Random DAG Generator

python3 -m pip install PyYAML matplotlib pandas

python3 main.py -c acc.yaml

For dag_num = 10,000,

python3 main.py --c density.yaml

python3 main.py --c std.yaml

• Accuracy plot: python3 viz/acc.py
• Density plot: python3 viz/density.py

• exp (str): select experiment type (acc, density, std)

• exp_range ([start, end, step]): same as python range()

  • density_range: The values are on a scale of 100 times (Required in density experiment)
  • std_range: The values are on a tenfold scale (Required in std experiment)

• dag_num (int): set the number of DAGs

• instance_num (int): set the number of instances

• core_num (int): set the number of cores

• node_num ([mean, dev]): set the number of nodes between [mean-dev, mean+dev]

• depth ([mean, dev]): set the depth of DAG between [mean-dev, mean+dev]

• exec_t ([mean, dev]): set the execution time of task between [mean-dev, mean+dev]

• backup_ratio (float): execution time ratio of backup node

• sl : Self-looping node's accuracy function is $A(L) = 1 - e^{-L/sl_exp + ln0.3} - \left| N(0, sl_std) \right|$

  • sl_unit (float) : $e_{S, 1}$
  • sl_exp (float)
  • sl_std (float): (Not required in std experiment)

• acceptance_threshold (int): Acceptance threshold for score function

• baseline ([small, large]): loop count for BaseLine Small and BaseLine Large

• density (float): (Not required in density experiment)

• dangling_ratio (float): dangling DAG node # / total node #