Compilation of Fortran 90 code that uses PETSc

See also questions close to this topic

• Compiling MPI with make, several namespace errors such as "error: unknown type name ‘using’?

  I'm trying to compile MPI on Xubuntu 16.04 LTS and I'm quite positive I have all of my g++ and gcc includes and libraries already installed and yet I get this weird mess...

      make[3]: Entering directory '/media/verthex/32gig/openmpi-3.1.2/oshmem/shmem/c/profile'
    LN_S     pshmem_init.c
    CC       pshmem_init.lo
  In file included from ../../../../oshmem/include/shmem.h:26:0,
                   from ../../../../oshmem/include/oshmem/constants.h:15,
                   from pshmem_init.c:22:
  /usr/local/include/complex.h:10:1: error: unknown type name ‘using’
   using namespace std;
   ^
  /usr/local/include/complex.h:10:17: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘std’
   using namespace std;
                   ^
  /usr/local/include/complex.h:13:17: fatal error: cmath: No such file or directory
  compilation terminated.
  Makefile:1950: recipe for target 'pshmem_init.lo' failed
  make[3]: *** [pshmem_init.lo] Error 1
  make[3]: Leaving directory '/media/verthex/32gig/openmpi-3.1.2/oshmem/shmem/c/profile'
  Makefile:2012: recipe for target 'all-recursive' failed
  make[2]: *** [all-recursive] Error 1
  make[2]: Leaving directory '/media/verthex/32gig/openmpi-3.1.2/oshmem/shmem/c'
  Makefile:2578: recipe for target 'all-recursive' failed
  make[1]: *** [all-recursive] Error 1
  make[1]: Leaving directory '/media/verthex/32gig/openmpi-3.1.2/oshmem'
  Makefile:1897: recipe for target 'all-recursive' failed
  make: *** [all-recursive] Error 1
  

  This is the configure output towards the end.

  Open MPI configuration:
  -----------------------
  Version: 3.1.2
  Build MPI C bindings: yes
  Build MPI C++ bindings (deprecated): no
  Build MPI Fortran bindings: no
  MPI Build Java bindings (experimental): no
  Build Open SHMEM support: yes
  Debug build: no
  Platform file: (none)
  
  Miscellaneous
  -----------------------
  CUDA support: no
  PMIx support: internal
  
  Transports
  -----------------------
  Cisco usNIC: no
  Cray uGNI (Gemini/Aries): no
  Intel Omnipath (PSM2): no
  Intel SCIF: no
  Intel TrueScale (PSM): no
  Mellanox MXM: no
  Open UCX: no
  OpenFabrics Libfabric: no
  OpenFabrics Verbs: yes
  Portals4: no
  Shared memory/copy in+copy out: yes
  Shared memory/Linux CMA: yes
  Shared memory/Linux KNEM: no
  Shared memory/XPMEM: no
  TCP: yes
  
  Resource Managers
  -----------------------
  Cray Alps: no
  Grid Engine: no
  LSF: no
  Moab: no
  Slurm: yes
  ssh/rsh: yes
  Torque: no
  
• Nothing to be done for dev makefile

  I have a simple Makefil here:

  .PHONY: dev
  
  dev:
          php -S localhost:8000
  

  And this error comme out: Nothing to be done for dev
• How to exclude object files while linking in makefile

  I have few object files which I do not want to be included while linking process.

  OBJ = one.obj tow.obj three.obj four.obj five.obj

  I want to exclude two.obj and three.obj in linking stage. How can I achieve this? Thank you.
• SuperLU dgstrf returns a memory allocation failure when I increase the size of my sparse matrix

  I am creating a FEA program and I am dealing with matrices on the order 290-96 thousand squared or 27-9 billion elements. These matrices are largely sparse so I am using SuperLU to solve them. I have been able to successfully use SuperLU to solve the problem which has matched up with validation data reasonably well with smaller matrices. However, as I increase the size of my matrices, SuperLU's dgstrf function is outputting an info value of roughly 900 million (914459273 one time, 893813121 another).

  The documentation says that this info value is "number of bytes allocated when memory allocation failure occurred, plus A->ncol." However this does not give any information on how to work through this error. What is limiting the memory in this case? Does the library limit the memory? Is it hard coded into the library or is it determined during compilation? Is the memory limited in the compilation of my Fortran code?

  I am writing my code in Fortran and am using the prebuilt c_fortran_dgssv.c file to link with SuperLU. This file does allow the system to "allocate space internally by system malloc" (lwork=0). Is this something that I could change in order to have more space.

  I am calling the code using similar calls as the fortran example.

  nrhs = 1
  ldb = Dim3DFull
  iopt = 1
  call c_fortran_dgssv(iopt,Dim3DFull,TotalNonZeroElements_BCs,nrhs, &
      Global_Matrix_T_Value_BC_CSC,Global_Matrix_T_Row_BC_CSC, &
      Global_Matrix_T_Col_BC_CSC,Global_Temp,ldb,factors,info)
  if (info .eq. 0) then
      write (*,*) 'Factorization succeeded'
  else
      write(*,*) 'INFO from factorization = ', info
  endif
  !Second, solve the system using the existing factors.
  iopt = 2
  call c_fortran_dgssv(iopt,Dim3DFull,TotalNonZeroElements_BCs,nrhs, &
      Global_Matrix_T_Value_BC_CSC,Global_Matrix_T_Row_BC_CSC, &
      Global_Matrix_T_Col_BC_CSC,Global_Temp,ldb,factors,info)
  if (info .eq. 0) then
      write (*,*) 'Solve succeeded'
  else
      write(*,*) 'INFO from triangular solve = ', info
  endif
  !Last, free the storage allocated inside SuperLU
  iopt = 3
  call c_fortran_dgssv(iopt,Dim3DFull,TotalNonZeroElements_BCs,nrhs, &
      Global_Matrix_T_Value_BC_CSC,Global_Matrix_T_Row_BC_CSC, &
      Global_Matrix_T_Col_BC_CSC,Global_Temp,ldb,factors,info)
  
• Can I create an array without variable assignment with a Fortran?

