def main():
if len(sys.argv) != 4:
print "Error: exactly 3 arguments are required"
config_file_in = sys.argv[1]
config_file_out = sys.argv[2]
masspoint = float(sys.argv[3])
confhandler = ConfigFileHandler()
confhandler.load_configuration(config_file_in)
confhandler.set_field('global', 'mass_point', str(masspoint))
confhandler.save_configuration(config_file_out)
def save_params(out_path, params, evalcnt):
confhandler = ConfigFileHandler()
if os.path.exists(out_path):
confhandler.load_configuration(out_path)
section_name = 'evaluation_' + str(evalcnt)
confhandler.new_section(section_name)
for key, value in params.iteritems():
confhandler.set_field(section_name, key, str(value))
confhandler.save_configuration(out_path)
def punzi_target(WP_VBF2j, WP_VBF1j, WP_WHh, WP_ZHh):
global evalcnt
bin_dir = "/home/llr/cms/wind/cmssw/CMSSW_9_4_2/bin/slc6_amd64_gcc630/"
cost_function_evaluator = "run_WP_evaluator"
output = check_output([
bin_dir + cost_function_evaluator, ref_dir, out_dir,
str(lumi),
str(WP_VBF2j),
str(WP_VBF1j),
str(WP_WHh),
str(WP_ZHh)
])
costval = 0.0
for line in output.split('\n'):
if "cost = " in line:
costval = float(line.replace("cost = ", ""))
break
if math.isnan(costval):
costval = -8.75
# save the sampled point such that later they can be used as exploration points (if the need occurs)
confhandler = ConfigFileHandler()
evaluations_path = out_dir + 'evaluations.txt'
if os.path.exists(evaluations_path):
confhandler.load_configuration(evaluations_path)
print "saving evaluation for iteration " + str(evalcnt)
section_name = 'evaluation_' + str(evalcnt)
confhandler.new_section(section_name)
confhandler.set_field(section_name, 'cost', str(costval))
confhandler.set_field(section_name, 'WP_VBF2j', str(WP_VBF2j))
confhandler.set_field(section_name, 'WP_VBF1j', str(WP_VBF1j))
confhandler.set_field(section_name, 'WP_WHh', str(WP_WHh))
confhandler.set_field(section_name, 'WP_ZHh', str(WP_ZHh))
confhandler.save_configuration(evaluations_path)
evalcnt += 1
return costval
def save_priors(out_path, priors):
# combine all the results into the final prior and save it again
confhandler = ConfigFileHandler()
confhandler.config.optionxform = str
confhandler.new_section('Priors')
confhandler.set_field('Priors', 'VBF_prior', str(1.0))
confhandler.set_field('Priors', 'ggH_prior', str(priors["ggh_prior"]))
confhandler.set_field('Priors', 'ttHlept_prior', str(priors["tthlept_prior"]))
confhandler.set_field('Priors', 'ttHhadr_prior', str(priors["tthhadr_prior"]))
confhandler.set_field('Priors', 'ZHlept_prior', str(priors["zhlept_prior"]))
confhandler.set_field('Priors', 'WHlept_prior', str(priors["whlept_prior"]))
confhandler.set_field('Priors', 'ZHhadr_prior', str(priors["zhhadr_prior"]))
confhandler.set_field('Priors', 'WHhadr_prior', str(priors["whhadr_prior"]))
confhandler.set_field('Priors', 'ZHMET_prior', str(priors["zhmet_prior"]))
confhandler.set_field('Priors', "ZX_prior", str(priors["bkg_prior"]))
confhandler.set_field('Priors', "qq_prior", str(priors["qq_prior"]))
confhandler.save_configuration(out_path)
def main():
def append_variables_raw(confhandler, impdict):
section_name = impdict["discriminant"]
confhandler.new_section(section_name)
periodic_inputs = []
nonperiodic_inputs = []
for key, val in impdict.iteritems():
if key is not "discriminant":
if "phi" in key or "Phi" in key or "xi" in key or "xistar" in key:
periodic_inputs.append(key)
else:
nonperiodic_inputs.append(key)
confhandler.set_field(section_name, "nonperiodic_columns", ConfigFileUtils.serialize_list(nonperiodic_inputs, lambda x: x))
confhandler.set_field(section_name, "periodic_columns", ConfigFileUtils.serialize_list(periodic_inputs, lambda x: x))
def convert_varname(raw):
raw = raw.replace('(', '[')
raw = raw.replace(')', ']')
return raw
