def main():
# Specify number of poitns to be used in the training set
# Validation data can be provided optionally
ndata = 50
nval = 500
lb = [-15, -15]
ub = [15, 15]
x = np.random.uniform(lb, ub, (ndata, 2))
xval = np.random.uniform(lb, ub, (nval, 2))
z = [0] * ndata
zval = [0] * nval
# specify simulator as examples.sixcamel
sim = examples.ackley
for i in range(ndata):
z[i] = sim(x[i][0], x[i][1])
for i in range(nval):
zval[i] = sim(xval[i][0], xval[i][1])
# Use alamopy's python function wrapper to avoid using ALAMO's I/O format
almsim = alamopy.wrapwriter(sim)
# Call alamo through the alamopy wrapper
res = alamopy.doalamo(x,
z,
xval=xval,
zval=zval,
almname='ackley',
monomialpower=(1, 2, 3, 4, 5, 6),
expfcns=1,
multi2power=(1, 2),
expandoutput=True)
# Calculate confidence intervals
conf_inv = alamopy.almconfidence(res)
print('Model: {}'.format(res['model']))
print('Confidence Intervals : {}'.format(conf_inv['conf_inv']))
def main():
# Specify number of poitns to be used in the training set
# Validation data can be provided optionally
ndata = 20
nval = 100
x = np.random.uniform([-5, 0], [10, 15], (ndata, 2))
xval = np.random.uniform([-5, 0], [10, 15], (nval, 2))
z = [0] * ndata
zval = [0] * nval
# specify simulator as examples.sixcamel
sim = examples.branin
for i in range(ndata):
z[i] = sim(x[i][0], x[i][1])
for i in range(nval):
zval[i] = sim(xval[i][0], xval[i][1])
# Use alamopy's python function wrapper to avoid using ALAMO's I/O format
# Call alamo through the alamopy wrapper
res = alamopy.doalamo(x,
z,
xval=xval,
zval=zval,
almname='branin',
xmin=(-5, 0),
xmax=(10, 15),
monomialpower=(1, 2, 3, 4),
expfcns=1,
multi2power=(1, 2),
expandoutput=True)
conf_inv = alamopy.almconfidence(res)
print('Model: {}'.format(res['model']))
print('Confidence Intervals : {}'.format(conf_inv['conf_inv']))
def test_basic(**kwargs):
x = [[ 0.17361977, -0.44326123], [-0.30192964, 0.68955226],[-1.98112458, -0.75686176],[0.68299634, 0.65170551],[-1.45317364, 0.15018666],[ 1.56528782, -0.58159576] ,[-1.25868712, -0.78324622],[-1.12121003, 0.95724757] ,[ 1.2467326, -0.65611797],[ 1.26489899, -0.45185251]]
z = [-0.58978634828943055, -0.85834512885363479, 4.0241154669754113, 0.91057814668811488, 1.9147616212616931, 0.29103827202206878, 2.4290896722960778, 0.99199475534877579, 0.59688699266830847, 1.167850366995701]
xival = [[5,5], [2,2]]
zival = [5,2]
doalamo(x, z) #, xval=xival, zval=zival, mock=True)
doalamo(x, z, mock=True, xlabels=["T", "V"], zlabels= ["P"])
doalamo(x, z, xval=xival, zval=zival, mock=True, lmo=5)
ndata=10
x = np.random.uniform([-2,-1],[2,1],(ndata,2))
z = [0]*ndata
# specify simulator as examples.sixcamel
sim = examples.sixcamel
for i in range(ndata):
z[i]=sim(x[i][0],x[i][1])
# Use alamopy's python function wrapper to avoid using ALAMO's I/O format
almsim = wrapwriter(sim)
# Call alamo through the alamopy wrapper
res = doalamo(x,z,almname='cam6',monomialpower=(1,2,3,4,5,6), multi2power=(1,2), simulator=almsim, expandoutput=True,maxiter=20, mock=True)#,cvfun=True)
conf_inv = almconfidence(res)
res = doalamo(x, z, xval = xival, zval = zival) #, expandoutput=True) #, mock=True, loo=True) #, xval=xival, zval=zival, mock=True, loo=True)
conf_inv = almconfidence(res)
print('Model: {}'.format(res['model']))
print('Confidence Intervals : {}'.format(conf_inv['conf_inv']))
almplot(res)
try:
deletefile("logscratch")
deletfile("../cam6")
deletefile("../cam6alm.py")
except:
pass
def build_alamo_model(self, xdata, zdata, xlabels = None, zlabels='e', **kwargs):
