Parse hdf5 as inputs

[edoyango]

Currently hardcoded virtual and real particle generation is sub-optimal. Greater coding burden on the user.

Allowing an input hdf5 file to be parsed would make things much easier.

A user could generate one via python/c++/fortran as needed. Maybe helper python package could be created too.

[edoyango]

parallel hdf5 implemented, still need serial/hl version 03bb957

[edoyango]

CAF version now only utilizes hdf5 files to define particle arrangement and initial density, and velocty 6f38500

See example h5dump headers output

$ h5dump -H example/dambreak.h5 
HDF5 "example/dambreak.h5" {
GROUP "/" {
   GROUP "halo" {
      ATTRIBUTE "n" {
         DATATYPE  H5T_STD_I32LE
         DATASPACE  SCALAR
      }
   }
   GROUP "real" {
      ATTRIBUTE "n" {
         DATATYPE  H5T_STD_I32LE
         DATASPACE  SCALAR
      }
      DATASET "ind" {
         DATATYPE  H5T_STD_I32LE
         DATASPACE  SIMPLE { ( 62500 ) / ( 62500 ) }
      }
      DATASET "p" {
         DATATYPE  H5T_IEEE_F64LE
         DATASPACE  SIMPLE { ( 62500 ) / ( 62500 ) }
      }
      DATASET "procid" {
         DATATYPE  H5T_STD_I32LE
         DATASPACE  SIMPLE { ( 62500 ) / ( 62500 ) }
      }
      DATASET "rho" {
         DATATYPE  H5T_IEEE_F64LE
         DATASPACE  SIMPLE { ( 62500 ) / ( 62500 ) }
      }
      DATASET "type" {
         DATATYPE  H5T_STD_I32LE
         DATASPACE  SIMPLE { ( 62500 ) / ( 62500 ) }
      }
      DATASET "v" {
         DATATYPE  H5T_IEEE_F64LE
         DATASPACE  SIMPLE { ( 62500, 3 ) / ( 62500, 3 ) }
      }
      DATASET "x" {
         DATATYPE  H5T_IEEE_F64LE
         DATASPACE  SIMPLE { ( 62500, 3 ) / ( 62500, 3 ) }
      }
   }
   GROUP "virt" {
      ATTRIBUTE "n" {
         DATATYPE  H5T_STD_I32LE
         DATASPACE  SCALAR
      }
      DATASET "ind" {
         DATATYPE  H5T_STD_I32LE
         DATASPACE  SIMPLE { ( 188112 ) / ( 188112 ) }
      }
      DATASET "p" {
         DATATYPE  H5T_IEEE_F64LE
         DATASPACE  SIMPLE { ( 188112 ) / ( 188112 ) }
      }
      DATASET "procid" {
         DATATYPE  H5T_STD_I32LE
         DATASPACE  SIMPLE { ( 188112 ) / ( 188112 ) }
      }
      DATASET "rho" {
         DATATYPE  H5T_IEEE_F64LE
         DATASPACE  SIMPLE { ( 188112 ) / ( 188112 ) }
      }
      DATASET "type" {
         DATATYPE  H5T_STD_I32LE
         DATASPACE  SIMPLE { ( 188112 ) / ( 188112 ) }
      }
      DATASET "v" {
         DATATYPE  H5T_IEEE_F64LE
         DATASPACE  SIMPLE { ( 188112, 3 ) / ( 188112, 3 ) }
      }
      DATASET "x" {
         DATATYPE  H5T_IEEE_F64LE
         DATASPACE  SIMPLE { ( 188112, 3 ) / ( 188112, 3 ) }
      }
   }
}
}

Work still needed:

• README and detailed instructions on webpage
  • how to install hdf5 (apt, from source, spack, brew etc) for both serial and parallel
  • how to compile with hdf5 (setting flags)
  • other considerations e.g., ensuring same Fortran compiler version used
  • explanation of structure of hdf5 file
• Same IO functionality for GPU code
• Potentially unify IO subroutines between src_CAF and src_GPU
• Examples scripts for preparing input file (Python, Matlab)

[edoyango]

Same IO functionality for GPU code DONE 66a7fd2

[edoyango]

Examples scripts for preparing input file (Python, Matlab) DONE 92eed18

[edoyango]

README and detailed instructions on webpage (first version) DONE https://ed-yang.com/docs/docs/input.

Of the tasks I set, I just need to merge the IO subroutines. I think I'll put that into a bigger project of merging similar subroutines.