[petsc-users] PETSc binary write format

[Smith, Barry F.]

> On Dec 3, 2018, at 1:57 PM, Sajid Ali <sajidsyed2021 at u.northwestern.edu> wrote:
> 
> Apologies for the error on my part. 
> 
> Does the vector binary format also work the same way :  VEC_FILE_CLASSID (32 bit int), num_elements (32 bit int) , value of the elements in double (num_elements*64 bit double) ?
 
   Yes.

> 
> Are the doubles stored in IEEE_754 format ?

   I don't know the format. From the manual page for PetscBinaryRead() 

Notes:
   PetscBinaryRead() uses byte swapping to work on all machines; the files
   are written to file ALWAYS using big-endian ordering.


> 
> On Mon, Dec 3, 2018 at 1:43 PM Smith, Barry F. <bsmith at mcs.anl.gov> wrote:
> 
>    You saved a Vec to the file, not a Mat.
> 
> 
> 
> > On Dec 3, 2018, at 1:38 PM, Sajid Ali via petsc-users <petsc-users at mcs.anl.gov> wrote:
> >
> > Hi,
> >
> > I ran ex10 from /vec/examples/tutorials and saved the matrix in binary format.
> >
> > Looking at the matrix in binary using xxd, I see
> >
> > [sajid at xrm temp]$ xxd -b vector.dat
> > 0000000: 00000000 00010010 01111011 01001110 00000000 00000000  ..{N..
> > 0000006: 00000000 00010100 00000000 00000000 00000000 00000000  ......
> > 000000c: 00000000 00000000 00000000 00000000 00111111 11110000  ....?.
> > 0000012: 00000000 00000000 00000000 00000000 00000000 00000000  ......
> > 0000018: 01000000 00000000 00000000 00000000 00000000 00000000  @.....
> > 000001e: 00000000 00000000 01000000 00001000 00000000 00000000  .. at ...
> > 0000024: 00000000 00000000 00000000 00000000 01000000 00010000  .... at .
> > 000002a: 00000000 00000000 00000000 00000000 00000000 00000000  ......
> > 0000030: 01000000 00010100 00000000 00000000 00000000 00000000  @.....
> > 0000036: 00000000 00000000 01000000 00011000 00000000 00000000  .. at ...
> > 000003c: 00000000 00000000 00000000 00000000 01000000 00011100  .... at .
> > 0000042: 00000000 00000000 00000000 00000000 00000000 00000000  ......
> > 0000048: 01000000 00100000 00000000 00000000 00000000 00000000  @ ....
> > 000004e: 00000000 00000000 01000000 00100010 00000000 00000000  ..@"..
> > 0000054: 00000000 00000000 00000000 00000000 01000000 00100100  ....@$
> > 000005a: 00000000 00000000 00000000 00000000 00000000 00000000  ......
> > 0000060: 01000000 00100110 00000000 00000000 00000000 00000000  @&....
> > 0000066: 00000000 00000000 01000000 00101000 00000000 00000000  ..@(..
> > 000006c: 00000000 00000000 00000000 00000000 01000000 00101010  ....@*
> > 0000072: 00000000 00000000 00000000 00000000 00000000 00000000  ......
> > 0000078: 01000000 00101100 00000000 00000000 00000000 00000000  @,....
> > 000007e: 00000000 00000000 01000000 00101110 00000000 00000000  .. at ...
> > 0000084: 00000000 00000000 00000000 00000000 01000000 00110000  .... at 0
> > 000008a: 00000000 00000000 00000000 00000000 00000000 00000000  ......
> > 0000090: 01000000 00110001 00000000 00000000 00000000 00000000  @1....
> > 0000096: 00000000 00000000 01000000 00110010 00000000 00000000  .. at 2..
> > 000009c: 00000000 00000000 00000000 00000000 01000000 00110011  .... at 3
> > 00000a2: 00000000 00000000 00000000 00000000 00000000 00000000  ......
> >
> >
> > The format of the binary file is supposed to be
> > MAT_FILE_CLASSID, num_rows, num_cols, -1 followed by data.
> >
> > Converting the first 4 bytes (32 bits) of binary to int, the number
> > 00000000000100100111101101001110 becomes 1211214, but the value of MAT_FILE_CLASSID according to the header file petscmat.h is 1211216 (petsc at 3.10.2.1 built without complex support). What causes this discrepancy ?
> >
> > Also, what are the special characters at the end of every line ? Am i not reading the binary file correctly ?
> >
> >
> > Thank You,
> > Sajid Ali
> > Applied Physics
> > Northwestern University
> 
> 
> 
> -- 
> Sajid Ali
> Applied Physics
> Northwestern University