + &
-                                                  (     &
-                              (fh(i,j-3,k)+fh(i,j+3,k)) - &
-                          SIX*(fh(i,j-2,k)+fh(i,j+2,k)) + &
-                          FIT*(fh(i,j-1,k)+fh(i,j+1,k)) - &
-                          TWT* fh(i,j,k)            )/dY + &
-                                                  (     &
-                              (fh(i,j,k-3)+fh(i,j,k+3)) - &
-                          SIX*(fh(i,j,k-2)+fh(i,j,k+2)) + &
-                          FIT*(fh(i,j,k-1)+fh(i,j,k+1)) - &
-                          TWT* fh(i,j,k)            )/dZ )
-  enddo
-  enddo
-  enddo
-  !$OMP END PARALLEL DO
-
-! Boundary shell: original branching logic for points near edges
   do k=1,ex(3)
   do j=1,ex(2)
   do i=1,ex(1)
-  if(i >= 4 .and. i <= ex(1)-3 .and. &
+
-     j >= 4 .and. j <= ex(2)-3 .and. &
-     k >= 4 .and. k <= ex(3)-3) cycle
   if(i-3 >= imin .and. i+3 <= imax .and. &
      j-3 >= jmin .and. j+3 <= jmax .and. &
      k-3 >= kmin .and. k+3 <= kmax) then
+#if 0     
+! x direction
+   f_rhs(i,j,k)       = f_rhs(i,j,k) + eps/dX/cof * (     &
+                              (fh(i-3,j,k)+fh(i+3,j,k)) - &
+                          SIX*(fh(i-2,j,k)+fh(i+2,j,k)) + &
+                          FIT*(fh(i-1,j,k)+fh(i+1,j,k)) - &
+                          TWT* fh(i,j,k)            )
+! y direction
+
+   f_rhs(i,j,k)       = f_rhs(i,j,k) + eps/dY/cof * (     &
+                              (fh(i,j-3,k)+fh(i,j+3,k)) - &
+                          SIX*(fh(i,j-2,k)+fh(i,j+2,k)) + &
+                          FIT*(fh(i,j-1,k)+fh(i,j+1,k)) - &
+                          TWT* fh(i,j,k)            )
+! z direction
+
+   f_rhs(i,j,k)       = f_rhs(i,j,k) + eps/dZ/cof * (     &
+                              (fh(i,j,k-3)+fh(i,j,k+3)) - &
+                          SIX*(fh(i,j,k-2)+fh(i,j,k+2)) + &
+                          FIT*(fh(i,j,k-1)+fh(i,j,k+1)) - &
+                          TWT* fh(i,j,k)            )
+#else
+! calculation order if important ?
    f_rhs(i,j,k)       = f_rhs(i,j,k) + eps/cof *( (     &
                               (fh(i-3,j,k)+fh(i+3,j,k)) - &
                           SIX*(fh(i-2,j,k)+fh(i+2,j,k)) + &
@@ -210,7 +204,9 @@ integer, parameter :: NO_SYMM=0, OCTANT=2
                           SIX*(fh(i,j,k-2)+fh(i,j,k+2)) + &
                           FIT*(fh(i,j,k-1)+fh(i,j,k+1)) - &
                           TWT* fh(i,j,k)            )/dZ )
+#endif
   endif
+
   enddo
   enddo
   enddo

AMSS_NCKU_source/lopsidediff.f90

@@ -233,7 +233,6 @@ subroutine lopsided(ex,X,Y,Z,f,f_rhs,Sfx,Sfy,Sfz,Symmetry,SoA)
 
 ! upper bound set ex-1 only for efficiency, 
 ! the loop body will set ex 0 also
-  !$OMP PARALLEL DO COLLAPSE(2) SCHEDULE(static) PRIVATE(i,j,k)
   do k=1,ex(3)-1
   do j=1,ex(2)-1
   do i=1,ex(1)-1
@@ -483,7 +482,6 @@ subroutine lopsided(ex,X,Y,Z,f,f_rhs,Sfx,Sfy,Sfz,Symmetry,SoA)
   enddo
   enddo
   enddo
-  !$OMP END PARALLEL DO
 
   return
 

AMSS_NCKU_source/macrodef.h

@@ -2,7 +2,7 @@
 #ifndef MICRODEF_H
 #define MICRODEF_H
 
-#include "macrodef.fh"
+#include "microdef.fh"
 
 // application parameters
 

AMSS_NCKU_source/makefile.inc

@@ -8,7 +8,7 @@ filein  = -I/usr/include/ -I${MKLROOT}/include
 
 ## Using sequential MKL (OpenMP disabled for better single-threaded performance)
 ## Added -lifcore for Intel Fortran runtime and -limf for Intel math library
-LDLIBS  = -L${MKLROOT}/lib -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core -lifcore -limf -lpthread -lm -ldl -qopenmp
+LDLIBS  = -L${MKLROOT}/lib -lmkl_intel_lp64 -lmkl_sequential -lmkl_core -lifcore -limf -lpthread -lm -ldl
 
