def main(args):
#Start with clean set of parameters
all_params={}
#get predictor modules
predictors=User.predictors.get(args.predictors)
predictor_modules=import_predictor_modules(predictors)
all_params.update(predictor_modules)
#FIXME: need to come up with generalised parsing of options to the predictors
#get versions
versions=User.versions.get(args.versions)
for predictor , versions in versions.items():
if predictor==predictors['spectrum_generator']:
try:
all_params[predictor]['version']=versions
except KeyError:
all_params[predictor]={'version':versions}
else:
try:
all_params[predictor]['versions']=versions
except KeyError:
all_params[predictor]={'versions':versions}
#this is afterburner style
if args.mc_cmssm :
all_params.update(inputs.get_mc_cmssm_inputs(*(args.mc_cmssm)))
elif args.mc_cmssm_default:
all_params.update(inputs.get_mc_cmssm_inputs(271.378279475, 920.368119935, 14.4499538001,
-1193.57068242, 173.474173, 91.1877452551, 0.0274821578423))
elif args.mc_neg_mu_cmssm :
all_params.update(inputs.get_mc_neg_mu_cmssm_inputs(*(args.mc_neg_mu_cmssm)))
elif args.mc_nuhm1 :
all_params.update(inputs.get_mc_nuhm1_inputs(*(args.mc_nuhm1)))
elif args.mc_neg_mu_nuhm1 :
all_params.update(inputs.get_mc_neg_mu_nuhm1_inputs(*(args.mc_neg_mu_nuhm1)))
elif args.mc_nuhm1_default :
all_params.update(inputs.get_mc_nuhm1_inputs(237.467776964, 968.808711245, 15.649644, -1858.78698798, -6499529.79661,
173.385870186, 91.1875000682, 0.0274949856504))
elif args.mc_pmssm8 :
all_params.update(inputs.get_mc_pmssm8_inputs(*(args.mc_pmssm8)))
elif args.mc_pmssm10 :
all_params.update(inputs.get_mc_pmssm10_inputs(*(args.mc_pmssm10)))
elif args.mc_pmssm10_default :
all_params.update(inputs.get_mc_pmssm10_inputs(1663.99,1671.75,414.131,294.935,311.199,1712.73,
1841.21,718.489,43.4923,775.09,173.233,91.1874,0.0275018))
elif args.run_softsusy_input_slha:
all_params.update({'SoftSUSY':{'file':args.run_softsusy_input_slha}})
elif args.run_spectrum:
all_params.update({'spectrumfile':args.run_spectrum})
#check for command line input parameters
if args.input_pars is not None:
command_line_dict=eval(args.input_pars)
for key, value in command_line_dict.items():
# if input_pars.get(key) is None:
# input_pars.update
if isinstance(value,dict) and (all_params.get(key) is not None):
all_params[key].update(value)
else:
all_params[key]=value
#check for tmp_dir
if args.tmp_dir:
all_params.update({'tmp_dir':args.tmp_dir})
#print inputs like
if 'inputs' in args.verbose:
print(all_params)
#check verbosity
if args.verbose:
all_params['verbose']=args.verbose
try:
slha_obj, point ,stdouts = POINT.run_point(**all_params)
except TypeError:
print("ERROR: Point failed to run")
exit()
if __name__=="__main__" and slha_obj is None:
print('ERROR: Point fails\nExiting')
exit()
if not args.suppress_chi2_calc:
all_constraints=Constraints_list.constraints
#mc8 data set
try:
data_set=data_sets[args.data_set]
except KeyError:
print("WARNING: \"{}\" invalid data set. No X^2 is calculated".format(args.data_set))
data_set=[]
if not args.suppress_chi2_calc:
constraints={name: all_constraints[name] for name in data_set}
#FIXME: this should become a separate file
#pass this constraints list to the chi2 function
if not args.suppress_chi2_calc:
total, breakdown = Analyse.chi2(point,constraints)
# optional printing
if args.obs:
pp.pprint(point)
if args.breakdown:
Analyse.print_chi2_breakdown(point, constraints,data_set)
# save to root
if args.root_save:
# NOTE: for old_mc_rootstorage, need X^2
point[('tot_X2','all')]=total
root.root_open(args.root_save)
VARS=old_mc_rootstorage.get_VARS(point,point[('m','in_o')])
root.root_write(VARS)
root.root_close()
if args.json_breakdown:
l=[]
for d in data_set:
l.append([d,breakdown[d]])
with open(args.json_breakdown,'w') as f:
json.dump(l,f)
# print only observable keys
if args.observable_keys:
pp.pprint([key for key in point.keys()])
# store observables to piclked file
if args.store_pickle:
with open(args.store_pickle,'wb') as pickle_file:
pickle.dump(point,pickle_file)
#create json file with [('oid1','oid2',array_id), ... ] for storage array ids
if args.create_storage_dict:
point=OrderedDict([(('tot_X2', 'all'),0)]+list(point.items()))
l=[]
i=0
#make list
for key,val in point.items():
oid1,oid2=key
l.append([oid1,oid2,i])
i+=1
#store as json file
with open(args.create_storage_dict,'w') as f:
json.dump(l,f,indent=3)
if args.numpy_out :
if args.storage_dict:
with open(args.storage_dict, 'r') as f:
l=json.load(f)
d={(oid1,oid2):array_id for oid1,oid2, array_id in l}
#start with list of None's
vars=[None]*len(d)
#Fill with values
for oids, val in point.items():
vars[d[oids]]=val
dt=numpy.dtype(len(vars)*[('','f')])
vars=numpy.array(tuple(vars),dtype=dt)
try:
a=numpy.load(args.numpy_out)
print(args.numpy_out,'exists. Appending')
a=numpy.append(a,vars)
except FileNotFoundError:
print('creating: ', args.numpy_out)
a=vars
numpy.save(args.numpy_out,a)
#FIXME: we may want a better way of doing this
return slha_obj, point ,stdouts
param_ranges=get_param_ranges()
#nuitsance central values and sigmas
nuisance_mus_sigmas=[('mt',173.20,0.78),('mz',91.1875,0.0021),('delta_alpha_had',0.02756,0.0001)]
#default X^2 penalty in case of errors
#default_chi=-10*args.log_zero
#lookup dictionary for initiating SLHA() objects
lookup=SLHA().get_lookup()
#constraint objects
all_constraints=Constraints_list.constraints
#pretty printer
my_pprint = pprint.PrettyPrinter(indent=4, depth=3)
#constraints list
data_set=data_sets[args.data_set]
#predictors
predictors=User.predictors.get(args.predictors)
predictor_modules=import_predictor_modules(predictors)
#handle signal from batch etc.
signal.signal(signal.SIGINT, signal_handler)
signal.signal(signal.SIGUSR2, signal_handler)
signal.signal(signal.SIGXCPU, signal_handler)
signal.signal(signal.SIGUSR1, signal_handler)
#use storage dict
if args.storage_dict:
with open(args.storage_dict, 'r') as f:
l=json.load(f)
storage_dict={(oid1,oid2):array_id for oid1,oid2, array_id in l}
###############################################
def soft_flat_prior(X):
#takes number between 0 and 1
#transforms this number to lie between 0 and 1 with probability p
#otherwise ends up in gaussian tails.