def select_input_features_cumulative(H1_stream, H0_stream, discriminant_name, scalar_branches, list_branches, list_pt_limits, confhandler, df_scores, cumulative_threshold = 0.99):
print "using cumulative threshold = " + str(cumulative_threshold)
# temporary fix: for any discriminant that involves Z+X, do not use PF-MET (for the data / MC comparison would otherwise be potentially biased)
if "ZX" in model.name:
print "blocking PFMET for model " + model.name
scalar_branches.remove("PFMET")
print "will select input features from: " + str(scalar_branches + list_branches)
implist = scorer.get_sorted_feature_importance_list(H1_stream, H0_stream, scalar_branches, list_branches, list_pt_limits)
print "implist: " + str(implist)
# now iterate through the sorted list that has been returned and put only the highest-ranked variables, up to the threshold
running_sum = 0
impdict = {}
for key, val in implist:
if running_sum < cumulative_threshold:
impdict[convert_varname(key)] = [val]
running_sum += val
impdict["discriminant"] = discriminant_name
print "impdict: " + str(impdict)
append_variables_raw(confhandler, impdict)
df = df_scores.append(pd.DataFrame.from_dict(impdict))
print str(implist)
return df
if len(sys.argv) != 6:
print "Error: exactly 5 arguments are required"
return
out_dir = sys.argv[1]
campaign_name = sys.argv[2]
MC_path = sys.argv[3]
usemela = sys.argv[4]
threshold = float(sys.argv[5])
# input variables that are stored as lists
list_branches = ["Jet", "ExtraLep"]
# limit pt values for these lists: here, jets are only used if their pt > 30GeV, no restrictions are placed on leptons
list_pt_limits = [30, 0]
# scalar (i.e. non-list) input variables
production_branches = ["PFMET", "nCleanedJetsPt30", "nCleanedJetsPt30BTagged_bTagSF", "nExtraLep",
"ZZMass_masked", "nExtraZ", "Z1Mass", "Z2Mass", "Z1Pt", "Z2Pt", "ZZMassErr", "ZZPt", "ZZEta", "ZZPhi", "Z1Flav", "Z2Flav"]
decay_branches = ["costhetastar", "helphi", "helcosthetaZ1", "helcosthetaZ2", "phistarZ1", "xi", "xistar"]
print "using cumulative_threshold = " + str(threshold)
if "y" in usemela:
MELA_branches = ["D_bkg_ME", "D_VBF2j_ggH_ME", "D_VBF1j_ggH_ME", "D_WHh_ggH_ME", "D_ZHh_ggH_ME", "D_WHh_ZHh_ME", "D_VBF2j_WHh_ME", "D_VBF2j_ZHh_ME"]
else:
MELA_branches = []
scorer = BDTscorer(MC_path)
confhandler = ConfigFileHandler()
df_fscores = pd.DataFrame()
# create a model collection with default input variables and default hyperparameters, just to get a list of all the models and their training data files
mcolls = SimpleModelFactoryDynamic.GenerateSimpleModelCollections(MC_path, input_config_file = None, hyperparam_config_file = None)
# iterate over all models that are contained in the list of model collections
for mcoll in mcolls:
for model_name in mcoll.model_dict.keys():
model = mcoll.model_dict[model_name]
pre = mcoll.preprocessor_dict[model_name]
H1_stream = {}
H0_stream = {}
print "=========================================================================="
print "selecting input variables for model " + model_name
for key, val in mcoll.H1_stream.iteritems():
H1_stream[key] = lambda row, val = val, pre = pre: val(row) and pre.cuts(row)
print "adding preprocessor cuts on top for " + key + ": " + pre.cuts_s
for key, val in mcoll.H0_stream.iteritems():
H0_stream[key] = lambda row, val = val, pre = pre: val(row) and pre.cuts(row)
print "adding preprocessor cuts on top for " + key + ": " + pre.cuts_s
df_fscores = select_input_features_cumulative(H1_stream, H0_stream, model_name, production_branches + decay_branches + MELA_branches, list_branches, list_pt_limits, confhandler, df_fscores, threshold)
print "=========================================================================="
confhandler.save_configuration(os.path.join(out_dir, campaign_name + "_inputs.conf"))
