[](http://dx.doi.org/10.5281/zenodo.832052) # Regional and sectoral disaggregation of multi-regional input-output tables C++-Implementation of the flexible algorithm for regional and sectoral disaggregation of multi-regional input-output tables as described in: L. Wenz, S.N. Willner, A. Radebach, R. Bierkandt, J.C. Steckel, A. Levermann **[Regional and sectoral disaggregation of multi-regional input-output tables: a flexible algorithm](http://www.pik-potsdam.de/~anders/publications/wenz_willner15.pdf)** *Economic Systems Research* 27 (2015), [DOI: 10.1080/09535314.2014.987731](http://dx.doi.org/10.1080/09535314.2014.987731). It includes a library for handling heterogeneous MRIO tables with up to one level of hierarchy. If you want to use it and have trouble with it just drop me an [email](mailto:sven.willner@pik-potsdam.de). ## Compiling Just use cmake: ``` mkdir build cd build cmake .. ``` to create the `mrio_disaggregate` binary. Compiler has to support C++11. To use `libmrio` as a static library you can include `libmrio.cmake` from `cmake`. ## Dependencies Libmrio depends on my [https://github.com/swillner/cpp-library](CSV parser) and [https://github.com/swillner/settingsnode](Settings wrapper). These are included as subtrees. It also uses [https://github.com/jbeder/yaml-cpp.git](yaml-cpp), which you either need to have installed or get as a submodule: ``` git submodule update --init --recursive ``` The implementation optionally uses the [https://github.com/Unidata/netcdf-cxx4](NetCDF-CXX4-library) (e.g. package `libnetcdf-c++4-dev` in Ubuntu/Debian). Its use is controlled via the cmake option `LIBMRIO_WITH_NETCDF` (e.g. use `ccmake ..`). ## Usage `mrio_disaggregate` expects the path of a YAML control file as parameter (or `-` and the file is read from stdin): - YAML control file see example in `examples/simple`. - Proxy files CSV-files with proxy data. Column numbers depend on proxy level (as documented in the paper). First column: Year; Then columns of either region/sector name or column pairs of region/sector name and index (starting with 0) of subregion/subsector; Then value; Concluding with an optional column given the sum (only applies for GDP and population levels). - Formats Supported input and output formats are `csv`, `mrio` and `netcdf`.