A sparse outranking digraph model for HPC-ranking of big performance tableaux

In the context of the ongoing GDRI-Algodec "Algorithmic Decision Theory", supported o.a. by the CNRS (France) and the FNR (Luxembourg), we develop multicriteria ranking HPC algorithms for large sets of potential decision alternatives: up to several thousand of alternatives evaluated on multiple incommensurable ordinal performance criteria. By using Python3.5 multiprocessing resources and the Digraph3 multicriteria software library, we could linearly rank without ties on the UL HPC gaia-80 machine with 120 single threaded cores and a CPU memory of 2.3 TB, in about four hours (6h05') a huge set of 1'732'051 decision alternatives evaluated on 13 performance criteria by balancing economic, ecological and societal decision objectives. Data input is, on the one side, a 1'732'051 x 13 performance tableau of size 2.6GB, and on the other side, a theoretical outranking space consisting of three trillions (3 x 1012) of pairwise outranking situations. A "small" set of 1000 decision alternatives, in a similar setting, may thus be ranked typically in about 2 seconds.