df_fscores.to_csv(os.path.join(out_dir, campaign_name + "_fscore_table_bkg.csv"))
def run_bayesian_optimization(name, eval_file, target, var_ranges, init_points, max_iterations, patience, alpha):
global evalcnt
evalcnt = 0
print "now optimizing the following variables: " + str(var_ranges)
print "alpha = " + str(alpha)
# change the kernel to have a length scale more appropriate to this function
# alpha ... corresponds to the value added to the diagonal elements of the covariance matrix <-> the approximate noise level in the observations
gp_params = {'kernel': ConstantKernel(1.0, (1e-8, 1e2)) * Matern(length_scale = 0.01, length_scale_bounds = (1e-5, 1e5), nu = 1.5),
'alpha': alpha}
bo = BayesianOptimization(target, var_ranges)
# check if a file with previous evaluations of this utility function already exists, if so, use it for initialization
evaluations_path = os.path.join(out_dir, eval_file)
if os.path.exists(evaluations_path):
confhandler = ConfigFileHandler()
confhandler.load_configuration(evaluations_path)
init_dict = {}
for section_name in confhandler.get_sections():
cur_section = confhandler.get_section(section_name)
for key, value in cur_section.iteritems():
# only take those variables that are actually relevant
if key in var_ranges or key == "target":
if key not in init_dict:
init_dict[key] = []
init_dict[key].append(float(value))
evalcnt = int(re.sub('evaluation_', '', confhandler.get_sections()[-1])) + 1
print "resuming " + name + " at evaluation " + str(evalcnt)
init_points_loaded = len(init_dict["target"])
print "found " + str(init_points_loaded) + " initialization points: " + str(init_dict)
bo.initialize(init_dict)
bo.maximize(init_points = max(0, init_points - init_points_loaded), n_iter = 0, acq = 'poi', kappa = 3, xi = xi_scheduler(0.0, max_iterations), **gp_params)
print "initialization done"
else:
bo.maximize(init_points = init_points, n_iter = 0, acq = 'poi', kappa = 3, xi = xi_scheduler(0.0, max_iterations), **gp_params)
cur_iteration = 1
patience_cnt = 0
best_cost = -7.0
for it in range(max_iterations):
cur_xi = xi_scheduler(cur_iteration, max_iterations)
print "cur_iteration = " + str(cur_iteration) + ", using xi = " + str(cur_xi)
cur_iteration += 1
bo.maximize(init_points = 0, n_iter = 1, acq = 'poi', kappa = 3, xi = cur_xi, **gp_params)
# evaluate the current maximum
curval = bo.res['max']
cost = curval['max_val']
curparams = curval['max_params']
confhandler = ConfigFileHandler()
confhandler.config.optionxform = str
confhandler.new_section(name)
confhandler.set_field(name, 'target', str(cost))
for key, val in curparams.iteritems():
confhandler.set_field(name, key, str(val))
confhandler.save_configuration(os.path.join(out_dir, name + '.txt'))
# check if it is time to stop this optimization
if(cost > best_cost):
best_cost = cost
patience_cnt = 0
patience_cnt += 1
if(patience_cnt > patience):
break
return curparams
def main():
def _compute_class_weights_lengths(gen, preprocessor, MC_weighting=False):
# determine the actual size of the available dataset and adjust the sample weights correspondingly
H1_data = gen.H1_collection.get_data(Config.branches, 0.0, 1.0)
H0_data = gen.H0_collection.get_data(Config.branches, 0.0, 1.0)
H1_length = len(preprocessor.process(H1_data).values()[0])
H1_indices = preprocessor.get_last_indices()
H0_length = len(preprocessor.process(H0_data).values()[0])
H0_indices = preprocessor.get_last_indices()
print "H1_length = " + str(H1_length)
print "H0_length = " + str(H0_length)
# if per-sample weighting is enabled, also set up the normalization of the event weights
if MC_weighting:
H1_weight_sum = np.sum(
np.maximum(np.array(H1_data["training_weight"][H1_indices]),
0.0))
H0_weight_sum = np.sum(
np.maximum(np.array(H0_data["training_weight"][H0_indices]),
0.0))