 ## Aggressive optimization flags:
 ## -O3: Maximum optimization
@@ -16,10 +16,10 @@ LDLIBS  = -L${MKLROOT}/lib -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core -lifco
 ## -fp-model fast=2: Aggressive floating-point optimizations
 ## -fma: Enable fused multiply-add instructions
 ## Note: OpenMP has been disabled (-qopenmp removed) due to performance issues
-CXXAPPFLAGS  = -O3 -xHost -fp-model fast=2 -fma -ipo -qopenmp \
+CXXAPPFLAGS  = -O3 -xHost -fp-model fast=2 -fma \
                -Dfortran3 -Dnewc -I${MKLROOT}/include
-f90appflags  = -O3 -xHost -fp-model fast=2 -fma -ipo -qopenmp \
+f90appflags  = -O3 -xHost -fp-model fast=2 -fma \
-               -align array64byte -fpp -I${MKLROOT}/include
+               -fpp -I${MKLROOT}/include
 f90          = ifx
 f77          = ifx
 CXX          = icpx
@@ -30,3 +30,4 @@ Cu = nvcc
 CUDA_LIB_PATH = -L/usr/lib/cuda/lib64 -I/usr/include -I/usr/lib/cuda/include
 #CUDA_APP_FLAGS = -c -g -O3 --ptxas-options=-v -arch compute_13 -code compute_13,sm_13 -Dfortran3 -Dnewc
 CUDA_APP_FLAGS = -c -g -O3 --ptxas-options=-v -Dfortran3 -Dnewc
+

makefile_and_run.py

@@ -11,17 +11,6 @@
 import AMSS_NCKU_Input as input_data
 import subprocess
 
-## CPU core binding configuration using taskset
-## taskset ensures all child processes inherit the CPU affinity mask
-## This forces make and all compiler processes to use only nohz_full cores (4-55, 60-111)
-## Format: taskset -c 4-55,60-111 ensures processes only run on these cores
-NUMACTL_CPU_BIND = "taskset -c 0-111"
-
-## Build parallelism configuration
-## Use nohz_full cores (4-55, 60-111) for compilation: 52 + 52 = 104 cores
-## Set make -j to utilize available cores for faster builds
-BUILD_JOBS = 104
-
 
 ##################################################################
 
@@ -37,11 +26,11 @@ def makefile_ABE():
     print( " Compiling the AMSS-NCKU executable file ABE/ABEGPU " ) 
     print(                                                        )
 
-    ## Build command with CPU binding to nohz_full cores
+    ## Build command
     if (input_data.GPU_Calculation == "no"):
-        makefile_command  = f"{NUMACTL_CPU_BIND} make -j{BUILD_JOBS} ABE"
+        makefile_command  = "make -j4" + " ABE"
     elif (input_data.GPU_Calculation == "yes"):
-        makefile_command  = f"{NUMACTL_CPU_BIND} make -j{BUILD_JOBS} ABEGPU"
+        makefile_command  = "make -j4" + " ABEGPU"
     else:
         print( " CPU/GPU numerical calculation setting is wrong " )
         print(                                                    )
@@ -78,8 +67,8 @@ def makefile_TwoPunctureABE():
     print( " Compiling the AMSS-NCKU executable file TwoPunctureABE " )
     print(                                                            )
     
-    ## Build command with CPU binding to nohz_full cores
+    ## Build command
-    makefile_command = f"{NUMACTL_CPU_BIND} make -j{BUILD_JOBS} TwoPunctureABE"
+    makefile_command = "make" + " TwoPunctureABE"
 
     ## Execute the command with subprocess.Popen and stream output
     makefile_process = subprocess.Popen(makefile_command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, text=True) 
@@ -116,10 +105,10 @@ def run_ABE():
     ## Define the command to run; cast other values to strings as needed
     
     if (input_data.GPU_Calculation == "no"):
-        mpi_command         = NUMACTL_CPU_BIND + " mpirun -np " + str(input_data.MPI_processes) + " ./ABE"
+        mpi_command         = "mpirun -np " + str(input_data.MPI_processes) + " ./ABE"
         mpi_command_outfile = "ABE_out.log"
     elif (input_data.GPU_Calculation == "yes"):
-        mpi_command         = NUMACTL_CPU_BIND + " mpirun -np " + str(input_data.MPI_processes) + " ./ABEGPU"
+        mpi_command         = "mpirun -np " + str(input_data.MPI_processes) + " ./ABEGPU"
         mpi_command_outfile = "ABEGPU_out.log"
  
     ## Execute the MPI command and stream output
@@ -158,7 +147,7 @@ def run_TwoPunctureABE():
     print(                                                          )
     
     ## Define the command to run
-    TwoPuncture_command         = NUMACTL_CPU_BIND + " ./TwoPunctureABE"
+    TwoPuncture_command         = "./TwoPunctureABE"
     TwoPuncture_command_outfile = "TwoPunctureABE_out.log"
 
     ## Execute the command with subprocess.Popen and stream output