# results=alamopy.doalamo(xdata,zdata,xlabels=xlbls, zlabels='expr', monomialpower='1,2,3', multi3power='1,2,3', multi2power='1,2,3', ratiopower='1 2', logfcns=1, expfcns=1, savescratch=1, expandoutput=1); #logfcns=1, expfcns=1, ratiopower='1 2 ', multi3power='1 2', multi2power='1 2 '
# results=alamopy.doalamo(xdata,zdata,xlabels=xlbls, zlabels='e', monomialpower='1 2 3', multi3power='1 2 3', multi2power='1 2 3', savescratch=1, expandoutput=1); #logfcns=1, expfcns=1, ratiopower='1 2 ', multi3power='1 2', multi2power='1 2 '
# results=alamopy.doalamo(xdata,zdata, almname='SteamEnthalpy.alm', almopt='steamcustom.alm', monomialpower='1 2 3', multi3power='1 2 3', multi2power='1 2 3', savescratch=1, expandoutput=1); #logfcns=1, expfcns=1, ratiopower='1 2 ', multi3power='1 2', multi2power='1 2 '
if 'almopt' in kwargs:
results=alamopy.doalamo(xdata,zdata, xlabels=xlabels, zlabels='e', almname=self.almname, almopt=kwargs.get("almopt"), monomialpower='1 2 3', multi3power='1 2 3', multi2power='1 2 3', savescratch=1, expandoutput=1); #logfcns=1, expfcns=1, ratiopower='1 2 ', multi3power='1 2', multi2power='1 2 '
else:
results=alamopy.doalamo(xdata,zdata, almname=self.almname, monomialpower='1 2 3', multi3power='1 2 3', multi2power='1 2 3', savescratch=1, expandoutput=1); #logfcns=1, expfcns=1, ratiopower='1 2 ', multi3power='1 2', multi2power='1 2 '
print "Model function: %s" % results['model'];
print "Sum of squared error = %s" % results['ssr']
print "R^2 = %s" % results['R2'];
res = alamopy.almconfidence(results, xdata, zdata);
print "Confidence Interval =%s" % results['conf_inv']
alamopy.almplot(res)
print results['model'].replace("^","**")
alm_model = results['model']
self.models.append(results);
def writemodel(aterm, ccon):
import numpy as np
import os, sys
import alamopy
import ripe
from alamopy import deletefile
data, rspace = ripe.data, ripe.rspace
#First an activity term matrix is defined to make assignment easier later
if ('clc' in data['otherdata']):
aterm = np.zeros([
data['otherdata']['npc'], data['otherdata']['ntimedata'],
data['rxnet']['mechanisms']
])
for i in range(data['otherdata']['npc']):
for j in range(data['otherdata']['ntimedata']):
#t is an index on the current mechanism
t = 0
if ('clc' in data['otherdata']):
for m in range(len(data['otherdata']['clc'])):
mech = data['otherdata']['clc'][m]
# Check for CLC mechanisms
if (mech == 'c1'):
aterm[i, j, t] = 1.0 - data['concdata']['X'][i][
j] #random nucleation
if (mech == 'c2'):
aterm[
i, j,
t] = 2.0 / 3.0 * data['concdata']['X'][i][j]**(
2.0 / 3.0 - 3.0 / 2.0
) #Power law of oer 2/3
if (mech == 'c3'):
aterm[i, j,
t] = 2.0 * data['concdata']['X'][i][j]**(
2.0 - 1.0 / 2.0
) # Power law model of order 2
if (mech == 'c4'):
aterm[i, j,
t] = 4.0 * data['concdata']['X'][i][j]**(
4.0 - 1.0 / 4.0
) # power law model of order 4
if (mech == 'c5'):
aterm[i, j,
t] = 3.0 * data['concdata']['X'][i][j]**(
3.0 - 1.0 / 3.0
) # power law model of order 4
if (mech == 'c6'):
aterm[i, j, t] = 1.0 # power law model of order 1
if (mech == 'c7'):
aterm[i, j, t] = 2.0 * (
1.0 - data['concdata']['X'][i][j]
) * np.power(
-np.log(1.0 - data['concdata']['X'][i][j]), 1.5
) #nucleation mode - Avrami-Erofeev eq of order 2
if (mech == 'c8'):
aterm[i, j, t] = 3.0 * (
1.0 - data['concdata']['X'][i][j]) * np.power(
-np.log(1.0 - data['concdata']['X'][i][j]),
3.0 - (1.0 / 3.0)) # '' order 3
if (mech == 'c9'):
aterm[i, j, t] = 0.5 * (
1.0 - data['concdata']['X'][i][j]
) * (-np.log(1.0 - data['concdata']['X'][i][j]))**(
0.5 - 1.0 / 0.5) # '' order 0.5
if (mech == 'c10'):
aterm[i, j, t] = 1.5 * (
1.0 - data['concdata']['X'][i][j]) * np.power(
-np.log(1.0 - data['concdata']['X'][i][j]),
1.5 - (1.0 / 1.5)) # '' order 1.5