H1_class_weight = float(H0_length) / H1_weight_sum
H0_class_weight = float(H1_length) / H0_weight_sum
else:
# H1_class_weight = 1.0
# H0_class_weight = float(H1_length) / float(H0_length)
H1_class_weight = 1.0 + float(H0_length) / float(H1_length)
H0_class_weight = 1.0 + float(H1_length) / float(H0_length)
return H1_class_weight, H0_class_weight, H1_length, H0_length
# this computes low-level performance metrics for a model collection, i.e. the mean-quare error
# computed on the validation dataset for each discriminant. Since the validation datasets will be held constant,
# this is an easy way to directly compare different models
setting_dir = sys.argv[1]
training_dir = sys.argv[2]
out_dir = sys.argv[3]
# first, need to read in the trained ModelCollection:
mconfhandler = ModelCollectionConfigFileHandler()
mconfhandler.load_configuration(setting_dir + "settings.conf")
mcolls = mconfhandler.GetModelCollection(weightpath=training_dir)
confhandler = ConfigFileHandler()
out_path = out_dir + "model_benchmark.txt"
# for the evaluation, need to proceed in the same way as for training, but evaluate the models on the validation
# data instead of training them on the training data
for mcoll in mcolls:
models, preprocessors, settings = mcoll.get_models(
), mcoll.get_preprocessors(), mcoll.get_settings()
for cur_model, cur_preprocessor, cur_settings in zip(
models, preprocessors, settings):
val_gen = Generator(mcoll.H1_stream,
mcoll.H0_stream,
Config.branches,
preprocessor=cur_preprocessor,
chunks=1,
MC_weighting=False)
val_gen.setup_validation_data()
val_H1_classweight, val_H0_classweight, H1_length, H0_length = _compute_class_weights_lengths(
val_gen, cur_preprocessor, False)
print val_H1_classweight
print val_H0_classweight
print H1_length
print H0_length
val_gen.set_H1_weight(val_H1_classweight)
val_gen.set_H0_weight(val_H0_classweight)
val_gen.set_minimum_length(0)
cur_model.get_keras_model().compile(optimizer=optimizers.Adam(),
loss="mean_squared_error",
metrics=["binary_accuracy"])
res = cur_model.get_keras_model().evaluate_generator(
val_gen.preprocessed_generator(), steps=1)
print "statistics for model " + cur_model.name
print res
print cur_model.get_keras_model().metrics_names
confhandler.new_section(cur_model.name)
confhandler.set_field(cur_model.name, 'H0_val_length',
str(H0_length))
confhandler.set_field(cur_model.name, 'H1_val_length',
str(H1_length))
confhandler.set_field(cur_model.name, 'val_loss', str(res[0]))
confhandler.save_configuration(out_path)
def main():
# runs to check for (good) models (the first one passed is taken as reference run from which the available models
# are taken - it is expected that all others runs also follow this structure):
input_runs = []
print "==================================================================="
print "looking for models in the following runs:"
for campaign_dir in sys.argv[1:-2]:
for run_dir in next(os.walk(campaign_dir))[1]:
if not "bin" in run_dir:
run_path = os.path.join(campaign_dir, run_dir)
print run_path
input_runs.append(run_path)
print "==================================================================="
# output training campaign, this will consist of a combination of the models found in the campaigns listed above, in such a way that the overall performance is optimized
output_run = os.path.join(sys.argv[-1], "optimized")
# where the configuration file for the hyperparameter settings should be stored
hyperparam_output = os.path.join(output_run, "../hyperparameters.conf")
os.makedirs(output_run)
# load the available model names
reference_run = input_runs[0]
available_mcolls = os.walk(os.path.join(reference_run,
"training")).next()[1]
mcolls_winning = []
for mcoll in available_mcolls:
models = os.walk(os.path.join(reference_run, "training",
mcoll)).next()[1]
