def extended_distribution(pivotword_i):
pivotword=pivotwords[pivotword_i]
logphrase=identity+'/'+pivotword
try:
logger.info(logphrase+' : starting')
target_path='/data/nrekabsaz/experiments/randomness/'+simfunction+'/extendeddist/'+pivotword+'/'+identity+'.json'
tools._mkdir_recursive(os.path.dirname(target_path))
simword_path='/data/nrekabsaz/experiments/randomness/'+simfunction+'/simword/'+pivotword+'/'+identity+'.json'
logphrase=identity+'/'+pivotword
logger.info(logphrase+' : '+'loading' + simword_path)
with open(simword_path) as frr:
data=json.load(frr)
meanstd=data.values()
logger.info(logphrase+' : '+'calculating meanstdtopdown')
meanstdtopdown=[]
for meanstd_tuple in meanstd:
mean=meanstd_tuple[0]
std=meanstd_tuple[1]
topbrd=mean+3*std
downbrd=mean-3*std
meanstdtopdown.append((mean, std, topbrd, downbrd))
meanstdtopdown.sort(key=lambda x: x[2], reverse=True)#sort by topbrd
meanstd=None
logger.info(logphrase+' : '+'calculating mixture norms')
bins = np.arange(1, -0.2, -.001)
first_dropin_slide_cnt=0
mixpdflist=[]
for bin_i, bin in enumerate(bins):
mixpdf=0
if first_dropin_slide_cnt!=-1:
meanstdtopdown_cnt=first_dropin_slide_cnt
first_dropin_slide_cnt=-1
while (True):
if meanstdtopdown_cnt>=len(meanstdtopdown):
break
meanstdtopdown_tuple=meanstdtopdown[meanstdtopdown_cnt]
if meanstdtopdown_tuple[2]>=bin:
if meanstdtopdown_tuple[3]<=bin:
if first_dropin_slide_cnt==-1:
first_dropin_slide_cnt=meanstdtopdown_cnt
if meanstdtopdown_tuple[1]!=0:
#normdist=norm(loc = meanstdtopdown_tuple[0], scale = meanstdtopdown_tuple[1])
#normdist=normdistlist[meanstdtopdown_cnt]
if bin_i+1<len(bins):
mixpdf+=(norm.cdf(bins[bin_i+1], loc = meanstdtopdown_tuple[0], scale = meanstdtopdown_tuple[1])-
norm.cdf(bins[bin_i], loc = meanstdtopdown_tuple[0], scale = meanstdtopdown_tuple[1]))
else:
mixpdf+=norm.cdf(bins[bin_i], loc = meanstdtopdown_tuple[0], scale = meanstdtopdown_tuple[1])
else:
mixpdf+=1.0
meanstdtopdown_cnt+=1
else:
break
mixpdflist.append(mixpdf)
if bin_i%50==0:
logger.info(logphrase+' : calculating mixture value '+str(bin_i)+'/'+str(len(bins)))
logger.info(logphrase+' : saving into file')
wr=open(target_path,'w')
wr.write(' '.join(map(str, mixpdflist)))
wr.close()
except KeyboardInterrupt:
raise KeyboardInterruptError()
except Exception as e:
msgText=identity+' : extended_distribution('+simfunction+') error!'+'\n'+traceback.format_exc()
mail.sendemail_error(msgText)
logger.error(msgText)
simfunction=argv[0]
outputcode=argv[1]
simwordtype=argv[2]
if simwordtype=='orig':
simword_dir_path='/data/nrekabsaz/similarity/result/randomness/'+simfunction+'/simword/'
target_dir_path='/data/nrekabsaz/similarity/result/randomness/'+simfunction+'/simwordspan/'
elif values_ready[1]=='norm':
simword_dir_path='/data/nrekabsaz/similarity/result/randomness/'+simfunction+'/simwordnorm/'
target_dir_path='/data/nrekabsaz/similarity/result/randomness/'+simfunction+'/simwordspannorm/'
tools._mkdir_recursive(target_dir_path)
try:
logger.info('simwordspan_calc ('+str(outputcode)+') '+simfunction+' started...')
simwordspan_calc(outputcode)
msgText='simwordspan_calc ('+str(outputcode)+') '+simfunction+' finished!'
logger.info(msgText)
except Exception as e:
msgText='run_simwordspan_calc('+str(outputcode)+') '+simfunction+'\n'+traceback.format_exc()
logger.error(msgText)
mail.sendemail_error(msgText)
logger.info('email sent!')
dur = datetime.now() - begin_time
logger.info('Whole Duration: '+str(dur))
logger.info('fertig!');
import sys
from src.misc import mail
if __name__ == "__main__":
argv=sys.argv[1:]
msgtype=argv[0]
msg=argv[1]
if msgtype=='suc':
mail.sendemail_success(msg)
elif msgtype=='err':
mail.sendemail_error(msg)