Open jasonlyik opened 2 months ago
Currently, the target costs (c_optimal) for the QUBO workloads have only been globally solved for graphs of size 10 and 25, while sizes 50 to 5000 are solved with the DWave Tabu sampler. https://github.com/NeuroBench/system_benchmarks/blob/main/QUBO/default_config.csv
Graphs of size 50, 100, 250, 500 can be attempted to be globally solved as well.
Note: up until n=250, d=0.01 matches tabu sampler.
n=250, d=0.05, s=0 is the first instance of finding a larger independent set, and it is also the first to take a considerable runtime.
Currently, the target costs (c_optimal) for the QUBO workloads have only been globally solved for graphs of size 10 and 25, while sizes 50 to 5000 are solved with the DWave Tabu sampler. https://github.com/NeuroBench/system_benchmarks/blob/main/QUBO/default_config.csv
Graphs of size 50, 100, 250, 500 can be attempted to be globally solved as well.