# load a representative version of the current model collection...
mconfhandler = ModelCollectionConfigFileHandler()
mconfhandler.load_configuration(
os.path.join(reference_run, "settings_training", mcoll,
"settings.conf"))
mcoll_template = mconfhandler.GetModelCollection()[0]
# ... but strip away all the actual model components
mcoll_template.model_dict = {}
mcoll_template.preprocessor_dict = {}
mcoll_template.settings_dict = {}
for model in models:
# compare this model across the different runs
losses = [get_loss(run, mcoll, model) for run in input_runs]
winner = np.argmin(losses)
winning_run = input_runs[winner]
# copy the winning model into the output run
shutil.copytree(
os.path.join(winning_run, "training", mcoll, model),
os.path.join(output_run, "training", mcoll, model))
print "--------------------------------------------"
print " take " + model + " from " + winning_run
print "--------------------------------------------"
# load the winning model to keep track of its settings
mconfhandler = ModelCollectionConfigFileHandler()
mconfhandler.load_configuration(
os.path.join(winning_run, "settings_training", mcoll,
"settings.conf"))
mcoll_winning = mconfhandler.GetModelCollection()[0]
# then pull the winning model over into the template
winning_model = mcoll_winning.model_dict[model]
winning_preprocessor = mcoll_winning.preprocessor_dict[model]
winning_settings = mcoll_winning.settings_dict[model]
mcoll_template.add_model(winning_preprocessor, winning_model,
winning_settings)
mcolls_winning.append(mcoll_template)
# now save the put-together config file also into the output run
mconfhandler = ModelCollectionConfigFileHandler()
mconfhandler.ToConfiguration(mcolls_winning)
mconfhandler.save_configuration(os.path.join(output_run, "settings.conf"))
# now distriute again the training settings, as usual:
distribute_training_settings(output_run + '/')
# now create the hyperparameter config file for each model, taken from the winners
hp_confhandler = ConfigFileHandler()
for mcoll in mcolls_winning:
for model_name, model in mcoll.model_dict.iteritems():
hp_confhandler.new_section(model_name)
hp_confhandler.set_field(
model_name, "hyperparameters",
ConfigFileUtils.serialize_dict(model.hyperparameters,
lambda x: str(x)))
hp_confhandler.save_configuration(hyperparam_output)
print "==================================================================="
print "hyperparameter configuration file written to " + hyperparam_output
print "==================================================================="
def punzi_target(ggH_prior, WHhadr_prior, ZHhadr_prior, WHlept_prior,
ZHlept_prior, ZHMET_prior, ttHhadr_prior, ttHlept_prior):
global evalcnt
bin_dir = "/home/llr/cms/wind/cmssw/CMSSW_9_4_2/bin/slc6_amd64_gcc630/"
cost_function_evaluator = "run_prior_evaluator"
output = check_output([
bin_dir + cost_function_evaluator, run_dir, out_dir, engine,
str(ggH_prior),
str(WHhadr_prior),
str(ZHhadr_prior),
str(WHlept_prior),
str(ZHlept_prior),
str(ZHMET_prior),
str(ttHhadr_prior),
str(ttHlept_prior)
])
costval = 0.0
for line in output.split('\n'):
if "cost = " in line:
costval = float(line.replace("cost = ", ""))
break
if math.isnan(costval):
costval = -8.75
# add a regularization term that prefers default priors (i.e. close to 1.0)
reg_term = 1.0 / 8.0 * (
(ggH_prior - 1.0)**2.0 + (WHhadr_prior - 1.0)**2.0 +
(ZHhadr_prior - 1.0)**2.0 + (WHlept_prior - 1.0)**2.0 +
(ZHlept_prior - 1.0)**2.0 + (ZHMET_prior - 1.0)**2.0 +
(ttHhadr_prior - 1.0)**2.0 + (ttHlept_prior - 1.0)**2.0)
costval -= reg_term * lambda_reg
