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 pythonrange()
-
density_range
: The values are on a scale of 100 times (Required indensity
experiment) -
std_range
: The values are on a tenfold scale (Required instd
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 instd
experiment)
-
-
acceptance_threshold
(int): Acceptance threshold for score function -
baseline
([small, large]): loop count forBaseLine Small
andBaseLine Large
-
density
(float): (Not required indensity
experiment) -
dangling_ratio
(float): dangling DAG node # / total node #