  I want to create an NxN array whose size is the number of N values I enter.

  Can you create an array that has the size of the number I enter, not the same as the existing Fortran variable declaration real*8 Matrix (100, 100)?
• Rank mismatch in array reference at (1) (1/2)

  Can you look into my code and help me out with what's wrong? I keep facing one problem or the other when I work with IF statements in Fortran.

  Another thing is I want to check if the matrices (A*B) are consistent. I'm trying to do that with the IF statements but whenever I try to check whether the column of matrix A equals the row of matrix B I run into errors such as "rank mismatch in array reference". If there are any other problems lurking around in my code can anyone point them out and tell me how to fix them?

  !Computing C=A*B
  
  program matrixmul
  implicit none
  
  real, allocatable, dimension(:,:):: A,B,C
  integer:: m,n,p,i,j
  
  print*, 'allocate the space for the array at run time'
  print*, 'enter the size of the arrays'
  
  read*, m,n,p
  
  allocate(A(m,n),B(n,p),C(m,p))
  
  print*, 'enter matrix elements for A'
  
  do i=1,m
    read*,(A(i,j),j=1,n)
  end do
  
  print*,'enter matrix elements for B'
  
  do i=1,n
    read*,(B(i,j),j=1,p)
  end do 
  
  call matmlply(A,B,C,m,n,p)
  
  print*, 'output',C
  
  deallocate (A,B,C)
  
  end program matrixmul
  
  
  subroutine matmlply(R,P,Q,x,y,z)
  implicit none
  integer:: x, y,z,i,j 
  real,dimension(x,y),intent(in)::R
  real,dimension(y,z),intent(in)::P
  real,dimension(x,z),intent(out)::Q
  
  !Checking if matrix multiplication is possible
  if(R(y)==P(y)) then 
    do i=1,x
      Q(i)=0
      do j=1,y
        Q(i)=Q(i)+R(i,j)*P(j)
      end do 
    end do 
  else 
    print*,'matrices are not conformable'
    print*,'the column of matrix R should be equal to row of matrix P'
  end if
  
  end subroutine matmlply
  
• How to call PetscFinalize from the root processor only?

  In my Fortran program, I call an external function only at the root processor. I have a few checks to ensure that the function worked. If it doesn't work, then I stopthe program and call PetscFinalize. For example:

  if(rank==0) then
     call_status = external_function(x)
  
     ! function failed
     if (call_status/=1) then
        write(6,*) "external_function failed"
        call PetscFinalize(ierr)
        stop
     end if
  end if
  

  However, I noticed that during runtime, if there is an error with external_function, the program hangs and is not finalized properly.

  My question is: what is the appropriate way to terminate the program at the root processor?
• Very simple PETSc Mat construction gives SEGV 11 error - C++

  I want to create a very simple 2x2 identity matrix in C++ using PETSc library. I have searched various tutorials on the internet and found that all of them indicate one way of doing this. However I get a "Caught signal number 11 SEGV: Segmentation Violation" error. This code is part of a much larger program where I need to create some matrices used in conjunction with other elements of it. Moreover, this code is a mere test, in which I am trying to find out how exactly does PETSc work with matrices. I could say I am still in the learning phase.

  The code is the following:

  Mat Y11;
  
   MatCreate(MPI_COMM_WORLD, &Y11);
   MatSetSizes(Y11, PETSC_DECIDE, PETSC_DECIDE, 2, 2);
   std::cout << "set mat type" << std::endl;
   MatSetType(Y11, MATSEQMAIJ);
  
   MatSetFromOptions(Y11);
  
   PetscScalar temp22[] = {std::complex<double>(1.0,0), std::complex<double>(0.0,0), std::complex<double>(0.0,0.0), std::complex<double>(1.0,0.0)};
   PetscInt tempis[] = {0, 1};
   std::cout << "MatSetValues" << std::endl;
   MatSetValues(Y11, 1, &tempis[0], 1, &tempis[0], &temp22[0], INSERT_VALUES);
   MatSetValues(Y11, 1, &tempis[0], 1, &tempis[1], &temp22[1], INSERT_VALUES);
   MatSetValues(Y11, 1, &tempis[1], 1, &tempis[0], &temp22[2], INSERT_VALUES);
   MatSetValues(Y11, 1, &tempis[1], 1, &tempis[1], &temp22[3], INSERT_VALUES);
  
   MatAssemblyBegin(Y11,MAT_FINAL_ASSEMBLY);
   MatAssemblyEnd(Y11,MAT_FINAL_ASSEMBLY);
  
   std::cout << std::endl << "Y11 block:" << std::endl;
   MatView(Y11,PETSC_VIEWER_STDOUT_WORLD);
  

  The error stems from MatSetValues function.

  Could anyone please tell me what could probably be wrong with this code? There are no simple examples like this on the internet (at least of what I can search, in C++). Thank you in advance!