# save the sampled point such that later they can be used as exploration points (if the need occurs)
confhandler = ConfigFileHandler()
evaluations_path = out_dir + 'evaluations.txt'
if os.path.exists(evaluations_path):
confhandler.load_configuration(evaluations_path)
print "saving evaluation for iteration " + str(evalcnt)
section_name = 'evaluation_' + str(evalcnt)
confhandler.new_section(section_name)
confhandler.set_field(section_name, 'cost', str(costval))
confhandler.set_field(section_name, 'ggH_prior', str(ggH_prior))
confhandler.set_field(section_name, 'WHhadr_prior', str(WHhadr_prior))
confhandler.set_field(section_name, 'ZHhadr_prior', str(ZHhadr_prior))
confhandler.set_field(section_name, 'WHlept_prior', str(WHlept_prior))
confhandler.set_field(section_name, 'ZHlept_prior', str(ZHlept_prior))
confhandler.set_field(section_name, 'ZHMET_prior', str(ZHMET_prior))
confhandler.set_field(section_name, 'ttHhadr_prior',
str(ttHhadr_prior))
confhandler.set_field(section_name, 'ttHlept_prior',
str(ttHlept_prior))
confhandler.save_configuration(evaluations_path)
evalcnt += 1
return costval
def main():
global evalcnt
if len(sys.argv) != 4:
print "Error: exactly 3 arguments are required"
run_dir = sys.argv[1]
out_dir = sys.argv[2]
engine = sys.argv[3]
print run_dir
print out_dir
print engine
# punzi_target_2d = lambda WHlept_prior, ZHlept_prior: punzi_target(ggH_prior_default, WHhadr_prior_default, ZHhadr_prior_default,
# WHlept_prior, ZHlept_prior, ZHMET_prior_default,
# ttHhadr_prior_default, ttHlept_prior_default)
def punzi_target(ggH_prior, WHhadr_prior, ZHhadr_prior, WHlept_prior,
ZHlept_prior, ZHMET_prior, ttHhadr_prior, ttHlept_prior):
global evalcnt
bin_dir = "/home/llr/cms/wind/cmssw/CMSSW_9_4_2/bin/slc6_amd64_gcc630/"
cost_function_evaluator = "run_prior_evaluator"
output = check_output([
bin_dir + cost_function_evaluator, run_dir, out_dir, engine,
str(ggH_prior),
str(WHhadr_prior),
str(ZHhadr_prior),
str(WHlept_prior),
str(ZHlept_prior),
str(ZHMET_prior),
str(ttHhadr_prior),
str(ttHlept_prior)
])
costval = 0.0
for line in output.split('\n'):
if "cost = " in line:
costval = float(line.replace("cost = ", ""))
break
if math.isnan(costval):
costval = -8.75
# add a regularization term that prefers default priors (i.e. close to 1.0)
reg_term = 1.0 / 8.0 * (
(ggH_prior - 1.0)**2.0 + (WHhadr_prior - 1.0)**2.0 +
(ZHhadr_prior - 1.0)**2.0 + (WHlept_prior - 1.0)**2.0 +
(ZHlept_prior - 1.0)**2.0 + (ZHMET_prior - 1.0)**2.0 +
(ttHhadr_prior - 1.0)**2.0 + (ttHlept_prior - 1.0)**2.0)
costval -= reg_term * lambda_reg
# save the sampled point such that later they can be used as exploration points (if the need occurs)
confhandler = ConfigFileHandler()
evaluations_path = out_dir + 'evaluations.txt'
if os.path.exists(evaluations_path):
confhandler.load_configuration(evaluations_path)
print "saving evaluation for iteration " + str(evalcnt)
section_name = 'evaluation_' + str(evalcnt)
confhandler.new_section(section_name)
confhandler.set_field(section_name, 'cost', str(costval))
confhandler.set_field(section_name, 'ggH_prior', str(ggH_prior))
confhandler.set_field(section_name, 'WHhadr_prior', str(WHhadr_prior))
confhandler.set_field(section_name, 'ZHhadr_prior', str(ZHhadr_prior))
confhandler.set_field(section_name, 'WHlept_prior', str(WHlept_prior))
confhandler.set_field(section_name, 'ZHlept_prior', str(ZHlept_prior))
confhandler.set_field(section_name, 'ZHMET_prior', str(ZHMET_prior))
confhandler.set_field(section_name, 'ttHhadr_prior',
str(ttHhadr_prior))
confhandler.set_field(section_name, 'ttHlept_prior',
str(ttHlept_prior))