if (mech == 'c11'):
aterm[i, j, t] = 4.0 * (
1.0 - data['concdata']['X'][i][j]
) * (-np.log(1.0 - data['concdata']['X'][i][j]))**(
4.0 - 1.0 / 4.0) # '' order 4
if (mech == 'c12'):
aterm[i, j, t] = data['concdata']['X'][i][j] * (
1.0 - data['concdata']['X'][i][j]
) #Prout tompkins
# diffusion models
if (mech == 'c13'):
aterm[i, j, t] = 3.0 * (
1.0 - data['concdata']['X'][i][j]
)**(1.0 / 3.0) * (
2.0 * ((1.0 - data['concdata']['X'][i][j])**
(-1.0 / 3.0) - 1.0)
)**(
-1
) #Jander equation (3d difsusion spherical sym)
if (mech == 'c14'):
aterm[i, j, t] = 3.0 / 2.0 * (
1.0 +
data['concdata']['X'][i][j])**(2.0 / 3.0) * (
((1.0 + data['concdata']['X'][i][j])**
(-1.0 / 3.0) - 1.0))**(-1) #antijander
if (mech == 'c15'):
aterm[i, j, t] = 1.0 / (
-np.log(1.0 - data['concdata']['X'][i][j])
) # valensi
if (mech == 'c16'):
aterm[i, j, t] = 1.0 / (
2.0 * data['concdata']['X'][i][j]) # parabolic
if (mech == 'c17'):
aterm[i, j, t] = (3.0 / 2.0) * ((
(1.0 - data['concdata']['X'][i][j])**
(-1.0 / 3.0) - 1.0))**(
-1
) #Ginstling-Brounstein eq 3d diff cylindrical
if (mech == 'c18'):
aterm[i, j, t] = 3.0 / 2.0 * (
1.0 -
data['concdata']['X'][i][j])**(4.0 / 3.0) * (
((1.0 - data['concdata']['X'][i][j])**
(-1.0 / 3.0) - 1.0))**(
-1) #zhuralev, lesohkhin, tempelman
if (mech == 'c19'):
aterm[i, j,
t] = (1.0 -
data['concdata']['X'][i][j])**(2.0 /
3.0)
# if(mech=='c19'): aterm[i,j,t]= (0.2)**(0.535) * 1.5 * (0.112)**(1.0/1.5)*((1.0)-data['concdata']['X'][i][j])*(-np.log(1.0-data['concdata']['X'][i][j]))**(1.0-1.0/1.5) # if(mech=='c20'): aterm[i,j,t]= (0.3)**(0.312) * 2.76 * (11.24)**(1.0/2.76)*((0.18 *1.479)-data['concdata']['X'][i][j])*(-np.log(1.0-data['concdata']['X'][i][j])/(0.18 *1.479))**(1.0-1.0/2.76) # if(mech=='c21'): aterm[i,j,t]= (0.3)**(2.13) * 1.63 * (28.4)**(1.0/1.63)*((0.135*1.479)-data['concdata']['X'][i][j])*(-np.log(1.0-data['concdata']['X'][i][j])/(0.135*1.479))**(1.0-1.0/1.63)
t = t + 1
else:
for m in range(data['rxnet']['mechanisms']):
tempa = 1
for s in data['otherdata']['slabels']:
if data['rxnet'][i, s] < 0:
tempa = tempa * float(data['concdata'][s][i][j])**float(
abs(data['rxnet'][i, s]))
aterm[i, j, m] = tempa
for q in range(data['rxnet']['mechanisms']):
if (np.isinf(aterm[i, j, q])):
aterm[i, j, q] = 0
if (np.isnan(aterm[i, j, q])):
aterm[i, j, q] = 0
if ('time' not in data['procdata']):
for q in range(data['rxnet']['nrx']):
temp = 1
for s in data['otherdata']['slabels']:
if (data['rxnet'][q, s] < 0):
temp = temp * data['concdata'][s][i]**abs(
data['rxnet'][q, s])
aterm[i, q] = temp
# check for isothermal, single species observations
if len(data['otherdata']['slabels']) == 1:
if 'T' not in data['otherdata']['pclabels']:
sys.stdout = open(os.devnull, 'w')
r = alamopy.doalamo(np.asarray(aterm[0, :, :]),
np.asarray(data['ratedata'][str(
data['otherdata']['slabels'][0])][0, :]),
linfcns=1,
constant=0,
maxterms=1,
expandoutput=1)
r = alamopy.almconfidence(r)
sys.stdout = sys.__stdout__
selected = r['model'].split(' * ')[-1]
selected = selected[1:]
coeff = r['model'].split(' * ')[0]
coeff = coeff.split(' = ')[-1]
cfi = r['conf_inv'][0].split(' : ')[-1]
slist = [
'Selected model is : ' + rspace['clcnames'][int(selected) - 1],
'Selected model form : ' +
rspace['clcforms'][int(selected) - 1],
'Estimated rate constant : ' + cfi
]
sys.stdout.write(slist[0] + '\n')
sys.stdout.write(slist[1] + '\n')
sys.stdout.write(slist[2] + '\n')
with open('output.txt', 'w') as o:
o.write(slist[0] + '\n')
o.write(slist[1] + '\n')
o.write(slist[2] + '\n')
deletefile('*logscratch*', 'tempalm.py', 'sv.alm', 'tmpscratch')
return [aterm, 0]
else:
writeminlp(aterm, ccon)
return [aterm, 0]