def experimentalDesign():
base_params = {"x": -1.0, "y": -1.0}
S = Strategy.create_strategy('opt_test', 'python3.6 opt_test.py '+sd, '{x} {y}', base_params)
#factorial design
f4e.output = ""
for x,y in itertools.product(*[[-1.0, 0.0, 1.0],[-1.0, 0.0, 1.0]]):
params = {"x": x, "y": y}
if params == base_params: continue
S2 = Strategy.create_strategy('opt_test', 'python3.6 opt_test.py '+sd, '{x} {y}', params)
S = f4e.bestStrategy(S, S2)
f4e.output += str(S.params)
f4e.terminate()
def parameter_tuning(S, param, param_values):
original_value = S.params[param]
original_name = S.name
params = S.params.copy()
for value in param_values:
params[param] = value
if original_value == value:
continue
else:
S2 = Strategy.create_strategy(original_name, S.exe, S.params_str,
params)
S = f4e.bestStrategy(S, S2)
return S
def experimentalDesign():
S = Strategy.create_strategy(
'BSG_CLP', './BSG_CLP',
'--alpha={a} --beta={b} --gamma={g} -p {p} -t 30', {
"a": 0.0,
"b": 0.0,
"g": 0.0,
"p": 0.0
})
f4e.output = ""
S = parameter_tuning(S, "a", [0.0, 1.0, 2.0, 4.0, 8.0])
f4e.output = str(S.params["a"]) + " "
S = parameter_tuning(S, "b", [0.0, 0.5, 1.0, 2.0, 4.0])
f4e.output += str(S.params["b"]) + " "
S = parameter_tuning(S, "g", [0.0, 0.1, 0.2, 0.3, 0.4])
f4e.output += str(S.params["g"]) + " "
S = parameter_tuning(S, "p", [0.00, 0.01, 0.02, 0.03, 0.04])
f4e.output += str(S.params["p"]) + " "
f4e.terminate()