confhandler.save_configuration(evaluations_path)
evalcnt += 1
return costval
eps = 1e-1
delta = 0.2
bo = BayesianOptimization(
punzi_target, {
'ggH_prior': (1.0 - delta, 1.0 + delta),
'WHhadr_prior': (eps, 1.0),
'ZHhadr_prior': (eps, 1.0),
'WHlept_prior': (eps, 1.0),
'ZHlept_prior': (eps, 1.0),
'ZHMET_prior': (eps, 1.0),
'ttHhadr_prior': (eps, 1.0),
'ttHlept_prior': (eps, 1.0)
})
# bo = BayesianOptimization(punzi_target_2d, {'WHlept_prior': (eps, WHlept_prior_default + delta),
# 'ZHlept_prior': (eps, ZHlept_prior_default + delta)})
# check if a file with previously evaluated points exists, if so, use them for initialization
confhandler = ConfigFileHandler()
evaluations_path = out_dir + 'evaluations.txt'
if os.path.exists(evaluations_path):
confhandler.load_configuration(evaluations_path)
ggH_priors_init = []
WHhadr_priors_init = []
ZHhadr_priors_init = []
WHlept_priors_init = []
ZHlept_priors_init = []
ZHMET_priors_init = []
ttHhadr_priors_init = []
ttHlept_priors_init = []
targets_init = []
for section_name in confhandler.get_sections():
cur_section = confhandler.get_section(section_name)
targets_init.append(float(cur_section['cost']))
ggH_priors_init.append(float(cur_section['ggH_prior']))
WHhadr_priors_init.append(float(cur_section['WHhadr_prior']))
ZHhadr_priors_init.append(float(cur_section['ZHhadr_prior']))
WHlept_priors_init.append(float(cur_section['WHlept_prior']))
ZHlept_priors_init.append(float(cur_section['ZHlept_prior']))
ZHMET_priors_init.append(float(cur_section['ZHMET_prior']))
ttHhadr_priors_init.append(float(cur_section['ttHhadr_prior']))
ttHlept_priors_init.append(float(cur_section['ttHlept_prior']))
init_dict = {
'target': targets_init,
'ggH_prior': ggH_priors_init,
'WHhadr_prior': WHhadr_priors_init,
'ZHhadr_prior': ZHhadr_priors_init,
'WHlept_prior': WHlept_priors_init,
'ZHlept_prior': ZHlept_priors_init,
'ZHMET_prior': ZHMET_priors_init,
'ttHhadr_prior': ttHhadr_priors_init,
'ttHlept_prior': ttHlept_priors_init
}
evalcnt = int(re.sub('evaluation_', '',
confhandler.get_sections()[-1])) + 1
print "resuming at evaluation " + str(evalcnt)
bo.initialize(init_dict)
initialized = True
else:
initialized = False
# change the kernel to have a length scale more appropriate to this function
# alpha ... corresponds to the value added to the diagonal elements of the covariance matrix <-> the approximate noise level in the observations
gp_params = {
'kernel':
1.0 *
Matern(length_scale=0.05, length_scale_bounds=(1e-5, 1e5), nu=1.5),
'alpha':
1e-1
}
# perform the standard initialization and setup
if initialized:
bo.maximize(init_points=0,
n_iter=0,
acq='poi',
kappa=3,
xi=xi_scheduler(0.0),
**gp_params)
else:
bo.maximize(init_points=6,
n_iter=0,
acq='poi',
kappa=3,
xi=xi_scheduler(0.0),
**gp_params)
cur_iteration = 1
for it in range(1000):
cur_iteration += 1
cur_xi = xi_scheduler(cur_iteration)
print "using xi = " + str(cur_xi)
bo.maximize(init_points=6,
n_iter=1,
acq='poi',
kappa=3,
xi=cur_xi,
**gp_params)
# evaluate the current maximum
curval = bo.res['max']
cost = curval['max_val']
priors = curval['max_params']
confhandler = ConfigFileHandler()
confhandler.config.optionxform = str
confhandler.new_section('Priors')
confhandler.set_field('Priors', 'cost', str(cost))
confhandler.set_field('Priors', 'VBF_prior', str(1.0))
for key, val in priors.iteritems():
confhandler.set_field('Priors', key, str(val))
confhandler.save_configuration(out_dir + 'priors.txt')
def main():
global evalcnt
if len(sys.argv) != 4:
print "Error: exactly 3 arguments are required"
ref_dir = sys.argv[1]
out_dir = sys.argv[2]
lumi = float(sys.argv[3])
