def IDA_postprocess(IM, EDP, limits, bUthd, rec_data, g):
Pex = []
[Nstripe, Nrec] = rec_data
for i in range(0, len(limits)):
for edp in EDP:
if bUthd[i] > 0:
allPex = stat.norm(np.log(limits[i]),
bUthd[i]).cdf(np.log(edp))
Pex.append(allPex.sum() / Nrec)
else:
Pex.append(
np.divide(float(len(edp[np.log(edp) > np.log(limits[i])])),
float(Nrec)))
Pex = np.array(Pex)
Pex = Pex.reshape((len(limits), Nstripe))
# <codecell>
log_meanSa, log_stSa = [], []
for i in range(0, len(Pex)):
data = Pex[i] * Nrec
[mu, sigma] = mle(data.astype(int), IM, Nrec, g)
log_meanSa.append(mu)
log_stSa.append(sigma)
print np.exp(mu), sigma
print 'medians=', [np.exp(ele) for ele in log_meanSa]
print 'total dispersion=', log_stSa
return [log_meanSa, log_stSa, Pex]
def IDA_postprocess(IM, EDP, limits, bUthd, rec_data, g):
Pex = []
[Nstripe,Nrec] = rec_data;
for i in range(0,len(limits)):
for edp in EDP:
if bUthd[i]>0:
allPex = stat.norm(np.log(limits[i]),bUthd[i]).cdf(np.log(edp))
Pex.append(allPex.sum()/Nrec)
else:
Pex.append(np.divide(float(len(edp[np.log(edp)>np.log(limits[i])])),float(Nrec)))
Pex = np.array(Pex)
Pex = Pex.reshape((len(limits),Nstripe))
# <codecell>
log_meanSa, log_stSa = [], []
for i in range(0,len(Pex)):
data = Pex[i]*Nrec
[mu, sigma] = mle(data.astype(int), IM, Nrec,g)
log_meanSa.append(mu)
log_stSa.append(sigma)
print np.exp(mu), sigma
print 'medians=', [np.exp(ele) for ele in log_meanSa]
print 'total dispersion=', log_stSa
return [log_meanSa, log_stSa, Pex]
def fragility_process(dcm, totblg, im, noLS, g):
[fr] = count_to_poe(dcm, totblg)
# Export fragility points to csv
#cd = os.getcwd()
#output_file = cd+'/outputs/cumulative_damage_matrix.csv'
#header = ['IML', 'no.blgs', 'Damage States']
#dat1 = np.zeros(len(fr))
#dat3 = []
#for j in range(0, len(fr)):
# dat1[j] = im.tolist()[j][0]
# dat3.append([fr[j,:].tolist()[n] for n in range(0,noLS+1)]) # I don't know how to print fr in separeted columns
#col_data = [dat1, [totblg]*len(dat1), dat3]
#n_lines = len(fr)
#print_outputs(output_file,header,n_lines,col_data)
#Sort the matrices according to Intensity Measure Levels
dcm = np.matrix(np.empty(fr.shape))
I = np.argsort(im, axis=0)
FR = fr[I, :]
IM = im[I, :]
log_meanSa, log_stSa = [], []
# Fit number of buildings exceeding each level of damage with Maximum Likelihood regression method
for j in range(1, FR.shape[2]):
damage = FR[:, 0, j] * totblg #number of buildings exceeding damage j
dint = damage.astype(int)
nc = dint.transpose()
num_collapse = nc.tolist()
IML = [IM[i, 0, 0] for i in range(0, len(IM))]
[mu, sigma] = mle(num_collapse, IML, totblg, g)
log_meanSa.append(mu)
log_stSa.append(sigma)
print "LS", j
print "theta=", np.exp(mu)
print "beta=", sigma
return [log_meanSa, log_stSa, FR, IML]
def fragility_process(dcm,totblg,im,noLS, g):
[fr] = count_to_poe(dcm,totblg)
# Export fragility points to csv
#cd = os.getcwd()
#output_file = cd+'/outputs/cumulative_damage_matrix.csv'
#header = ['IML', 'no.blgs', 'Damage States']
#dat1 = np.zeros(len(fr))
#dat3 = []
#for j in range(0, len(fr)):
# dat1[j] = im.tolist()[j][0]
# dat3.append([fr[j,:].tolist()[n] for n in range(0,noLS+1)]) # I don't know how to print fr in separeted columns
#col_data = [dat1, [totblg]*len(dat1), dat3]
#n_lines = len(fr)
#print_outputs(output_file,header,n_lines,col_data)
#Sort the matrices according to Intensity Measure Levels
dcm = np.matrix(np.empty(fr.shape))
I = np.argsort(im, axis=0)
FR = fr[I,:]
IM = im[I,:]
log_meanSa, log_stSa = [],[]
# Fit number of buildings exceeding each level of damage with Maximum Likelihood regression method
for j in range(1,FR.shape[2]):
damage = FR[:,0,j]*totblg #number of buildings exceeding damage j
dint = damage.astype(int)
nc = dint.transpose()
num_collapse = nc.tolist()
IML = [IM[i,0,0] for i in range(0,len(IM))]
[mu, sigma]=mle(num_collapse, IML, totblg, g)
log_meanSa.append(mu)
log_stSa.append(sigma)
print "LS", j
print "theta=",np.exp(mu)
print "beta=",sigma
return [log_meanSa, log_stSa, FR, IML]
# -*- coding: utf-8 -*-
"""
Created on Thu May 8 18:18:30 2014
@author: chiaracasotto
"""
import numpy as np
from rmtk.vulnerability.common.mle import mle
#INPUTS
totblg = 100
g = 981
num_collaspe = np.array([0.54, 0.31, 0.31, 0.5, 0.37, 0, 0.46, 0.58,
1, 0.95, 1, 0.88, 0.72, 0.08, 0.65, 0.43,
0.16, 0.59, 0.41, 0.6, 0])*totblg
IML = [49.852, 47.056 , 33.012 , 82.125 , 37.499 , 3.159 , 52.205 , 88.363 ,
290.11 , 241.78 , 273.37 , 151.24 , 92.957 , 40.101 , 97.535 , 54.723 ,
26.081 , 105.61 , 156.81 , 82.401 , 26.807 ]
num_collapse = num_collaspe.astype(int)
[mu, sigma]=mle(num_collapse.tolist(), IML, totblg, g)
print mu, sigma
* totblg
)
IML = [
49.852,
47.056,
33.012,
82.125,
37.499,
3.159,
52.205,
88.363,
290.11,
241.78,
273.37,
151.24,
92.957,
40.101,
97.535,
54.723,
26.081,
105.61,
156.81,
82.401,
26.807,
]
num_collapse = num_collaspe.astype(int)
[mu, sigma] = mle(num_collapse.tolist(), IML, totblg, g)
print mu, sigma
