|Donator||Arne Stolbjerg Drud|
|Root LP Basis||arki001.bas.gz|
the model originates in the metalurgic industry
The first reported solution is due to Anureet Saxena and Egon Balas, reported in Optimizing over the Split Closure, MSRR-674, December 2005. Solving time was about 64 hours.
Bob Bixby reports solving this instance by letting CPLEX run for nearly a month, computing more than 100,000,000 nodes.
Using SCIP with CPLEX 10.0 as LP-Solver Tobias Achterberg solved arki001 in less than 9 hours on an AMD Opteron 2.2 GHz computing only about 3 million nodes. Again strong branching seems to be the key to success, as the following settings were used:
Alkis Vazacopoulos reports solving this instance using XPRESS 2006A in 4 hours on a 2 processor Xeon system, computing about 1,200,000 nodes.