SDPB takes preprocessed input generated by sdp2input or pvm2sdp.
If you want to modify these files or generate your own input files
with a separate tool, this documents the format you will need to follow.
sdp2input and pvm2sdp generate a main zip file containing multiple
JSON files. They do not enable compression, because that can be quite
slow. Putting everything into a single file makes it easier to
manage. Also, the zip format has a built-in checksum to detect
corruption.
Inside the zip file, SDPB expects to find control.json,
objectives.json, and two files for every block (block_info_0.json, block_data_0.json,
block_info_1.json, block_data_2.json, ...)
The main part of control.json is listing the number of blocks.
sdp2input and pvm2sdp will also include a copy of the command used
to create the files, but SDPB does not require that.
objectives.json includes the constant contribution to the objective,
and 'b', the optimization vector.
block_info_XXX.json contains the dimension and number of points.
block_data_XXX.bin or block_data_XXX.json contains bilinear
bases, the 'B' matrix, and the 'c' vector.
Block data can be stored either in a human-readable JSON format, or in a more efficient binary format (which
uses Boost.Serialization library,
see write_block_data.cxx for details).
SDPB can also read the input if you pack these files into different
archive formats. It is limited by the support for
libarchive,
but it includes zip, tar, tar.gz, 7z. If you like, you can also leave
everything in a plain old directory.
The JSON schema for these input files are in
sdpb_input_control_schema.json,
sdpb_input_objectives_schema.json,
sdpb_input_block_info_schema.json,
sdpb_input_block_data_schema.json. Running SDPB on the
test example
pvm2sdp 1024 test/data/pvm2sdp/pvm.xml test/out/pvm2sdp/sdp.zip
will generate a simple example you can look at.