print ref_dir
print out_dir
print lumi
def punzi_target(WP_VBF2j, WP_VBF1j, WP_WHh, WP_ZHh):
global evalcnt
bin_dir = "/home/llr/cms/wind/cmssw/CMSSW_9_4_2/bin/slc6_amd64_gcc630/"
cost_function_evaluator = "run_WP_evaluator"
output = check_output([
bin_dir + cost_function_evaluator, ref_dir, out_dir,
str(lumi),
str(WP_VBF2j),
str(WP_VBF1j),
str(WP_WHh),
str(WP_ZHh)
])
costval = 0.0
for line in output.split('\n'):
if "cost = " in line:
costval = float(line.replace("cost = ", ""))
break
if math.isnan(costval):
costval = -8.75
# save the sampled point such that later they can be used as exploration points (if the need occurs)
confhandler = ConfigFileHandler()
evaluations_path = out_dir + 'evaluations.txt'
if os.path.exists(evaluations_path):
confhandler.load_configuration(evaluations_path)
print "saving evaluation for iteration " + str(evalcnt)
section_name = 'evaluation_' + str(evalcnt)
confhandler.new_section(section_name)
confhandler.set_field(section_name, 'cost', str(costval))
confhandler.set_field(section_name, 'WP_VBF2j', str(WP_VBF2j))
confhandler.set_field(section_name, 'WP_VBF1j', str(WP_VBF1j))
confhandler.set_field(section_name, 'WP_WHh', str(WP_WHh))
confhandler.set_field(section_name, 'WP_ZHh', str(WP_ZHh))
confhandler.save_configuration(evaluations_path)
evalcnt += 1
return costval
eps = 1e-3
delta = 0.2
bo = BayesianOptimization(
punzi_target, {
'WP_VBF2j': (eps, 1.0 - eps),
'WP_VBF1j': (eps, 1.0 - eps),
'WP_WHh': (eps, 1.0 - eps),
'WP_ZHh': (eps, 1.0 - eps)
})
# check if a file with previously evaluated points exists, if so, use them for initialization
confhandler = ConfigFileHandler()
evaluations_path = out_dir + 'evaluations.txt'
if os.path.exists(evaluations_path):
confhandler.load_configuration(evaluations_path)
targets_init = []
WP_VBF2j_init = []
WP_VBF1j_init = []
WP_WHh_init = []
WP_ZHh_init = []
for section_name in confhandler.get_sections():
cur_section = confhandler.get_section(section_name)
targets_init.append(float(cur_section['cost']))
WP_VBF2j_init.append(float(cur_section['WP_VBF2j']))
WP_VBF1j_init.append(float(cur_section['WP_VBF1j']))
WP_WHh_init.append(float(cur_section['WP_WHh']))
WP_ZHh_init.append(float(cur_section['WP_ZHh']))
init_dict = {
'target': targets_init,
'WP_VBF2j': WP_VBF2j_init,
'WP_VBF1j': WP_VBF1j_init,
'WP_WHh': WP_WHh_init,
'WP_ZHh': WP_ZHh_init
}
evalcnt = int(re.sub('evaluation_', '',
confhandler.get_sections()[-1])) + 1
print "resuming at evaluation " + str(evalcnt)
bo.initialize(init_dict)
initialized = True
else:
initialized = False
# change the kernel to have a length scale more appropriate to this function
gp_params = {
'kernel':
1.0 *
Matern(length_scale=0.05, length_scale_bounds=(1e-5, 1e5), nu=1.5),
'alpha':
1e-5
}
# perform the standard initialization and setup
if initialized:
bo.maximize(init_points=0,
n_iter=0,
acq='poi',
kappa=3,
xi=xi_scheduler(0.0),
**gp_params)
else:
bo.maximize(init_points=6,
n_iter=0,
acq='poi',
kappa=3,
xi=xi_scheduler(0.0),
**gp_params)
cur_iteration = 1
for it in range(1000):
cur_xi = xi_scheduler(cur_iteration)
cur_iteration += 1
print "using xi = " + str(cur_xi)
bo.maximize(init_points=6,
n_iter=1,
acq='poi',
kappa=3,
xi=cur_xi,
**gp_params)
# evaluate the current maximum
curval = bo.res['max']
cost = curval['max_val']
WPs = curval['max_params']
confhandler = ConfigFileHandler()
confhandler.config.optionxform = str
confhandler.new_section('WPs')
confhandler.set_field('WPs', 'cost', str(cost))
for key, val in WPs.iteritems():
confhandler.set_field('WPs', key, str(val))
confhandler.save_configuration(out_dir + 'WPs.txt')